obu.c:clz: 186| 7.45k|static inline int clz(const unsigned int mask) { 187| 7.45k| return __builtin_clz(mask); 188| 7.45k|} thread_task.c:ctz: 182| 1.79M|static inline int ctz(const unsigned int mask) { 183| 1.79M| return __builtin_ctz(mask); 184| 1.79M|} decode.c:ctz: 182| 192k|static inline int ctz(const unsigned int mask) { 183| 192k| return __builtin_ctz(mask); 184| 192k|} decode.c:clz: 186| 26.8M|static inline int clz(const unsigned int mask) { 187| 26.8M| return __builtin_clz(mask); 188| 26.8M|} getbits.c:clz: 186| 178k|static inline int clz(const unsigned int mask) { 187| 178k| return __builtin_clz(mask); 188| 178k|} lf_mask.c:clz: 186| 6.75M|static inline int clz(const unsigned int mask) { 187| 6.75M| return __builtin_clz(mask); 188| 6.75M|} warpmv.c:clz: 186| 311k|static inline int clz(const unsigned int mask) { 187| 311k| return __builtin_clz(mask); 188| 311k|} warpmv.c:clzll: 190| 161k|static inline int clzll(const unsigned long long mask) { 191| 161k| return __builtin_clzll(mask); 192| 161k|} looprestoration_tmpl.c:clz: 186| 2.19M|static inline int clz(const unsigned int mask) { 187| 2.19M| return __builtin_clz(mask); 188| 2.19M|} recon_tmpl.c:clz: 186| 103M|static inline int clz(const unsigned int mask) { 187| 103M| return __builtin_clz(mask); 188| 103M|} cdef_apply_tmpl.c:clz: 186| 2.44M|static inline int clz(const unsigned int mask) { 187| 2.44M| return __builtin_clz(mask); 188| 2.44M|} ipred_prepare_tmpl.c:clz: 186| 16.4M|static inline int clz(const unsigned int mask) { 187| 16.4M| return __builtin_clz(mask); 188| 16.4M|} fg_apply_tmpl.c:PXSTRIDE: 79| 53.3k|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 53.3k| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 53.3k, False: 4] ------------------ 81| 53.3k| return x >> 1; 82| 53.3k|} itx_tmpl.c:PXSTRIDE: 79| 2.36M|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 2.36M| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 2.36M, False: 18.4E] ------------------ 81| 2.36M| return x >> 1; 82| 2.36M|} looprestoration_tmpl.c:PXSTRIDE: 79| 5.24M|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 5.24M| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 5.24M, False: 18.4E] ------------------ 81| 5.24M| return x >> 1; 82| 5.24M|} recon_tmpl.c:PXSTRIDE: 79| 40.2M|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 40.2M| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 40.3M, False: 18.4E] ------------------ 81| 40.3M| return x >> 1; 82| 40.2M|} cdef_apply_tmpl.c:PXSTRIDE: 79| 78.5M|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 78.5M| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 78.5M, False: 53.5k] ------------------ 81| 78.5M| return x >> 1; 82| 78.5M|} ipred_prepare_tmpl.c:PXSTRIDE: 79| 106M|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 106M| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 105M, False: 603k] ------------------ 81| 105M| return x >> 1; 82| 106M|} ipred_prepare_tmpl.c:pixel_set: 66| 2.14M|static inline void pixel_set(pixel *const dst, const int val, const int num) { 67| 22.1M| for (int n = 0; n < num; n++) ------------------ | Branch (67:21): [True: 20.0M, False: 2.14M] ------------------ 68| 20.0M| dst[n] = val; 69| 2.14M|} lf_apply_tmpl.c:PXSTRIDE: 79| 42.1M|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 42.1M| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 42.1M, False: 18.4E] ------------------ 81| 42.1M| return x >> 1; 82| 42.1M|} lr_apply_tmpl.c:PXSTRIDE: 79| 1.60M|static inline ptrdiff_t PXSTRIDE(const ptrdiff_t x) { 80| 1.60M| assert(!(x & 1)); ------------------ | Branch (80:5): [True: 1.60M, False: 2] ------------------ 81| 1.60M| return x >> 1; 82| 1.60M|} lib.c:umin: 47| 9.41k|static inline unsigned umin(const unsigned a, const unsigned b) { 48| 9.41k| return a < b ? a : b; ------------------ | Branch (48:12): [True: 0, False: 9.41k] ------------------ 49| 9.41k|} obu.c:ulog2: 67| 7.45k|static inline int ulog2(const unsigned v) { 68| 7.45k| return 31 ^ clz(v); 69| 7.45k|} obu.c:imin: 39| 1.10M|static inline int imin(const int a, const int b) { 40| 1.10M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 983k, False: 123k] ------------------ 41| 1.10M|} obu.c:imax: 35| 1.01M|static inline int imax(const int a, const int b) { 36| 1.01M| return a > b ? a : b; ------------------ | Branch (36:12): [True: 118k, False: 892k] ------------------ 37| 1.01M|} obu.c:iclip_u8: 55| 182k|static inline int iclip_u8(const int v) { 56| 182k| return iclip(v, 0, 255); 57| 182k|} obu.c:iclip: 51| 182k|static inline int iclip(const int v, const int min, const int max) { 52| 182k| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 6.21k, False: 176k] | Branch (52:28): [True: 3.04k, False: 173k] ------------------ 53| 182k|} refmvs.c:imin: 39| 71.4M|static inline int imin(const int a, const int b) { 40| 71.4M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 30.2M, False: 41.1M] ------------------ 41| 71.4M|} refmvs.c:apply_sign: 59| 2.76M|static inline int apply_sign(const int v, const int s) { 60| 2.76M| return s < 0 ? -v : v; ------------------ | Branch (60:12): [True: 1.72M, False: 1.04M] ------------------ 61| 2.76M|} refmvs.c:imax: 35| 31.0M|static inline int imax(const int a, const int b) { 36| 31.0M| return a > b ? a : b; ------------------ | Branch (36:12): [True: 3.84M, False: 27.2M] ------------------ 37| 31.0M|} refmvs.c:iclip: 51| 30.0M|static inline int iclip(const int v, const int min, const int max) { 52| 30.0M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 331k, False: 29.7M] | Branch (52:28): [True: 394k, False: 29.3M] ------------------ 53| 30.0M|} thread_task.c:imax: 35| 10.6M|static inline int imax(const int a, const int b) { 36| 10.6M| return a > b ? a : b; ------------------ | Branch (36:12): [True: 1.47M, False: 9.18M] ------------------ 37| 10.6M|} thread_task.c:iclip: 51| 1.63M|static inline int iclip(const int v, const int min, const int max) { 52| 1.63M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 626, False: 1.63M] | Branch (52:28): [True: 27.9k, False: 1.60M] ------------------ 53| 1.63M|} thread_task.c:umin: 47| 13.1M|static inline unsigned umin(const unsigned a, const unsigned b) { 48| 13.1M| return a < b ? a : b; ------------------ | Branch (48:12): [True: 115k, False: 13.0M] ------------------ 49| 13.1M|} wedge.c:imax: 35| 256|static inline int imax(const int a, const int b) { 36| 256| return a > b ? a : b; ------------------ | Branch (36:12): [True: 128, False: 128] ------------------ 37| 256|} wedge.c:imin: 39| 2.48k|static inline int imin(const int a, const int b) { 40| 2.48k| return a < b ? a : b; ------------------ | Branch (40:12): [True: 1.41k, False: 1.06k] ------------------ 41| 2.48k|} fg_apply_tmpl.c:imin: 39| 44.9k|static inline int imin(const int a, const int b) { 40| 44.9k| return a < b ? a : b; ------------------ | Branch (40:12): [True: 10.7k, False: 34.1k] ------------------ 41| 44.9k|} cdf.c:imin: 39| 118k|static inline int imin(const int a, const int b) { 40| 118k| return a < b ? a : b; ------------------ | Branch (40:12): [True: 29.5k, False: 88.6k] ------------------ 41| 118k|} decode.c:iclip: 51| 4.32M|static inline int iclip(const int v, const int min, const int max) { 52| 4.32M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 207k, False: 4.11M] | Branch (52:28): [True: 284k, False: 3.83M] ------------------ 53| 4.32M|} decode.c:apply_sign: 59| 1.83M|static inline int apply_sign(const int v, const int s) { 60| 1.83M| return s < 0 ? -v : v; ------------------ | Branch (60:12): [True: 1.37M, False: 461k] ------------------ 61| 1.83M|} decode.c:apply_sign64: 63| 1.80M|static inline int apply_sign64(const int v, const int64_t s) { 64| 1.80M| return s < 0 ? -v : v; ------------------ | Branch (64:12): [True: 293k, False: 1.50M] ------------------ 65| 1.80M|} decode.c:ulog2: 67| 26.8M|static inline int ulog2(const unsigned v) { 68| 26.8M| return 31 ^ clz(v); 69| 26.8M|} decode.c:imax: 35| 33.4M|static inline int imax(const int a, const int b) { 36| 33.4M| return a > b ? a : b; ------------------ | Branch (36:12): [True: 13.7M, False: 19.6M] ------------------ 37| 33.4M|} decode.c:imin: 39| 76.1M|static inline int imin(const int a, const int b) { 40| 76.1M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 40.6M, False: 35.4M] ------------------ 41| 76.1M|} decode.c:iclip_u8: 55| 3.04M|static inline int iclip_u8(const int v) { 56| 3.04M| return iclip(v, 0, 255); 57| 3.04M|} getbits.c:ulog2: 67| 178k|static inline int ulog2(const unsigned v) { 68| 178k| return 31 ^ clz(v); 69| 178k|} getbits.c:inv_recenter: 75| 178k|static inline unsigned inv_recenter(const unsigned r, const unsigned v) { 76| 178k| if (v > (r << 1)) ------------------ | Branch (76:9): [True: 3.41k, False: 175k] ------------------ 77| 3.41k| return v; 78| 175k| else if ((v & 1) == 0) ------------------ | Branch (78:14): [True: 128k, False: 46.8k] ------------------ 79| 128k| return (v >> 1) + r; 80| 46.8k| else 81| 46.8k| return r - ((v + 1) >> 1); 82| 178k|} lf_mask.c:imin: 39| 158M|static inline int imin(const int a, const int b) { 40| 158M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 15.8M, False: 142M] ------------------ 41| 158M|} lf_mask.c:ulog2: 67| 6.75M|static inline int ulog2(const unsigned v) { 68| 6.75M| return 31 ^ clz(v); 69| 6.75M|} lf_mask.c:imax: 35| 2.17M|static inline int imax(const int a, const int b) { 36| 2.17M| return a > b ? a : b; ------------------ | Branch (36:12): [True: 2.03M, False: 140k] ------------------ 37| 2.17M|} lf_mask.c:iclip: 51| 7.29M|static inline int iclip(const int v, const int min, const int max) { 52| 7.29M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 728k, False: 6.57M] | Branch (52:28): [True: 261k, False: 6.30M] ------------------ 53| 7.29M|} msac.c:inv_recenter: 75| 231k|static inline unsigned inv_recenter(const unsigned r, const unsigned v) { 76| 231k| if (v > (r << 1)) ------------------ | Branch (76:9): [True: 80.6k, False: 150k] ------------------ 77| 80.6k| return v; 78| 150k| else if ((v & 1) == 0) ------------------ | Branch (78:14): [True: 74.0k, False: 76.8k] ------------------ 79| 74.0k| return (v >> 1) + r; 80| 76.8k| else 81| 76.8k| return r - ((v + 1) >> 1); 82| 231k|} warpmv.c:apply_sign: 59| 1.55M|static inline int apply_sign(const int v, const int s) { 60| 1.55M| return s < 0 ? -v : v; ------------------ | Branch (60:12): [True: 401k, False: 1.15M] ------------------ 61| 1.55M|} warpmv.c:ulog2: 67| 311k|static inline int ulog2(const unsigned v) { 68| 311k| return 31 ^ clz(v); 69| 311k|} warpmv.c:apply_sign64: 63| 1.42M|static inline int apply_sign64(const int v, const int64_t s) { 64| 1.42M| return s < 0 ? -v : v; ------------------ | Branch (64:12): [True: 247k, False: 1.17M] ------------------ 65| 1.42M|} warpmv.c:iclip: 51| 2.45M|static inline int iclip(const int v, const int min, const int max) { 52| 2.45M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 57.2k, False: 2.39M] | Branch (52:28): [True: 72.4k, False: 2.32M] ------------------ 53| 2.45M|} warpmv.c:u64log2: 71| 161k|static inline int u64log2(const uint64_t v) { 72| 161k| return 63 ^ clzll(v); 73| 161k|} itx_tmpl.c:iclip: 51| 148M|static inline int iclip(const int v, const int min, const int max) { 52| 148M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 10.2M, False: 138M] | Branch (52:28): [True: 15.0M, False: 123M] ------------------ 53| 148M|} itx_tmpl.c:imin: 39| 61.8k|static inline int imin(const int a, const int b) { 40| 61.8k| return a < b ? a : b; ------------------ | Branch (40:12): [True: 8.40k, False: 53.4k] ------------------ 41| 61.8k|} looprestoration_tmpl.c:iclip: 51| 165M|static inline int iclip(const int v, const int min, const int max) { 52| 165M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 461k, False: 165M] | Branch (52:28): [True: 439k, False: 164M] ------------------ 53| 165M|} looprestoration_tmpl.c:imax: 35| 186M|static inline int imax(const int a, const int b) { 36| 186M| return a > b ? a : b; ------------------ | Branch (36:12): [True: 152M, False: 34.0M] ------------------ 37| 186M|} looprestoration_tmpl.c:umin: 47| 186M|static inline unsigned umin(const unsigned a, const unsigned b) { 48| 186M| return a < b ? a : b; ------------------ | Branch (48:12): [True: 180M, False: 6.41M] ------------------ 49| 186M|} recon_tmpl.c:ulog2: 67| 103M|static inline int ulog2(const unsigned v) { 68| 103M| return 31 ^ clz(v); 69| 103M|} recon_tmpl.c:imin: 39| 253M|static inline int imin(const int a, const int b) { 40| 253M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 226M, False: 27.1M] ------------------ 41| 253M|} recon_tmpl.c:imax: 35| 18.1M|static inline int imax(const int a, const int b) { 36| 18.1M| return a > b ? a : b; ------------------ | Branch (36:12): [True: 12.6M, False: 5.47M] ------------------ 37| 18.1M|} recon_tmpl.c:umin: 47| 447M|static inline unsigned umin(const unsigned a, const unsigned b) { 48| 447M| return a < b ? a : b; ------------------ | Branch (48:12): [True: 306M, False: 141M] ------------------ 49| 447M|} recon_tmpl.c:apply_sign64: 63| 1.33M|static inline int apply_sign64(const int v, const int64_t s) { 64| 1.33M| return s < 0 ? -v : v; ------------------ | Branch (64:12): [True: 98.6k, False: 1.23M] ------------------ 65| 1.33M|} recon_tmpl.c:iclip: 51| 1.00M|static inline int iclip(const int v, const int min, const int max) { 52| 1.00M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 57.4k, False: 948k] | Branch (52:28): [True: 11.9k, False: 936k] ------------------ 53| 1.00M|} itx_1d.c:iclip: 51| 343M|static inline int iclip(const int v, const int min, const int max) { 52| 343M| return v < min ? min : v > max ? max : v; ------------------ | Branch (52:12): [True: 7.04M, False: 336M] | Branch (52:28): [True: 7.01M, False: 329M] ------------------ 53| 343M|} scan.c:imax: 35| 3.34k|static inline int imax(const int a, const int b) { 36| 3.34k| return a > b ? a : b; ------------------ | Branch (36:12): [True: 2.82k, False: 523] ------------------ 37| 3.34k|} cdef_apply_tmpl.c:imin: 39| 10.6M|static inline int imin(const int a, const int b) { 40| 10.6M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 9.41M, False: 1.21M] ------------------ 41| 10.6M|} cdef_apply_tmpl.c:ulog2: 67| 2.44M|static inline int ulog2(const unsigned v) { 68| 2.44M| return 31 ^ clz(v); 69| 2.44M|} ipred_prepare_tmpl.c:imin: 39| 53.6M|static inline int imin(const int a, const int b) { 40| 53.6M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 50.9M, False: 2.72M] ------------------ 41| 53.6M|} lf_apply_tmpl.c:imin: 39| 11.9M|static inline int imin(const int a, const int b) { 40| 11.9M| return a < b ? a : b; ------------------ | Branch (40:12): [True: 1.25M, False: 10.6M] ------------------ 41| 11.9M|} lr_apply_tmpl.c:imin: 39| 375k|static inline int imin(const int a, const int b) { 40| 375k| return a < b ? a : b; ------------------ | Branch (40:12): [True: 256k, False: 118k] ------------------ 41| 375k|} dav1d_cdef_brow_8bpc: 102| 129k|{ 103| 129k| Dav1dFrameContext *const f = (Dav1dFrameContext *)tc->f; 104| 129k| const int bitdepth_min_8 = BITDEPTH == 8 ? 0 : f->cur.p.bpc - 8; ------------------ | Branch (104:32): [True: 129k, Folded] ------------------ 105| 129k| const Dav1dDSPContext *const dsp = f->dsp; 106| 129k| enum CdefEdgeFlags edges = CDEF_HAVE_BOTTOM | (by_start > 0 ? CDEF_HAVE_TOP : 0); ------------------ | Branch (106:52): [True: 114k, False: 14.7k] ------------------ 107| 129k| pixel *ptrs[3] = { p[0], p[1], p[2] }; 108| 129k| const int sbsz = 16; 109| 129k| const int sb64w = f->sb128w << 1; 110| 129k| const int damping = f->frame_hdr->cdef.damping + bitdepth_min_8; 111| 129k| const enum Dav1dPixelLayout layout = f->cur.p.layout; 112| 129k| const int uv_idx = DAV1D_PIXEL_LAYOUT_I444 - layout; 113| 129k| const int ss_ver = layout == DAV1D_PIXEL_LAYOUT_I420; 114| 129k| const int ss_hor = layout != DAV1D_PIXEL_LAYOUT_I444; 115| 129k| static const uint8_t uv_dirs[2][8] = { { 0, 1, 2, 3, 4, 5, 6, 7 }, 116| 129k| { 7, 0, 2, 4, 5, 6, 6, 6 } }; 117| 129k| const uint8_t *uv_dir = uv_dirs[layout == DAV1D_PIXEL_LAYOUT_I422]; 118| 129k| const int have_tt = f->c->n_tc > 1; 119| 129k| const int sb128 = f->seq_hdr->sb128; 120| 129k| const int resize = f->frame_hdr->width[0] != f->frame_hdr->width[1]; 121| 129k| const ptrdiff_t y_stride = PXSTRIDE(f->cur.stride[0]); ------------------ | | 53| 129k|#define PXSTRIDE(x) (x) ------------------ 122| 129k| const ptrdiff_t uv_stride = PXSTRIDE(f->cur.stride[1]); ------------------ | | 53| 129k|#define PXSTRIDE(x) (x) ------------------ 123| | 124| 885k| for (int bit = 0, by = by_start; by < by_end; by += 2, edges |= CDEF_HAVE_TOP) { ------------------ | Branch (124:38): [True: 753k, False: 131k] ------------------ 125| 753k| const int tf = tc->top_pre_cdef_toggle; 126| 753k| const int by_idx = (by & 30) >> 1; 127| 753k| if (by + 2 >= f->bh) edges &= ~CDEF_HAVE_BOTTOM; ------------------ | Branch (127:13): [True: 14.4k, False: 739k] ------------------ 128| | 129| 753k| if ((!have_tt || sbrow_start || by + 2 < by_end) && ------------------ | Branch (129:14): [True: 16, False: 753k] | Branch (129:26): [True: 57.2k, False: 696k] | Branch (129:41): [True: 624k, False: 72.3k] ------------------ 130| 681k| edges & CDEF_HAVE_BOTTOM) ------------------ | Branch (130:13): [True: 682k, False: 18.4E] ------------------ 131| 682k| { 132| | // backup pre-filter data for next iteration 133| 682k| pixel *const cdef_top_bak[3] = { 134| 682k| f->lf.cdef_line[!tf][0] + have_tt * sby * 4 * y_stride, 135| 682k| f->lf.cdef_line[!tf][1] + have_tt * sby * 8 * uv_stride, 136| 682k| f->lf.cdef_line[!tf][2] + have_tt * sby * 8 * uv_stride 137| 682k| }; 138| 682k| backup2lines(cdef_top_bak, ptrs, f->cur.stride, layout); 139| 682k| } 140| | 141| 753k| ALIGN_STK_16(pixel, lr_bak, 2 /* idx */, [3 /* plane */][8 /* y */][2 /* x */]); ------------------ | | 100| 753k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 753k| line __attribute__((aligned(align))) | | ------------------ ------------------ 142| 753k| pixel *iptrs[3] = { ptrs[0], ptrs[1], ptrs[2] }; 143| 753k| edges &= ~CDEF_HAVE_LEFT; 144| 753k| edges |= CDEF_HAVE_RIGHT; 145| 753k| enum Backup2x8Flags prev_flag = 0; 146| 3.06M| for (int sbx = 0; sbx < sb64w; sbx++, edges |= CDEF_HAVE_LEFT) { ------------------ | Branch (146:27): [True: 2.30M, False: 755k] ------------------ 147| 2.30M| const int sb128x = sbx >> 1; 148| 2.30M| const int sb64_idx = ((by & sbsz) >> 3) + (sbx & 1); 149| 2.30M| const int cdef_idx = lflvl[sb128x].cdef_idx[sb64_idx]; 150| 2.30M| if (cdef_idx == -1 || ------------------ | Branch (150:17): [True: 1.43M, False: 874k] ------------------ 151| 874k| (!f->frame_hdr->cdef.y_strength[cdef_idx] && ------------------ | Branch (151:18): [True: 375k, False: 499k] ------------------ 152| 375k| !f->frame_hdr->cdef.uv_strength[cdef_idx])) ------------------ | Branch (152:18): [True: 316k, False: 58.7k] ------------------ 153| 1.75M| { 154| 1.75M| prev_flag = 0; 155| 1.75M| goto next_sb; 156| 1.75M| } 157| | 158| | // Create a complete 32-bit mask for the sb row ahead of time. 159| 555k| const uint16_t (*noskip_row)[2] = &lflvl[sb128x].noskip_mask[by_idx]; 160| 555k| const unsigned noskip_mask = (unsigned) noskip_row[0][1] << 16 | 161| 555k| noskip_row[0][0]; 162| | 163| 555k| const int y_lvl = f->frame_hdr->cdef.y_strength[cdef_idx]; 164| 555k| const int uv_lvl = f->frame_hdr->cdef.uv_strength[cdef_idx]; 165| 555k| const enum Backup2x8Flags flag = !!y_lvl + (!!uv_lvl << 1); 166| | 167| 555k| const int y_pri_lvl = (y_lvl >> 2) << bitdepth_min_8; 168| 555k| int y_sec_lvl = y_lvl & 3; 169| 555k| y_sec_lvl += y_sec_lvl == 3; 170| 555k| y_sec_lvl <<= bitdepth_min_8; 171| | 172| 555k| const int uv_pri_lvl = (uv_lvl >> 2) << bitdepth_min_8; 173| 555k| int uv_sec_lvl = uv_lvl & 3; 174| 555k| uv_sec_lvl += uv_sec_lvl == 3; 175| 555k| uv_sec_lvl <<= bitdepth_min_8; 176| | 177| 555k| pixel *bptrs[3] = { iptrs[0], iptrs[1], iptrs[2] }; 178| 4.80M| for (int bx = sbx * sbsz; bx < imin((sbx + 1) * sbsz, f->bw); ------------------ | Branch (178:39): [True: 4.24M, False: 560k] ------------------ 179| 4.24M| bx += 2, edges |= CDEF_HAVE_LEFT) 180| 4.24M| { 181| 4.24M| if (bx + 2 >= f->bw) edges &= ~CDEF_HAVE_RIGHT; ------------------ | Branch (181:21): [True: 57.8k, False: 4.18M] ------------------ 182| | 183| | // check if this 8x8 block had any coded coefficients; if not, 184| | // go to the next block 185| 4.24M| const uint32_t bx_mask = 3U << (bx & 30); 186| 4.24M| if (!(noskip_mask & bx_mask)) { ------------------ | Branch (186:21): [True: 910k, False: 3.33M] ------------------ 187| 910k| prev_flag = 0; 188| 910k| goto next_b; 189| 910k| } 190| 3.33M| const enum Backup2x8Flags do_left = (prev_flag ^ flag) & flag; 191| 3.33M| prev_flag = flag; 192| 3.33M| if (do_left && edges & CDEF_HAVE_LEFT) { ------------------ | Branch (192:21): [True: 203k, False: 3.13M] | Branch (192:32): [True: 153k, False: 49.9k] ------------------ 193| | // we didn't backup the prefilter data because it wasn't 194| | // there, so do it here instead 195| 153k| backup2x8(lr_bak[bit], bptrs, f->cur.stride, 0, layout, do_left); 196| 153k| } 197| 3.33M| if (edges & CDEF_HAVE_RIGHT) { ------------------ | Branch (197:21): [True: 3.29M, False: 40.5k] ------------------ 198| | // backup pre-filter data for next iteration 199| 3.29M| backup2x8(lr_bak[!bit], bptrs, f->cur.stride, 8, layout, flag); 200| 3.29M| } 201| | 202| 3.33M| int dir; 203| 3.33M| unsigned variance; 204| 3.33M| if (y_pri_lvl || uv_pri_lvl) ------------------ | Branch (204:21): [True: 2.52M, False: 812k] | Branch (204:34): [True: 517k, False: 295k] ------------------ 205| 3.01M| dir = dsp->cdef.dir(bptrs[0], f->cur.stride[0], 206| 3.01M| &variance HIGHBD_CALL_SUFFIX); 207| | 208| 3.33M| const pixel *top, *bot; 209| 3.33M| ptrdiff_t offset; 210| | 211| 3.33M| if (!have_tt) goto st_y; ------------------ | Branch (211:21): [True: 0, False: 3.33M] ------------------ 212| 3.33M| if (sbrow_start && by == by_start) { ------------------ | Branch (212:21): [True: 208k, False: 3.12M] | Branch (212:36): [True: 208k, False: 18.4E] ------------------ 213| 208k| if (resize) { ------------------ | Branch (213:25): [True: 16.3k, False: 192k] ------------------ 214| 16.3k| offset = (sby - 1) * 4 * y_stride + bx * 4; 215| 16.3k| top = &f->lf.cdef_lpf_line[0][offset]; 216| 192k| } else { 217| 192k| offset = (sby * (4 << sb128) - 4) * y_stride + bx * 4; 218| 192k| top = &f->lf.lr_lpf_line[0][offset]; 219| 192k| } 220| 208k| bot = bptrs[0] + 8 * y_stride; 221| 3.12M| } else if (!sbrow_start && by + 2 >= by_end) { ------------------ | Branch (221:28): [True: 3.11M, False: 5.85k] | Branch (221:44): [True: 248k, False: 2.87M] ------------------ 222| 248k| top = &f->lf.cdef_line[tf][0][sby * 4 * y_stride + bx * 4]; 223| 248k| if (resize) { ------------------ | Branch (223:25): [True: 18.0k, False: 230k] ------------------ 224| 18.0k| offset = (sby * 4 + 2) * y_stride + bx * 4; 225| 18.0k| bot = &f->lf.cdef_lpf_line[0][offset]; 226| 230k| } else { 227| 230k| const int line = sby * (4 << sb128) + 4 * sb128 + 2; 228| 230k| offset = line * y_stride + bx * 4; 229| 230k| bot = &f->lf.lr_lpf_line[0][offset]; 230| 230k| } 231| 2.87M| } else { 232| 2.87M| st_y:; 233| 2.87M| offset = sby * 4 * y_stride; 234| 2.87M| top = &f->lf.cdef_line[tf][0][have_tt * offset + bx * 4]; 235| 2.87M| bot = bptrs[0] + 8 * y_stride; 236| 2.87M| } 237| 3.33M| if (y_pri_lvl) { ------------------ | Branch (237:21): [True: 2.50M, False: 822k] ------------------ 238| 2.50M| const int adj_y_pri_lvl = adjust_strength(y_pri_lvl, variance); 239| 2.50M| if (adj_y_pri_lvl || y_sec_lvl) ------------------ | Branch (239:25): [True: 1.82M, False: 685k] | Branch (239:42): [True: 233k, False: 451k] ------------------ 240| 2.05M| dsp->cdef.fb[0](bptrs[0], f->cur.stride[0], lr_bak[bit][0], 241| 2.05M| top, bot, adj_y_pri_lvl, y_sec_lvl, 242| 2.05M| dir, damping, edges HIGHBD_CALL_SUFFIX); 243| 2.50M| } else if (y_sec_lvl) ------------------ | Branch (243:28): [True: 469k, False: 352k] ------------------ 244| 469k| dsp->cdef.fb[0](bptrs[0], f->cur.stride[0], lr_bak[bit][0], 245| 469k| top, bot, 0, y_sec_lvl, 0, damping, 246| 469k| edges HIGHBD_CALL_SUFFIX); 247| | 248| 3.33M| if (!uv_lvl) goto skip_uv; ------------------ | Branch (248:21): [True: 278k, False: 3.05M] ------------------ 249| 3.33M| assert(layout != DAV1D_PIXEL_LAYOUT_I400); ------------------ | Branch (249:17): [True: 3.04M, False: 3.39k] ------------------ 250| | 251| 3.04M| const int uvdir = uv_pri_lvl ? uv_dir[dir] : 0; ------------------ | Branch (251:35): [True: 2.75M, False: 293k] ------------------ 252| 9.11M| for (int pl = 1; pl <= 2; pl++) { ------------------ | Branch (252:34): [True: 6.04M, False: 3.06M] ------------------ 253| 6.04M| if (!have_tt) goto st_uv; ------------------ | Branch (253:25): [True: 0, False: 6.04M] ------------------ 254| 6.04M| if (sbrow_start && by == by_start) { ------------------ | Branch (254:25): [True: 376k, False: 5.66M] | Branch (254:40): [True: 376k, False: 18.4E] ------------------ 255| 376k| if (resize) { ------------------ | Branch (255:29): [True: 31.1k, False: 345k] ------------------ 256| 31.1k| offset = (sby - 1) * 4 * uv_stride + (bx * 4 >> ss_hor); 257| 31.1k| top = &f->lf.cdef_lpf_line[pl][offset]; 258| 345k| } else { 259| 345k| const int line = sby * (4 << sb128) - 4; 260| 345k| offset = line * uv_stride + (bx * 4 >> ss_hor); 261| 345k| top = &f->lf.lr_lpf_line[pl][offset]; 262| 345k| } 263| 376k| bot = bptrs[pl] + (8 >> ss_ver) * uv_stride; 264| 5.67M| } else if (!sbrow_start && by + 2 >= by_end) { ------------------ | Branch (264:32): [True: 5.67M, False: 18.4E] | Branch (264:48): [True: 439k, False: 5.23M] ------------------ 265| 439k| const ptrdiff_t top_offset = sby * 8 * uv_stride + 266| 439k| (bx * 4 >> ss_hor); 267| 439k| top = &f->lf.cdef_line[tf][pl][top_offset]; 268| 439k| if (resize) { ------------------ | Branch (268:29): [True: 33.4k, False: 406k] ------------------ 269| 33.4k| offset = (sby * 4 + 2) * uv_stride + (bx * 4 >> ss_hor); 270| 33.4k| bot = &f->lf.cdef_lpf_line[pl][offset]; 271| 406k| } else { 272| 406k| const int line = sby * (4 << sb128) + 4 * sb128 + 2; 273| 406k| offset = line * uv_stride + (bx * 4 >> ss_hor); 274| 406k| bot = &f->lf.lr_lpf_line[pl][offset]; 275| 406k| } 276| 5.22M| } else { 277| 5.24M| st_uv:; 278| 5.24M| const ptrdiff_t offset = sby * 8 * uv_stride; 279| 5.24M| top = &f->lf.cdef_line[tf][pl][have_tt * offset + (bx * 4 >> ss_hor)]; 280| 5.24M| bot = bptrs[pl] + (8 >> ss_ver) * uv_stride; 281| 5.24M| } 282| 6.06M| dsp->cdef.fb[uv_idx](bptrs[pl], f->cur.stride[1], 283| 6.06M| lr_bak[bit][pl], top, bot, 284| 6.06M| uv_pri_lvl, uv_sec_lvl, uvdir, 285| 6.06M| damping - 1, edges HIGHBD_CALL_SUFFIX); 286| 6.06M| } 287| | 288| 3.34M| skip_uv: 289| 3.34M| bit ^= 1; 290| | 291| 4.24M| next_b: 292| 4.24M| bptrs[0] += 8; 293| 4.24M| bptrs[1] += 8 >> ss_hor; 294| 4.24M| bptrs[2] += 8 >> ss_hor; 295| 4.24M| } 296| | 297| 2.31M| next_sb: 298| 2.31M| iptrs[0] += sbsz * 4; 299| 2.31M| iptrs[1] += sbsz * 4 >> ss_hor; 300| 2.31M| iptrs[2] += sbsz * 4 >> ss_hor; 301| 2.31M| } 302| | 303| 755k| ptrs[0] += 8 * PXSTRIDE(f->cur.stride[0]); ------------------ | | 53| 755k|#define PXSTRIDE(x) (x) ------------------ 304| 755k| ptrs[1] += 8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver; ------------------ | | 53| 755k|#define PXSTRIDE(x) (x) ------------------ 305| 755k| ptrs[2] += 8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver; ------------------ | | 53| 755k|#define PXSTRIDE(x) (x) ------------------ 306| 755k| tc->top_pre_cdef_toggle ^= 1; 307| 755k| } 308| 129k|} cdef_apply_tmpl.c:backup2lines: 44| 12.1M|{ 45| 12.1M| const ptrdiff_t y_stride = PXSTRIDE(stride[0]); ------------------ | | 53| 12.1M|#define PXSTRIDE(x) (x) ------------------ 46| 12.1M| if (y_stride < 0) ------------------ | Branch (46:9): [True: 0, False: 12.1M] ------------------ 47| 0| pixel_copy(dst[0] + y_stride, src[0] + 7 * y_stride, -2 * y_stride); ------------------ | | 47| 0|#define pixel_copy memcpy ------------------ 48| 12.1M| else 49| 12.1M| pixel_copy(dst[0], src[0] + 6 * y_stride, 2 * y_stride); ------------------ | | 47| 12.1M|#define pixel_copy memcpy ------------------ 50| | 51| 12.1M| if (layout != DAV1D_PIXEL_LAYOUT_I400) { ------------------ | Branch (51:9): [True: 1.67M, False: 10.4M] ------------------ 52| 1.67M| const ptrdiff_t uv_stride = PXSTRIDE(stride[1]); ------------------ | | 53| 1.67M|#define PXSTRIDE(x) (x) ------------------ 53| 1.67M| if (uv_stride < 0) { ------------------ | Branch (53:13): [True: 0, False: 1.67M] ------------------ 54| 0| const int uv_off = layout == DAV1D_PIXEL_LAYOUT_I420 ? 3 : 7; ------------------ | Branch (54:32): [True: 0, False: 0] ------------------ 55| 0| pixel_copy(dst[1] + uv_stride, src[1] + uv_off * uv_stride, -2 * uv_stride); ------------------ | | 47| 0|#define pixel_copy memcpy ------------------ 56| 0| pixel_copy(dst[2] + uv_stride, src[2] + uv_off * uv_stride, -2 * uv_stride); ------------------ | | 47| 0|#define pixel_copy memcpy ------------------ 57| 1.67M| } else { 58| 1.67M| const int uv_off = layout == DAV1D_PIXEL_LAYOUT_I420 ? 2 : 6; ------------------ | Branch (58:32): [True: 640k, False: 1.03M] ------------------ 59| 1.67M| pixel_copy(dst[1], src[1] + uv_off * uv_stride, 2 * uv_stride); ------------------ | | 47| 1.67M|#define pixel_copy memcpy ------------------ 60| 1.67M| pixel_copy(dst[2], src[2] + uv_off * uv_stride, 2 * uv_stride); ------------------ | | 47| 1.67M|#define pixel_copy memcpy ------------------ 61| 1.67M| } 62| 1.67M| } 63| 12.1M|} cdef_apply_tmpl.c:backup2x8: 70| 5.95M|{ 71| 5.95M| ptrdiff_t y_off = 0; 72| 5.95M| if (flag & BACKUP_2X8_Y) { ------------------ | Branch (72:9): [True: 5.52M, False: 431k] ------------------ 73| 49.1M| for (int y = 0; y < 8; y++, y_off += PXSTRIDE(src_stride[0])) ------------------ | | 53| 43.6M|#define PXSTRIDE(x) (x) ------------------ | Branch (73:25): [True: 43.6M, False: 5.52M] ------------------ 74| 43.6M| pixel_copy(dst[0][y], &src[0][y_off + x_off - 2], 2); ------------------ | | 47| 43.6M|#define pixel_copy memcpy ------------------ 75| 5.52M| } 76| | 77| 5.95M| if (layout == DAV1D_PIXEL_LAYOUT_I400 || !(flag & BACKUP_2X8_UV)) ------------------ | Branch (77:9): [True: 1.10M, False: 4.85M] | Branch (77:46): [True: 333k, False: 4.51M] ------------------ 78| 1.41M| return; 79| | 80| 4.54M| const int ss_ver = layout == DAV1D_PIXEL_LAYOUT_I420; 81| 4.54M| const int ss_hor = layout != DAV1D_PIXEL_LAYOUT_I444; 82| | 83| 4.54M| x_off >>= ss_hor; 84| 4.54M| y_off = 0; 85| 25.6M| for (int y = 0; y < (8 >> ss_ver); y++, y_off += PXSTRIDE(src_stride[1])) { ------------------ | | 53| 21.1M|#define PXSTRIDE(x) (x) ------------------ | Branch (85:21): [True: 21.1M, False: 4.54M] ------------------ 86| 21.1M| pixel_copy(dst[1][y], &src[1][y_off + x_off - 2], 2); ------------------ | | 47| 21.1M|#define pixel_copy memcpy ------------------ 87| 21.1M| pixel_copy(dst[2][y], &src[2][y_off + x_off - 2], 2); ------------------ | | 47| 21.1M|#define pixel_copy memcpy ------------------ 88| 21.1M| } 89| 4.54M|} cdef_apply_tmpl.c:adjust_strength: 91| 4.87M|static int adjust_strength(const int strength, const unsigned var) { 92| 4.87M| if (!var) return 0; ------------------ | Branch (92:9): [True: 1.80M, False: 3.06M] ------------------ 93| 3.06M| const int i = var >> 6 ? imin(ulog2(var >> 6), 12) : 0; ------------------ | Branch (93:19): [True: 2.44M, False: 614k] ------------------ 94| 3.06M| return (strength * (4 + i) + 8) >> 4; 95| 4.87M|} dav1d_cdef_brow_16bpc: 102| 1.57M|{ 103| 1.57M| Dav1dFrameContext *const f = (Dav1dFrameContext *)tc->f; 104| 1.57M| const int bitdepth_min_8 = BITDEPTH == 8 ? 0 : f->cur.p.bpc - 8; ------------------ | Branch (104:32): [Folded, False: 1.57M] ------------------ 105| 1.57M| const Dav1dDSPContext *const dsp = f->dsp; 106| 1.57M| enum CdefEdgeFlags edges = CDEF_HAVE_BOTTOM | (by_start > 0 ? CDEF_HAVE_TOP : 0); ------------------ | Branch (106:52): [True: 1.42M, False: 143k] ------------------ 107| 1.57M| pixel *ptrs[3] = { p[0], p[1], p[2] }; 108| 1.57M| const int sbsz = 16; 109| 1.57M| const int sb64w = f->sb128w << 1; 110| 1.57M| const int damping = f->frame_hdr->cdef.damping + bitdepth_min_8; 111| 1.57M| const enum Dav1dPixelLayout layout = f->cur.p.layout; 112| 1.57M| const int uv_idx = DAV1D_PIXEL_LAYOUT_I444 - layout; 113| 1.57M| const int ss_ver = layout == DAV1D_PIXEL_LAYOUT_I420; 114| 1.57M| const int ss_hor = layout != DAV1D_PIXEL_LAYOUT_I444; 115| 1.57M| static const uint8_t uv_dirs[2][8] = { { 0, 1, 2, 3, 4, 5, 6, 7 }, 116| 1.57M| { 7, 0, 2, 4, 5, 6, 6, 6 } }; 117| 1.57M| const uint8_t *uv_dir = uv_dirs[layout == DAV1D_PIXEL_LAYOUT_I422]; 118| 1.57M| const int have_tt = f->c->n_tc > 1; 119| 1.57M| const int sb128 = f->seq_hdr->sb128; 120| 1.57M| const int resize = f->frame_hdr->width[0] != f->frame_hdr->width[1]; 121| 1.57M| const ptrdiff_t y_stride = PXSTRIDE(f->cur.stride[0]); 122| 1.57M| const ptrdiff_t uv_stride = PXSTRIDE(f->cur.stride[1]); 123| | 124| 13.8M| for (int bit = 0, by = by_start; by < by_end; by += 2, edges |= CDEF_HAVE_TOP) { ------------------ | Branch (124:38): [True: 12.2M, False: 1.58M] ------------------ 125| 12.2M| const int tf = tc->top_pre_cdef_toggle; 126| 12.2M| const int by_idx = (by & 30) >> 1; 127| 12.2M| if (by + 2 >= f->bh) edges &= ~CDEF_HAVE_BOTTOM; ------------------ | Branch (127:13): [True: 143k, False: 12.1M] ------------------ 128| | 129| 12.2M| if ((!have_tt || sbrow_start || by + 2 < by_end) && ------------------ | Branch (129:14): [True: 18.4E, False: 12.2M] | Branch (129:26): [True: 711k, False: 11.5M] | Branch (129:41): [True: 10.7M, False: 859k] ------------------ 130| 11.4M| edges & CDEF_HAVE_BOTTOM) ------------------ | Branch (130:13): [True: 11.4M, False: 18.4E] ------------------ 131| 11.4M| { 132| | // backup pre-filter data for next iteration 133| 11.4M| pixel *const cdef_top_bak[3] = { 134| 11.4M| f->lf.cdef_line[!tf][0] + have_tt * sby * 4 * y_stride, 135| 11.4M| f->lf.cdef_line[!tf][1] + have_tt * sby * 8 * uv_stride, 136| 11.4M| f->lf.cdef_line[!tf][2] + have_tt * sby * 8 * uv_stride 137| 11.4M| }; 138| 11.4M| backup2lines(cdef_top_bak, ptrs, f->cur.stride, layout); 139| 11.4M| } 140| | 141| 12.2M| ALIGN_STK_16(pixel, lr_bak, 2 /* idx */, [3 /* plane */][8 /* y */][2 /* x */]); ------------------ | | 100| 12.2M| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 12.2M| line __attribute__((aligned(align))) | | ------------------ ------------------ 142| 12.2M| pixel *iptrs[3] = { ptrs[0], ptrs[1], ptrs[2] }; 143| 12.2M| edges &= ~CDEF_HAVE_LEFT; 144| 12.2M| edges |= CDEF_HAVE_RIGHT; 145| 12.2M| enum Backup2x8Flags prev_flag = 0; 146| 37.3M| for (int sbx = 0; sbx < sb64w; sbx++, edges |= CDEF_HAVE_LEFT) { ------------------ | Branch (146:27): [True: 25.0M, False: 12.2M] ------------------ 147| 25.0M| const int sb128x = sbx >> 1; 148| 25.0M| const int sb64_idx = ((by & sbsz) >> 3) + (sbx & 1); 149| 25.0M| const int cdef_idx = lflvl[sb128x].cdef_idx[sb64_idx]; 150| 25.0M| if (cdef_idx == -1 || ------------------ | Branch (150:17): [True: 23.4M, False: 1.58M] ------------------ 151| 1.58M| (!f->frame_hdr->cdef.y_strength[cdef_idx] && ------------------ | Branch (151:18): [True: 1.11M, False: 471k] ------------------ 152| 1.11M| !f->frame_hdr->cdef.uv_strength[cdef_idx])) ------------------ | Branch (152:18): [True: 1.09M, False: 16.3k] ------------------ 153| 24.5M| { 154| 24.5M| prev_flag = 0; 155| 24.5M| goto next_sb; 156| 24.5M| } 157| | 158| | // Create a complete 32-bit mask for the sb row ahead of time. 159| 463k| const uint16_t (*noskip_row)[2] = &lflvl[sb128x].noskip_mask[by_idx]; 160| 463k| const unsigned noskip_mask = (unsigned) noskip_row[0][1] << 16 | 161| 463k| noskip_row[0][0]; 162| | 163| 463k| const int y_lvl = f->frame_hdr->cdef.y_strength[cdef_idx]; 164| 463k| const int uv_lvl = f->frame_hdr->cdef.uv_strength[cdef_idx]; 165| 463k| const enum Backup2x8Flags flag = !!y_lvl + (!!uv_lvl << 1); 166| | 167| 463k| const int y_pri_lvl = (y_lvl >> 2) << bitdepth_min_8; 168| 463k| int y_sec_lvl = y_lvl & 3; 169| 463k| y_sec_lvl += y_sec_lvl == 3; 170| 463k| y_sec_lvl <<= bitdepth_min_8; 171| | 172| 463k| const int uv_pri_lvl = (uv_lvl >> 2) << bitdepth_min_8; 173| 463k| int uv_sec_lvl = uv_lvl & 3; 174| 463k| uv_sec_lvl += uv_sec_lvl == 3; 175| 463k| uv_sec_lvl <<= bitdepth_min_8; 176| | 177| 463k| pixel *bptrs[3] = { iptrs[0], iptrs[1], iptrs[2] }; 178| 3.37M| for (int bx = sbx * sbsz; bx < imin((sbx + 1) * sbsz, f->bw); ------------------ | Branch (178:39): [True: 2.91M, False: 463k] ------------------ 179| 2.91M| bx += 2, edges |= CDEF_HAVE_LEFT) 180| 2.91M| { 181| 2.91M| if (bx + 2 >= f->bw) edges &= ~CDEF_HAVE_RIGHT; ------------------ | Branch (181:21): [True: 172k, False: 2.74M] ------------------ 182| | 183| | // check if this 8x8 block had any coded coefficients; if not, 184| | // go to the next block 185| 2.91M| const uint32_t bx_mask = 3U << (bx & 30); 186| 2.91M| if (!(noskip_mask & bx_mask)) { ------------------ | Branch (186:21): [True: 278k, False: 2.63M] ------------------ 187| 278k| prev_flag = 0; 188| 278k| goto next_b; 189| 278k| } 190| 2.63M| const enum Backup2x8Flags do_left = (prev_flag ^ flag) & flag; 191| 2.63M| prev_flag = flag; 192| 2.63M| if (do_left && edges & CDEF_HAVE_LEFT) { ------------------ | Branch (192:21): [True: 211k, False: 2.42M] | Branch (192:32): [True: 40.6k, False: 170k] ------------------ 193| | // we didn't backup the prefilter data because it wasn't 194| | // there, so do it here instead 195| 40.6k| backup2x8(lr_bak[bit], bptrs, f->cur.stride, 0, layout, do_left); 196| 40.6k| } 197| 2.63M| if (edges & CDEF_HAVE_RIGHT) { ------------------ | Branch (197:21): [True: 2.46M, False: 166k] ------------------ 198| | // backup pre-filter data for next iteration 199| 2.46M| backup2x8(lr_bak[!bit], bptrs, f->cur.stride, 8, layout, flag); 200| 2.46M| } 201| | 202| 2.63M| int dir; 203| 2.63M| unsigned variance; 204| 2.63M| if (y_pri_lvl || uv_pri_lvl) ------------------ | Branch (204:21): [True: 2.36M, False: 269k] | Branch (204:34): [True: 90.4k, False: 179k] ------------------ 205| 2.44M| dir = dsp->cdef.dir(bptrs[0], f->cur.stride[0], 206| 2.44M| &variance HIGHBD_CALL_SUFFIX); ------------------ | | 73| 2.44M|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 207| | 208| 2.63M| const pixel *top, *bot; 209| 2.63M| ptrdiff_t offset; 210| | 211| 2.63M| if (!have_tt) goto st_y; ------------------ | Branch (211:21): [True: 0, False: 2.63M] ------------------ 212| 2.63M| if (sbrow_start && by == by_start) { ------------------ | Branch (212:21): [True: 156k, False: 2.47M] | Branch (212:36): [True: 156k, False: 18.4E] ------------------ 213| 156k| if (resize) { ------------------ | Branch (213:25): [True: 48.2k, False: 108k] ------------------ 214| 48.2k| offset = (sby - 1) * 4 * y_stride + bx * 4; 215| 48.2k| top = &f->lf.cdef_lpf_line[0][offset]; 216| 108k| } else { 217| 108k| offset = (sby * (4 << sb128) - 4) * y_stride + bx * 4; 218| 108k| top = &f->lf.lr_lpf_line[0][offset]; 219| 108k| } 220| 156k| bot = bptrs[0] + 8 * y_stride; 221| 2.48M| } else if (!sbrow_start && by + 2 >= by_end) { ------------------ | Branch (221:28): [True: 2.48M, False: 18.4E] | Branch (221:44): [True: 222k, False: 2.25M] ------------------ 222| 222k| top = &f->lf.cdef_line[tf][0][sby * 4 * y_stride + bx * 4]; 223| 222k| if (resize) { ------------------ | Branch (223:25): [True: 66.8k, False: 155k] ------------------ 224| 66.8k| offset = (sby * 4 + 2) * y_stride + bx * 4; 225| 66.8k| bot = &f->lf.cdef_lpf_line[0][offset]; 226| 155k| } else { 227| 155k| const int line = sby * (4 << sb128) + 4 * sb128 + 2; 228| 155k| offset = line * y_stride + bx * 4; 229| 155k| bot = &f->lf.lr_lpf_line[0][offset]; 230| 155k| } 231| 2.25M| } else { 232| 2.26M| st_y:; 233| 2.26M| offset = sby * 4 * y_stride; 234| 2.26M| top = &f->lf.cdef_line[tf][0][have_tt * offset + bx * 4]; 235| 2.26M| bot = bptrs[0] + 8 * y_stride; 236| 2.26M| } 237| 2.64M| if (y_pri_lvl) { ------------------ | Branch (237:21): [True: 2.36M, False: 280k] ------------------ 238| 2.36M| const int adj_y_pri_lvl = adjust_strength(y_pri_lvl, variance); 239| 2.36M| if (adj_y_pri_lvl || y_sec_lvl) ------------------ | Branch (239:25): [True: 1.01M, False: 1.34M] | Branch (239:42): [True: 518k, False: 829k] ------------------ 240| 1.52M| dsp->cdef.fb[0](bptrs[0], f->cur.stride[0], lr_bak[bit][0], 241| 1.52M| top, bot, adj_y_pri_lvl, y_sec_lvl, 242| 1.52M| dir, damping, edges HIGHBD_CALL_SUFFIX); ------------------ | | 73| 1.52M|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 243| 2.36M| } else if (y_sec_lvl) ------------------ | Branch (243:28): [True: 198k, False: 82.3k] ------------------ 244| 198k| dsp->cdef.fb[0](bptrs[0], f->cur.stride[0], lr_bak[bit][0], 245| 198k| top, bot, 0, y_sec_lvl, 0, damping, 246| 198k| edges HIGHBD_CALL_SUFFIX); ------------------ | | 73| 198k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 247| | 248| 2.64M| if (!uv_lvl) goto skip_uv; ------------------ | Branch (248:21): [True: 1.20M, False: 1.44M] ------------------ 249| 2.64M| assert(layout != DAV1D_PIXEL_LAYOUT_I400); ------------------ | Branch (249:17): [True: 1.43M, False: 7.87k] ------------------ 250| | 251| 1.43M| const int uvdir = uv_pri_lvl ? uv_dir[dir] : 0; ------------------ | Branch (251:35): [True: 1.37M, False: 53.1k] ------------------ 252| 4.28M| for (int pl = 1; pl <= 2; pl++) { ------------------ | Branch (252:34): [True: 2.84M, False: 1.44M] ------------------ 253| 2.84M| if (!have_tt) goto st_uv; ------------------ | Branch (253:25): [True: 0, False: 2.84M] ------------------ 254| 2.84M| if (sbrow_start && by == by_start) { ------------------ | Branch (254:25): [True: 168k, False: 2.67M] | Branch (254:40): [True: 168k, False: 18.4E] ------------------ 255| 168k| if (resize) { ------------------ | Branch (255:29): [True: 48.8k, False: 119k] ------------------ 256| 48.8k| offset = (sby - 1) * 4 * uv_stride + (bx * 4 >> ss_hor); 257| 48.8k| top = &f->lf.cdef_lpf_line[pl][offset]; 258| 119k| } else { 259| 119k| const int line = sby * (4 << sb128) - 4; 260| 119k| offset = line * uv_stride + (bx * 4 >> ss_hor); 261| 119k| top = &f->lf.lr_lpf_line[pl][offset]; 262| 119k| } 263| 168k| bot = bptrs[pl] + (8 >> ss_ver) * uv_stride; 264| 2.68M| } else if (!sbrow_start && by + 2 >= by_end) { ------------------ | Branch (264:32): [True: 2.68M, False: 18.4E] | Branch (264:48): [True: 217k, False: 2.46M] ------------------ 265| 217k| const ptrdiff_t top_offset = sby * 8 * uv_stride + 266| 217k| (bx * 4 >> ss_hor); 267| 217k| top = &f->lf.cdef_line[tf][pl][top_offset]; 268| 217k| if (resize) { ------------------ | Branch (268:29): [True: 57.4k, False: 160k] ------------------ 269| 57.4k| offset = (sby * 4 + 2) * uv_stride + (bx * 4 >> ss_hor); 270| 57.4k| bot = &f->lf.cdef_lpf_line[pl][offset]; 271| 160k| } else { 272| 160k| const int line = sby * (4 << sb128) + 4 * sb128 + 2; 273| 160k| offset = line * uv_stride + (bx * 4 >> ss_hor); 274| 160k| bot = &f->lf.lr_lpf_line[pl][offset]; 275| 160k| } 276| 2.45M| } else { 277| 2.46M| st_uv:; 278| 2.46M| const ptrdiff_t offset = sby * 8 * uv_stride; 279| 2.46M| top = &f->lf.cdef_line[tf][pl][have_tt * offset + (bx * 4 >> ss_hor)]; 280| 2.46M| bot = bptrs[pl] + (8 >> ss_ver) * uv_stride; 281| 2.46M| } 282| 2.85M| dsp->cdef.fb[uv_idx](bptrs[pl], f->cur.stride[1], 283| 2.85M| lr_bak[bit][pl], top, bot, 284| 2.85M| uv_pri_lvl, uv_sec_lvl, uvdir, 285| 2.85M| damping - 1, edges HIGHBD_CALL_SUFFIX); ------------------ | | 73| 2.85M|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 286| 2.85M| } 287| | 288| 2.63M| skip_uv: 289| 2.63M| bit ^= 1; 290| | 291| 2.91M| next_b: 292| 2.91M| bptrs[0] += 8; 293| 2.91M| bptrs[1] += 8 >> ss_hor; 294| 2.91M| bptrs[2] += 8 >> ss_hor; 295| 2.91M| } 296| | 297| 25.0M| next_sb: 298| 25.0M| iptrs[0] += sbsz * 4; 299| 25.0M| iptrs[1] += sbsz * 4 >> ss_hor; 300| 25.0M| iptrs[2] += sbsz * 4 >> ss_hor; 301| 25.0M| } 302| | 303| 12.2M| ptrs[0] += 8 * PXSTRIDE(f->cur.stride[0]); 304| 12.2M| ptrs[1] += 8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver; 305| 12.2M| ptrs[2] += 8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver; 306| 12.2M| tc->top_pre_cdef_toggle ^= 1; 307| 12.2M| } 308| 1.57M|} dav1d_cdef_dsp_init_8bpc: 321| 3.46k|COLD void bitfn(dav1d_cdef_dsp_init)(Dav1dCdefDSPContext *const c) { 322| 3.46k| c->dir = cdef_find_dir_c; 323| 3.46k| c->fb[0] = cdef_filter_block_8x8_c; 324| 3.46k| c->fb[1] = cdef_filter_block_4x8_c; 325| 3.46k| c->fb[2] = cdef_filter_block_4x4_c; 326| | 327| 3.46k|#if HAVE_ASM 328| |#if ARCH_AARCH64 || ARCH_ARM 329| | cdef_dsp_init_arm(c); 330| |#elif ARCH_PPC64LE 331| | cdef_dsp_init_ppc(c); 332| |#elif ARCH_RISCV 333| | cdef_dsp_init_riscv(c); 334| |#elif ARCH_X86 335| | cdef_dsp_init_x86(c); 336| |#elif ARCH_LOONGARCH64 337| | cdef_dsp_init_loongarch(c); 338| |#endif 339| 3.46k|#endif 340| 3.46k|} dav1d_cdef_dsp_init_16bpc: 321| 5.11k|COLD void bitfn(dav1d_cdef_dsp_init)(Dav1dCdefDSPContext *const c) { 322| 5.11k| c->dir = cdef_find_dir_c; 323| 5.11k| c->fb[0] = cdef_filter_block_8x8_c; 324| 5.11k| c->fb[1] = cdef_filter_block_4x8_c; 325| 5.11k| c->fb[2] = cdef_filter_block_4x4_c; 326| | 327| 5.11k|#if HAVE_ASM 328| |#if ARCH_AARCH64 || ARCH_ARM 329| | cdef_dsp_init_arm(c); 330| |#elif ARCH_PPC64LE 331| | cdef_dsp_init_ppc(c); 332| |#elif ARCH_RISCV 333| | cdef_dsp_init_riscv(c); 334| |#elif ARCH_X86 335| | cdef_dsp_init_x86(c); 336| |#elif ARCH_LOONGARCH64 337| | cdef_dsp_init_loongarch(c); 338| |#endif 339| 5.11k|#endif 340| 5.11k|} dav1d_cdf_thread_update: 3918| 9.85k|{ 3919| 9.85k|#define update_cdf_1d(n1d, name) \ 3920| 9.85k| do { \ 3921| 9.85k| dst->name[n1d] = 0; \ 3922| 9.85k| } while (0) 3923| 9.85k|#define update_cdf_2d(n1d, n2d, name) \ 3924| 9.85k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) 3925| 9.85k|#define update_cdf_3d(n1d, n2d, n3d, name) \ 3926| 9.85k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) 3927| 9.85k|#define update_cdf_4d(n1d, n2d, n3d, n4d, name) \ 3928| 9.85k| for (int l = 0; l < (n1d); l++) update_cdf_3d(n2d, n3d, n4d, name[l]) 3929| | 3930| 9.85k| memcpy(dst, src, offsetof(CdfContext, m.intrabc)); 3931| | 3932| 9.85k| update_cdf_3d(2, 2, 4, coef.eob_bin_16); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 59.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 39.4k| do { \ | | | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | | | 3922| 39.4k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 39.4k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3933| 9.85k| update_cdf_3d(2, 2, 5, coef.eob_bin_32); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 59.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 39.4k| do { \ | | | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | | | 3922| 39.4k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 39.4k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3934| 9.85k| update_cdf_3d(2, 2, 6, coef.eob_bin_64); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 59.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 39.4k| do { \ | | | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | | | 3922| 39.4k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 39.4k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3935| 9.85k| update_cdf_3d(2, 2, 7, coef.eob_bin_128); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 59.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 39.4k| do { \ | | | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | | | 3922| 39.4k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 39.4k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3936| 9.85k| update_cdf_3d(2, 2, 8, coef.eob_bin_256); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 59.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 39.4k| do { \ | | | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | | | 3922| 39.4k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 39.4k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3937| 9.85k| update_cdf_2d(2, 9, coef.eob_bin_512); ------------------ | | 3924| 29.5k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 19.7k| do { \ | | | | 3921| 19.7k| dst->name[n1d] = 0; \ | | | | 3922| 19.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3938| 9.85k| update_cdf_2d(2, 10, coef.eob_bin_1024); ------------------ | | 3924| 29.5k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 19.7k| do { \ | | | | 3921| 19.7k| dst->name[n1d] = 0; \ | | | | 3922| 19.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3939| 9.85k| update_cdf_4d(N_TX_SIZES, 2, 4, 2, coef.eob_base_tok); ------------------ | | 3928| 59.1k| for (int l = 0; l < (n1d); l++) update_cdf_3d(n2d, n3d, n4d, name[l]) | | ------------------ | | | | 3926| 147k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | | | ------------------ | | | | | | 3924| 492k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | | | ------------------ | | | | | | | | 3920| 394k| do { \ | | | | | | | | 3921| 394k| dst->name[n1d] = 0; \ | | | | | | | | 3922| 394k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (3922:14): [Folded, False: 394k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | | Branch (3924:21): [True: 394k, False: 98.5k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3926:21): [True: 98.5k, False: 49.2k] | | | | ------------------ | | ------------------ | | | Branch (3928:21): [True: 49.2k, False: 9.85k] | | ------------------ ------------------ 3940| 9.85k| update_cdf_4d(N_TX_SIZES, 2, 41 /*42*/, 3, coef.base_tok); ------------------ | | 3928| 59.1k| for (int l = 0; l < (n1d); l++) update_cdf_3d(n2d, n3d, n4d, name[l]) | | ------------------ | | | | 3926| 147k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | | | ------------------ | | | | | | 3924| 4.13M| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | | | ------------------ | | | | | | | | 3920| 4.03M| do { \ | | | | | | | | 3921| 4.03M| dst->name[n1d] = 0; \ | | | | | | | | 3922| 4.03M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (3922:14): [Folded, False: 4.03M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | | Branch (3924:21): [True: 4.03M, False: 98.5k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3926:21): [True: 98.5k, False: 49.2k] | | | | ------------------ | | ------------------ | | | Branch (3928:21): [True: 49.2k, False: 9.85k] | | ------------------ ------------------ 3941| 9.85k| update_cdf_4d(4, 2, 21, 3, coef.br_tok); ------------------ | | 3928| 49.2k| for (int l = 0; l < (n1d); l++) update_cdf_3d(n2d, n3d, n4d, name[l]) | | ------------------ | | | | 3926| 118k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | | | ------------------ | | | | | | 3924| 1.73M| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | | | ------------------ | | | | | | | | 3920| 1.65M| do { \ | | | | | | | | 3921| 1.65M| dst->name[n1d] = 0; \ | | | | | | | | 3922| 1.65M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (3922:14): [Folded, False: 1.65M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | | Branch (3924:21): [True: 1.65M, False: 78.8k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3926:21): [True: 78.8k, False: 39.4k] | | | | ------------------ | | ------------------ | | | Branch (3928:21): [True: 39.4k, False: 9.85k] | | ------------------ ------------------ 3942| 9.85k| update_cdf_4d(N_TX_SIZES, 2, 9, 1, coef.eob_hi_bit); ------------------ | | 3928| 59.1k| for (int l = 0; l < (n1d); l++) update_cdf_3d(n2d, n3d, n4d, name[l]) | | ------------------ | | | | 3926| 147k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | | | ------------------ | | | | | | 3924| 985k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | | | ------------------ | | | | | | | | 3920| 886k| do { \ | | | | | | | | 3921| 886k| dst->name[n1d] = 0; \ | | | | | | | | 3922| 886k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (3922:14): [Folded, False: 886k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | | Branch (3924:21): [True: 886k, False: 98.5k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3926:21): [True: 98.5k, False: 49.2k] | | | | ------------------ | | ------------------ | | | Branch (3928:21): [True: 49.2k, False: 9.85k] | | ------------------ ------------------ 3943| 9.85k| update_cdf_3d(N_TX_SIZES, 13, 1, coef.skip); ------------------ | | 3926| 59.1k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 689k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 640k| do { \ | | | | | | 3921| 640k| dst->name[n1d] = 0; \ | | | | | | 3922| 640k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 640k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 640k, False: 49.2k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 49.2k, False: 9.85k] | | ------------------ ------------------ 3944| 9.85k| update_cdf_3d(2, 3, 1, coef.dc_sign); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 78.8k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 59.1k| do { \ | | | | | | 3921| 59.1k| dst->name[n1d] = 0; \ | | | | | | 3922| 59.1k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 59.1k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 59.1k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3945| | 3946| 9.85k| update_cdf_3d(2, N_INTRA_PRED_MODES, N_UV_INTRA_PRED_MODES - 1 - !k, m.uv_mode); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 275k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 256k| do { \ | | | | | | 3921| 256k| dst->name[n1d] = 0; \ | | | | | | 3922| 256k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 256k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 256k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3947| 9.85k| update_cdf_2d(4, N_PARTITIONS - 3, m.partition[BL_128X128]); ------------------ | | 3924| 49.2k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 39.4k| do { \ | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | 3922| 39.4k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 39.4k, False: 9.85k] | | ------------------ ------------------ 3948| 39.4k| for (int k = BL_64X64; k < BL_8X8; k++) ------------------ | Branch (3948:28): [True: 29.5k, False: 9.85k] ------------------ 3949| 29.5k| update_cdf_2d(4, N_PARTITIONS - 1, m.partition[k]); ------------------ | | 3924| 147k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 118k| do { \ | | | | 3921| 118k| dst->name[n1d] = 0; \ | | | | 3922| 118k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 118k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 118k, False: 29.5k] | | ------------------ ------------------ 3950| 9.85k| update_cdf_2d(4, N_SUB8X8_PARTITIONS - 1, m.partition[BL_8X8]); ------------------ | | 3924| 49.2k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 39.4k| do { \ | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | 3922| 39.4k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 39.4k, False: 9.85k] | | ------------------ ------------------ 3951| 9.85k| update_cdf_2d(6, 15, m.cfl_alpha); ------------------ | | 3924| 68.9k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 59.1k| do { \ | | | | 3921| 59.1k| dst->name[n1d] = 0; \ | | | | 3922| 59.1k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 59.1k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 59.1k, False: 9.85k] | | ------------------ ------------------ 3952| 9.85k| update_cdf_2d(2, 15, m.txtp_inter1); ------------------ | | 3924| 29.5k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 19.7k| do { \ | | | | 3921| 19.7k| dst->name[n1d] = 0; \ | | | | 3922| 19.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3953| 9.85k| update_cdf_1d(11, m.txtp_inter2); ------------------ | | 3920| 9.85k| do { \ | | 3921| 9.85k| dst->name[n1d] = 0; \ | | 3922| 9.85k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 9.85k] | | ------------------ ------------------ 3954| 9.85k| update_cdf_3d(2, N_INTRA_PRED_MODES, 6, m.txtp_intra1); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 275k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 256k| do { \ | | | | | | 3921| 256k| dst->name[n1d] = 0; \ | | | | | | 3922| 256k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 256k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 256k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3955| 9.85k| update_cdf_3d(3, N_INTRA_PRED_MODES, 4, m.txtp_intra2); ------------------ | | 3926| 39.4k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 413k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 384k| do { \ | | | | | | 3921| 384k| dst->name[n1d] = 0; \ | | | | | | 3922| 384k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 384k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 384k, False: 29.5k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 29.5k, False: 9.85k] | | ------------------ ------------------ 3956| 9.85k| update_cdf_1d(7, m.cfl_sign); ------------------ | | 3920| 9.85k| do { \ | | 3921| 9.85k| dst->name[n1d] = 0; \ | | 3922| 9.85k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 9.85k] | | ------------------ ------------------ 3957| 9.85k| update_cdf_2d(8, 6, m.angle_delta); ------------------ | | 3924| 88.6k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 78.8k| do { \ | | | | 3921| 78.8k| dst->name[n1d] = 0; \ | | | | 3922| 78.8k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 78.8k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 78.8k, False: 9.85k] | | ------------------ ------------------ 3958| 9.85k| update_cdf_1d(4, m.filter_intra); ------------------ | | 3920| 9.85k| do { \ | | 3921| 9.85k| dst->name[n1d] = 0; \ | | 3922| 9.85k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 9.85k] | | ------------------ ------------------ 3959| 9.85k| update_cdf_2d(3, DAV1D_MAX_SEGMENTS - 1, m.seg_id); ------------------ | | 3924| 39.4k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 29.5k| do { \ | | | | 3921| 29.5k| dst->name[n1d] = 0; \ | | | | 3922| 29.5k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 29.5k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 29.5k, False: 9.85k] | | ------------------ ------------------ 3960| 9.85k| update_cdf_3d(2, 7, 6, m.pal_sz); ------------------ | | 3926| 29.5k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 157k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 137k| do { \ | | | | | | 3921| 137k| dst->name[n1d] = 0; \ | | | | | | 3922| 137k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 137k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 137k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3961| 9.85k| update_cdf_4d(2, 7, 5, k + 1, m.color_map); ------------------ | | 3928| 29.5k| for (int l = 0; l < (n1d); l++) update_cdf_3d(n2d, n3d, n4d, name[l]) | | ------------------ | | | | 3926| 157k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | | | ------------------ | | | | | | 3924| 827k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | | | ------------------ | | | | | | | | 3920| 689k| do { \ | | | | | | | | 3921| 689k| dst->name[n1d] = 0; \ | | | | | | | | 3922| 689k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (3922:14): [Folded, False: 689k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | | Branch (3924:21): [True: 689k, False: 137k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3926:21): [True: 137k, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3928:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3962| 9.85k| update_cdf_3d(N_TX_SIZES - 1, 3, imin(k + 1, 2), m.txsz); ------------------ | | 3926| 49.2k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 157k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 118k| do { \ | | | | | | 3921| 118k| dst->name[n1d] = 0; \ | | | | | | 3922| 118k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 118k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 118k, False: 39.4k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 39.4k, False: 9.85k] | | ------------------ ------------------ 3963| 9.85k| update_cdf_1d(3, m.delta_q); ------------------ | | 3920| 9.85k| do { \ | | 3921| 9.85k| dst->name[n1d] = 0; \ | | 3922| 9.85k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 9.85k] | | ------------------ ------------------ 3964| 9.85k| update_cdf_2d(5, 3, m.delta_lf); ------------------ | | 3924| 59.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 49.2k| do { \ | | | | 3921| 49.2k| dst->name[n1d] = 0; \ | | | | 3922| 49.2k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 49.2k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 49.2k, False: 9.85k] | | ------------------ ------------------ 3965| 9.85k| update_cdf_1d(2, m.restore_switchable); ------------------ | | 3920| 9.85k| do { \ | | 3921| 9.85k| dst->name[n1d] = 0; \ | | 3922| 9.85k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 9.85k] | | ------------------ ------------------ 3966| 9.85k| update_cdf_1d(1, m.restore_wiener); ------------------ | | 3920| 9.85k| do { \ | | 3921| 9.85k| dst->name[n1d] = 0; \ | | 3922| 9.85k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 9.85k] | | ------------------ ------------------ 3967| 9.85k| update_cdf_1d(1, m.restore_sgrproj); ------------------ | | 3920| 9.85k| do { \ | | 3921| 9.85k| dst->name[n1d] = 0; \ | | 3922| 9.85k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 9.85k] | | ------------------ ------------------ 3968| 9.85k| update_cdf_2d(4, 1, m.txtp_inter3); ------------------ | | 3924| 49.2k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 39.4k| do { \ | | | | 3921| 39.4k| dst->name[n1d] = 0; \ | | | | 3922| 39.4k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 39.4k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 39.4k, False: 9.85k] | | ------------------ ------------------ 3969| 9.85k| update_cdf_2d(N_BS_SIZES, 1, m.use_filter_intra); ------------------ | | 3924| 226k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 216k| do { \ | | | | 3921| 216k| dst->name[n1d] = 0; \ | | | | 3922| 216k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 216k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 216k, False: 9.85k] | | ------------------ ------------------ 3970| 9.85k| update_cdf_3d(7, 3, 1, m.txpart); ------------------ | | 3926| 78.8k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 275k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 206k| do { \ | | | | | | 3921| 206k| dst->name[n1d] = 0; \ | | | | | | 3922| 206k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 206k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 206k, False: 68.9k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 68.9k, False: 9.85k] | | ------------------ ------------------ 3971| 9.85k| update_cdf_2d(3, 1, m.skip); ------------------ | | 3924| 39.4k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 29.5k| do { \ | | | | 3921| 29.5k| dst->name[n1d] = 0; \ | | | | 3922| 29.5k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 29.5k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 29.5k, False: 9.85k] | | ------------------ ------------------ 3972| 9.85k| update_cdf_3d(7, 3, 1, m.pal_y); ------------------ | | 3926| 78.8k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 275k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 206k| do { \ | | | | | | 3921| 206k| dst->name[n1d] = 0; \ | | | | | | 3922| 206k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 206k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 206k, False: 68.9k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 68.9k, False: 9.85k] | | ------------------ ------------------ 3973| 9.85k| update_cdf_2d(2, 1, m.pal_uv); ------------------ | | 3924| 29.5k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 19.7k| do { \ | | | | 3921| 19.7k| dst->name[n1d] = 0; \ | | | | 3922| 19.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 19.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 19.7k, False: 9.85k] | | ------------------ ------------------ 3974| | 3975| 9.85k| if (IS_KEY_OR_INTRA(hdr)) ------------------ | | 43| 9.85k| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 9.85k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (43:5): [True: 5.59k, False: 4.26k] | | ------------------ ------------------ 3976| 5.59k| return; 3977| | 3978| 4.26k| memcpy(dst->m.y_mode, src->m.y_mode, 3979| 4.26k| offsetof(CdfContext, kfym) - offsetof(CdfContext, m.y_mode)); 3980| | 3981| 4.26k| update_cdf_2d(4, N_INTRA_PRED_MODES - 1, m.y_mode); ------------------ | | 3924| 21.3k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 21.3k| do { \ | | | | 3921| 17.0k| dst->name[n1d] = 0; \ | | | | 3922| 17.0k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 17.0k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 17.0k, False: 4.26k] | | ------------------ ------------------ 3982| 4.26k| update_cdf_2d(9, 15, m.wedge_idx); ------------------ | | 3924| 42.6k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 42.6k| do { \ | | | | 3921| 38.3k| dst->name[n1d] = 0; \ | | | | 3922| 38.3k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 38.3k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 38.3k, False: 4.26k] | | ------------------ ------------------ 3983| 4.26k| update_cdf_2d(8, N_COMP_INTER_PRED_MODES - 1, m.comp_inter_mode); ------------------ | | 3924| 38.3k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 38.3k| do { \ | | | | 3921| 34.0k| dst->name[n1d] = 0; \ | | | | 3922| 34.0k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 34.0k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 34.0k, False: 4.26k] | | ------------------ ------------------ 3984| 4.26k| update_cdf_3d(2, 8, DAV1D_N_SWITCHABLE_FILTERS - 1, m.filter); ------------------ | | 3926| 12.7k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 76.7k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 72.4k| do { \ | | | | | | 3921| 68.1k| dst->name[n1d] = 0; \ | | | | | | 3922| 68.1k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 68.1k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 68.1k, False: 8.52k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 8.52k, False: 4.26k] | | ------------------ ------------------ 3985| 4.26k| update_cdf_2d(4, 3, m.interintra_mode); ------------------ | | 3924| 21.3k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 21.3k| do { \ | | | | 3921| 17.0k| dst->name[n1d] = 0; \ | | | | 3922| 17.0k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 17.0k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 17.0k, False: 4.26k] | | ------------------ ------------------ 3986| 4.26k| update_cdf_2d(N_BS_SIZES, 2, m.motion_mode); ------------------ | | 3924| 98.0k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 98.0k| do { \ | | | | 3921| 93.7k| dst->name[n1d] = 0; \ | | | | 3922| 93.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 93.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 93.7k, False: 4.26k] | | ------------------ ------------------ 3987| 4.26k| update_cdf_2d(3, 1, m.skip_mode); ------------------ | | 3924| 17.0k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 17.0k| do { \ | | | | 3921| 12.7k| dst->name[n1d] = 0; \ | | | | 3922| 12.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 12.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 12.7k, False: 4.26k] | | ------------------ ------------------ 3988| 4.26k| update_cdf_2d(6, 1, m.newmv_mode); ------------------ | | 3924| 29.8k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 29.8k| do { \ | | | | 3921| 25.5k| dst->name[n1d] = 0; \ | | | | 3922| 25.5k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 25.5k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 25.5k, False: 4.26k] | | ------------------ ------------------ 3989| 4.26k| update_cdf_2d(2, 1, m.globalmv_mode); ------------------ | | 3924| 12.7k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 12.7k| do { \ | | | | 3921| 8.52k| dst->name[n1d] = 0; \ | | | | 3922| 8.52k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 8.52k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 8.52k, False: 4.26k] | | ------------------ ------------------ 3990| 4.26k| update_cdf_2d(6, 1, m.refmv_mode); ------------------ | | 3924| 29.8k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 29.8k| do { \ | | | | 3921| 25.5k| dst->name[n1d] = 0; \ | | | | 3922| 25.5k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 25.5k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 25.5k, False: 4.26k] | | ------------------ ------------------ 3991| 4.26k| update_cdf_2d(3, 1, m.drl_bit); ------------------ | | 3924| 17.0k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 17.0k| do { \ | | | | 3921| 12.7k| dst->name[n1d] = 0; \ | | | | 3922| 12.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 12.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 12.7k, False: 4.26k] | | ------------------ ------------------ 3992| 4.26k| update_cdf_2d(4, 1, m.intra); ------------------ | | 3924| 21.3k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 21.3k| do { \ | | | | 3921| 17.0k| dst->name[n1d] = 0; \ | | | | 3922| 17.0k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 17.0k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 17.0k, False: 4.26k] | | ------------------ ------------------ 3993| 4.26k| update_cdf_2d(5, 1, m.comp); ------------------ | | 3924| 25.5k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 25.5k| do { \ | | | | 3921| 21.3k| dst->name[n1d] = 0; \ | | | | 3922| 21.3k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 21.3k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 21.3k, False: 4.26k] | | ------------------ ------------------ 3994| 4.26k| update_cdf_2d(5, 1, m.comp_dir); ------------------ | | 3924| 25.5k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 25.5k| do { \ | | | | 3921| 21.3k| dst->name[n1d] = 0; \ | | | | 3922| 21.3k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 21.3k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 21.3k, False: 4.26k] | | ------------------ ------------------ 3995| 4.26k| update_cdf_2d(6, 1, m.jnt_comp); ------------------ | | 3924| 29.8k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 29.8k| do { \ | | | | 3921| 25.5k| dst->name[n1d] = 0; \ | | | | 3922| 25.5k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 25.5k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 25.5k, False: 4.26k] | | ------------------ ------------------ 3996| 4.26k| update_cdf_2d(6, 1, m.mask_comp); ------------------ | | 3924| 29.8k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 29.8k| do { \ | | | | 3921| 25.5k| dst->name[n1d] = 0; \ | | | | 3922| 25.5k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 25.5k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 25.5k, False: 4.26k] | | ------------------ ------------------ 3997| 4.26k| update_cdf_2d(9, 1, m.wedge_comp); ------------------ | | 3924| 42.6k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 42.6k| do { \ | | | | 3921| 38.3k| dst->name[n1d] = 0; \ | | | | 3922| 38.3k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 38.3k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 38.3k, False: 4.26k] | | ------------------ ------------------ 3998| 4.26k| update_cdf_3d(6, 3, 1, m.ref); ------------------ | | 3926| 29.8k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 102k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 80.9k| do { \ | | | | | | 3921| 76.7k| dst->name[n1d] = 0; \ | | | | | | 3922| 76.7k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 76.7k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 76.7k, False: 25.5k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 25.5k, False: 4.26k] | | ------------------ ------------------ 3999| 4.26k| update_cdf_3d(3, 3, 1, m.comp_fwd_ref); ------------------ | | 3926| 17.0k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 51.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 42.6k| do { \ | | | | | | 3921| 38.3k| dst->name[n1d] = 0; \ | | | | | | 3922| 38.3k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 38.3k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 38.3k, False: 12.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 12.7k, False: 4.26k] | | ------------------ ------------------ 4000| 4.26k| update_cdf_3d(2, 3, 1, m.comp_bwd_ref); ------------------ | | 3926| 12.7k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 34.0k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 29.8k| do { \ | | | | | | 3921| 25.5k| dst->name[n1d] = 0; \ | | | | | | 3922| 25.5k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 25.5k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 25.5k, False: 8.52k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 8.52k, False: 4.26k] | | ------------------ ------------------ 4001| 4.26k| update_cdf_3d(3, 3, 1, m.comp_uni_ref); ------------------ | | 3926| 17.0k| for (int k = 0; k < (n1d); k++) update_cdf_2d(n2d, n3d, name[k]) | | ------------------ | | | | 3924| 51.1k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | | | ------------------ | | | | | | 3920| 42.6k| do { \ | | | | | | 3921| 38.3k| dst->name[n1d] = 0; \ | | | | | | 3922| 38.3k| } while (0) | | | | | | ------------------ | | | | | | | Branch (3922:14): [Folded, False: 38.3k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (3924:21): [True: 38.3k, False: 12.7k] | | | | ------------------ | | ------------------ | | | Branch (3926:21): [True: 12.7k, False: 4.26k] | | ------------------ ------------------ 4002| 4.26k| update_cdf_2d(3, 1, m.seg_pred); ------------------ | | 3924| 17.0k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 17.0k| do { \ | | | | 3921| 12.7k| dst->name[n1d] = 0; \ | | | | 3922| 12.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 12.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 12.7k, False: 4.26k] | | ------------------ ------------------ 4003| 4.26k| update_cdf_2d(4, 1, m.interintra); ------------------ | | 3924| 21.3k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 21.3k| do { \ | | | | 3921| 17.0k| dst->name[n1d] = 0; \ | | | | 3922| 17.0k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 17.0k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 17.0k, False: 4.26k] | | ------------------ ------------------ 4004| 4.26k| update_cdf_2d(7, 1, m.interintra_wedge); ------------------ | | 3924| 34.0k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 34.0k| do { \ | | | | 3921| 29.8k| dst->name[n1d] = 0; \ | | | | 3922| 29.8k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 29.8k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 29.8k, False: 4.26k] | | ------------------ ------------------ 4005| 4.26k| update_cdf_2d(N_BS_SIZES, 1, m.obmc); ------------------ | | 3924| 98.0k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 98.0k| do { \ | | | | 3921| 93.7k| dst->name[n1d] = 0; \ | | | | 3922| 93.7k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 93.7k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 93.7k, False: 4.26k] | | ------------------ ------------------ 4006| | 4007| 12.7k| for (int k = 0; k < 2; k++) { ------------------ | Branch (4007:21): [True: 8.52k, False: 4.26k] ------------------ 4008| 8.52k| update_cdf_1d(10, mv.comp[k].classes); ------------------ | | 3920| 8.52k| do { \ | | 3921| 8.52k| dst->name[n1d] = 0; \ | | 3922| 8.52k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 8.52k] | | ------------------ ------------------ 4009| 8.52k| update_cdf_1d(1, mv.comp[k].sign); ------------------ | | 3920| 8.52k| do { \ | | 3921| 8.52k| dst->name[n1d] = 0; \ | | 3922| 8.52k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 8.52k] | | ------------------ ------------------ 4010| 8.52k| update_cdf_1d(1, mv.comp[k].class0); ------------------ | | 3920| 8.52k| do { \ | | 3921| 8.52k| dst->name[n1d] = 0; \ | | 3922| 8.52k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 8.52k] | | ------------------ ------------------ 4011| 8.52k| update_cdf_2d(2, 3, mv.comp[k].class0_fp); ------------------ | | 3924| 25.5k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 17.0k| do { \ | | | | 3921| 17.0k| dst->name[n1d] = 0; \ | | | | 3922| 17.0k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 17.0k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 17.0k, False: 8.52k] | | ------------------ ------------------ 4012| 8.52k| update_cdf_1d(1, mv.comp[k].class0_hp); ------------------ | | 3920| 8.52k| do { \ | | 3921| 8.52k| dst->name[n1d] = 0; \ | | 3922| 8.52k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 8.52k] | | ------------------ ------------------ 4013| 8.52k| update_cdf_2d(10, 1, mv.comp[k].classN); ------------------ | | 3924| 93.7k| for (int j = 0; j < (n1d); j++) update_cdf_1d(n2d, name[j]) | | ------------------ | | | | 3920| 85.2k| do { \ | | | | 3921| 85.2k| dst->name[n1d] = 0; \ | | | | 3922| 85.2k| } while (0) | | | | ------------------ | | | | | Branch (3922:14): [Folded, False: 85.2k] | | | | ------------------ | | ------------------ | | | Branch (3924:21): [True: 85.2k, False: 8.52k] | | ------------------ ------------------ 4014| 8.52k| update_cdf_1d(3, mv.comp[k].classN_fp); ------------------ | | 3920| 8.52k| do { \ | | 3921| 8.52k| dst->name[n1d] = 0; \ | | 3922| 8.52k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 8.52k] | | ------------------ ------------------ 4015| 8.52k| update_cdf_1d(1, mv.comp[k].classN_hp); ------------------ | | 3920| 8.52k| do { \ | | 3921| 8.52k| dst->name[n1d] = 0; \ | | 3922| 8.52k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 8.52k] | | ------------------ ------------------ 4016| 8.52k| } 4017| 4.26k| update_cdf_1d(N_MV_JOINTS - 1, mv.joint); ------------------ | | 3920| 4.26k| do { \ | | 3921| 4.26k| dst->name[n1d] = 0; \ | | 3922| 4.26k| } while (0) | | ------------------ | | | Branch (3922:14): [Folded, False: 4.26k] | | ------------------ ------------------ 4018| 4.26k|} dav1d_cdf_thread_init_static: 4023| 278k|void dav1d_cdf_thread_init_static(CdfThreadContext *const cdf, const unsigned qidx) { 4024| | cdf->ref = NULL; 4025| 278k| cdf->data.qcat = (qidx > 20) + (qidx > 60) + (qidx > 120); 4026| 278k|} dav1d_cdf_thread_copy: 4028| 336k|void dav1d_cdf_thread_copy(CdfContext *const dst, const CdfThreadContext *const src) { 4029| 336k| if (src->ref) { ------------------ | Branch (4029:9): [True: 33.8k, False: 302k] ------------------ 4030| 33.8k| memcpy(dst, src->data.cdf, sizeof(*dst)); 4031| 302k| } else { 4032| 302k| dst->coef = default_coef_cdf[src->data.qcat]; 4033| 302k| memcpy(&dst->m, &default_cdf.m, 4034| 302k| offsetof(CdfDefaultContext, mv.joint)); 4035| 302k| memcpy(&dst->mv.comp[1], &default_cdf.mv.comp, 4036| | sizeof(default_cdf) - offsetof(CdfDefaultContext, mv.comp)); 4037| 302k| } 4038| 336k|} dav1d_cdf_thread_alloc: 4042| 55.3k|{ 4043| 55.3k| cdf->ref = dav1d_ref_create_using_pool(c->cdf_pool, 4044| 55.3k| sizeof(CdfContext) + sizeof(atomic_uint)); 4045| 55.3k| if (!cdf->ref) return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (4045:9): [True: 0, False: 55.3k] ------------------ 4046| 55.3k| cdf->data.cdf = cdf->ref->data; 4047| 55.3k| if (have_frame_mt) { ------------------ | Branch (4047:9): [True: 55.3k, False: 0] ------------------ 4048| 55.3k| cdf->progress = (atomic_uint *) &cdf->data.cdf[1]; 4049| | atomic_init(cdf->progress, 0); 4050| 55.3k| } 4051| 55.3k| return 0; 4052| 55.3k|} dav1d_cdf_thread_ref: 4056| 2.57M|{ 4057| 2.57M| *dst = *src; 4058| 2.57M| if (src->ref) ------------------ | Branch (4058:9): [True: 463k, False: 2.11M] ------------------ 4059| 463k| dav1d_ref_inc(src->ref); 4060| 2.57M|} dav1d_cdf_thread_unref: 4062| 3.16M|void dav1d_cdf_thread_unref(CdfThreadContext *const cdf) { 4063| | memset(&cdf->data, 0, sizeof(*cdf) - offsetof(CdfThreadContext, data)); 4064| 3.16M| dav1d_ref_dec(&cdf->ref); 4065| 3.16M|} dav1d_init_cpu: 63| 1|COLD void dav1d_init_cpu(void) { 64| 1|#if HAVE_ASM && !__has_feature(memory_sanitizer) 65| |// memory sanitizer is inherently incompatible with asm 66| |#if ARCH_AARCH64 || ARCH_ARM 67| | dav1d_cpu_flags = dav1d_get_cpu_flags_arm(); 68| |#elif ARCH_LOONGARCH 69| | dav1d_cpu_flags = dav1d_get_cpu_flags_loongarch(); 70| |#elif ARCH_PPC64LE 71| | dav1d_cpu_flags = dav1d_get_cpu_flags_ppc(); 72| |#elif ARCH_RISCV 73| | dav1d_cpu_flags = dav1d_get_cpu_flags_riscv(); 74| |#elif ARCH_X86 75| | dav1d_cpu_flags = dav1d_get_cpu_flags_x86(); 76| 1|#endif 77| 1|#endif 78| 1|} cpu.c:dav1d_get_default_cpu_flags: 58| 1|static ALWAYS_INLINE unsigned dav1d_get_default_cpu_flags(void) { 59| 1| unsigned flags = 0; 60| | 61| |#if ARCH_AARCH64 || ARCH_ARM 62| |#if defined(__ARM_NEON) || defined(__APPLE__) || defined(_WIN32) || ARCH_AARCH64 63| | flags |= DAV1D_ARM_CPU_FLAG_NEON; 64| |#endif 65| |#ifdef __ARM_FEATURE_DOTPROD 66| | flags |= DAV1D_ARM_CPU_FLAG_DOTPROD; 67| |#endif 68| |#ifdef __ARM_FEATURE_MATMUL_INT8 69| | flags |= DAV1D_ARM_CPU_FLAG_I8MM; 70| |#endif 71| |#if ARCH_AARCH64 72| |#ifdef __ARM_FEATURE_SVE 73| | flags |= DAV1D_ARM_CPU_FLAG_SVE; 74| |#endif 75| |#ifdef __ARM_FEATURE_SVE2 76| | flags |= DAV1D_ARM_CPU_FLAG_SVE2; 77| |#endif 78| |#endif /* ARCH_AARCH64 */ 79| |#elif ARCH_PPC64LE 80| |#if defined(__VSX__) 81| | flags |= DAV1D_PPC_CPU_FLAG_VSX; 82| |#endif 83| |#if defined(__POWER9_VECTOR__) 84| | flags |= DAV1D_PPC_CPU_FLAG_PWR9; 85| |#endif 86| |#elif ARCH_RISCV 87| |#if defined(__riscv_v) 88| | flags |= DAV1D_RISCV_CPU_FLAG_V; 89| |#endif 90| |#elif ARCH_X86 91| |#if defined(__AVX512F__) && defined(__AVX512CD__) && \ 92| | defined(__AVX512BW__) && defined(__AVX512DQ__) && \ 93| | defined(__AVX512VL__) && defined(__AVX512VNNI__) && \ 94| | defined(__AVX512IFMA__) && defined(__AVX512VBMI__) && \ 95| | defined(__AVX512VBMI2__) && defined(__AVX512VPOPCNTDQ__) && \ 96| | defined(__AVX512BITALG__) && defined(__GFNI__) && \ 97| | defined(__VAES__) && defined(__VPCLMULQDQ__) 98| | flags |= DAV1D_X86_CPU_FLAG_AVX512ICL | 99| | DAV1D_X86_CPU_FLAG_AVX2 | 100| | DAV1D_X86_CPU_FLAG_SSE41 | 101| | DAV1D_X86_CPU_FLAG_SSSE3 | 102| | DAV1D_X86_CPU_FLAG_SSE2; 103| |#elif defined(__AVX2__) 104| | flags |= DAV1D_X86_CPU_FLAG_AVX2 | 105| | DAV1D_X86_CPU_FLAG_SSE41 | 106| | DAV1D_X86_CPU_FLAG_SSSE3 | 107| | DAV1D_X86_CPU_FLAG_SSE2; 108| |#elif defined(__SSE4_1__) || defined(__AVX__) 109| | flags |= DAV1D_X86_CPU_FLAG_SSE41 | 110| | DAV1D_X86_CPU_FLAG_SSSE3 | 111| | DAV1D_X86_CPU_FLAG_SSE2; 112| |#elif defined(__SSSE3__) 113| | flags |= DAV1D_X86_CPU_FLAG_SSSE3 | 114| | DAV1D_X86_CPU_FLAG_SSE2; 115| |#elif ARCH_X86_64 || defined(__SSE2__) || \ 116| | (defined(_M_IX86_FP) && _M_IX86_FP >= 2) 117| | flags |= DAV1D_X86_CPU_FLAG_SSE2; 118| 1|#endif 119| 1|#endif 120| | 121| 1| return flags; 122| 1|} pal.c:dav1d_get_cpu_flags: 124| 9.41k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 9.41k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 9.41k| return flags; 134| 9.41k|} refmvs.c:dav1d_get_cpu_flags: 124| 9.41k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 9.41k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 9.41k| return flags; 134| 9.41k|} msac.c:dav1d_get_cpu_flags: 124| 315k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 315k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 315k| return flags; 134| 315k|} cdef_tmpl.c:dav1d_get_cpu_flags: 124| 8.57k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 8.57k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 8.57k| return flags; 134| 8.57k|} filmgrain_tmpl.c:dav1d_get_cpu_flags: 124| 8.57k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 8.57k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 8.57k| return flags; 134| 8.57k|} ipred_tmpl.c:dav1d_get_cpu_flags: 124| 8.57k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 8.57k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 8.57k| return flags; 134| 8.57k|} itx_tmpl.c:dav1d_get_cpu_flags: 124| 8.57k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 8.57k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 8.57k| return flags; 134| 8.57k|} loopfilter_tmpl.c:dav1d_get_cpu_flags: 124| 8.57k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 8.57k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 8.57k| return flags; 134| 8.57k|} looprestoration_tmpl.c:dav1d_get_cpu_flags: 124| 8.57k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 8.57k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 8.57k| return flags; 134| 8.57k|} mc_tmpl.c:dav1d_get_cpu_flags: 124| 8.57k|static ALWAYS_INLINE unsigned dav1d_get_cpu_flags(void) { 125| 8.57k| unsigned flags = dav1d_cpu_flags & dav1d_cpu_flags_mask; 126| | 127| |#if TRIM_DSP_FUNCTIONS 128| |/* Since this function is inlined, unconditionally setting a flag here will 129| | * enable dead code elimination in the calling function. */ 130| | flags |= dav1d_get_default_cpu_flags(); 131| |#endif 132| | 133| 8.57k| return flags; 134| 8.57k|} ctx.c:memset_w1: 34| 86.8M|static void memset_w1(void *const ptr, const int value) { 35| 86.8M| set_ctx1((uint8_t *) ptr, 0, value); ------------------ | | 56| 86.8M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 ------------------ 36| 86.8M|} ctx.c:memset_w2: 38| 25.7M|static void memset_w2(void *const ptr, const int value) { 39| 25.7M| set_ctx2((uint8_t *) ptr, 0, value); ------------------ | | 58| 25.7M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 ------------------ 40| 25.7M|} ctx.c:memset_w4: 42| 19.2M|static void memset_w4(void *const ptr, const int value) { 43| 19.2M| set_ctx4((uint8_t *) ptr, 0, value); ------------------ | | 60| 19.2M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U ------------------ 44| 19.2M|} ctx.c:memset_w8: 46| 14.7M|static void memset_w8(void *const ptr, const int value) { 47| 14.7M| set_ctx8((uint8_t *) ptr, 0, value); ------------------ | | 62| 14.7M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL ------------------ 48| 14.7M|} ctx.c:memset_w16: 50| 15.8M|static void memset_w16(void *const ptr, const int value) { 51| 15.8M| set_ctx16((uint8_t *) ptr, 0, value); ------------------ | | 63| 15.8M|#define set_ctx16(var, off, val) do { \ | | 64| 15.8M| memset(&(var)[off], val, 16); \ | | 65| 15.8M| } while (0) | | ------------------ | | | Branch (65:14): [Folded, False: 15.8M] | | ------------------ ------------------ 52| 15.8M|} ctx.c:memset_w32: 54| 4.03M|static void memset_w32(void *const ptr, const int value) { 55| 4.03M| set_ctx32((uint8_t *) ptr, 0, value); ------------------ | | 66| 4.03M|#define set_ctx32(var, off, val) do { \ | | 67| 4.03M| memset(&(var)[off], val, 32); \ | | 68| 4.03M| } while (0) | | ------------------ | | | Branch (68:14): [Folded, False: 4.03M] | | ------------------ ------------------ 56| 4.03M|} lf_mask.c:dav1d_memset_likely_pow2: 44| 7.37M|static inline void dav1d_memset_likely_pow2(void *const ptr, const int value, const int n) { 45| 7.37M| assert(n >= 1 && n <= 32); ------------------ | Branch (45:5): [True: 7.37M, False: 812] | Branch (45:5): [True: 7.37M, False: 366] ------------------ 46| 7.37M| if ((n&(n-1)) == 0) { ------------------ | Branch (46:9): [True: 6.75M, False: 623k] ------------------ 47| 6.75M| dav1d_memset_pow2[ulog2(n)](ptr, value); 48| 6.75M| } else { 49| 623k| memset(ptr, value, n); 50| 623k| } 51| 7.37M|} recon_tmpl.c:dav1d_memset_likely_pow2: 44| 99.1M|static inline void dav1d_memset_likely_pow2(void *const ptr, const int value, const int n) { 45| 99.1M| assert(n >= 1 && n <= 32); ------------------ | Branch (45:5): [True: 99.0M, False: 13.6k] | Branch (45:5): [True: 99.1M, False: 18.4E] ------------------ 46| 99.1M| if ((n&(n-1)) == 0) { ------------------ | Branch (46:9): [True: 97.5M, False: 1.56M] ------------------ 47| 97.5M| dav1d_memset_pow2[ulog2(n)](ptr, value); 48| 97.5M| } else { 49| 1.56M| memset(ptr, value, n); 50| 1.56M| } 51| 99.1M|} dav1d_data_create_internal: 43| 411k|uint8_t *dav1d_data_create_internal(Dav1dData *const buf, const size_t sz) { 44| 411k| validate_input_or_ret(buf != NULL, NULL); ------------------ | | 52| 411k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 411k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 45| | 46| 411k| if (sz > SIZE_MAX / 2) return NULL; ------------------ | Branch (46:9): [True: 0, False: 411k] ------------------ 47| 411k| buf->ref = dav1d_ref_create(ALLOC_DAV1DDATA, sz); ------------------ | | 49| 411k|#define dav1d_ref_create(type, size) dav1d_ref_create(size) ------------------ 48| 411k| if (!buf->ref) return NULL; ------------------ | Branch (48:9): [True: 0, False: 411k] ------------------ 49| 411k| buf->data = buf->ref->const_data; 50| 411k| buf->sz = sz; 51| 411k| dav1d_data_props_set_defaults(&buf->m); 52| 411k| buf->m.size = sz; 53| | 54| 411k| return buf->ref->data; 55| 411k|} dav1d_data_ref: 98| 771k|void dav1d_data_ref(Dav1dData *const dst, const Dav1dData *const src) { 99| 771k| assert(dst != NULL); ------------------ | Branch (99:5): [True: 771k, False: 0] ------------------ 100| 771k| assert(dst->data == NULL); ------------------ | Branch (100:5): [True: 771k, False: 0] ------------------ 101| 771k| assert(src != NULL); ------------------ | Branch (101:5): [True: 771k, False: 0] ------------------ 102| | 103| 771k| if (src->ref) { ------------------ | Branch (103:9): [True: 771k, False: 0] ------------------ 104| 771k| assert(src->data != NULL); ------------------ | Branch (104:9): [True: 771k, False: 0] ------------------ 105| 771k| dav1d_ref_inc(src->ref); 106| 771k| } 107| 771k| if (src->m.user_data.ref) dav1d_ref_inc(src->m.user_data.ref); ------------------ | Branch (107:9): [True: 0, False: 771k] ------------------ 108| 771k| *dst = *src; 109| 771k|} dav1d_data_props_copy: 113| 658k|{ 114| 658k| assert(dst != NULL); ------------------ | Branch (114:5): [True: 658k, False: 0] ------------------ 115| 658k| assert(src != NULL); ------------------ | Branch (115:5): [True: 658k, False: 0] ------------------ 116| | 117| 658k| dav1d_ref_dec(&dst->user_data.ref); 118| 658k| *dst = *src; 119| 658k| if (dst->user_data.ref) dav1d_ref_inc(dst->user_data.ref); ------------------ | Branch (119:9): [True: 0, False: 658k] ------------------ 120| 658k|} dav1d_data_props_set_defaults: 122| 6.75M|void dav1d_data_props_set_defaults(Dav1dDataProps *const props) { 123| 6.75M| assert(props != NULL); ------------------ | Branch (123:5): [True: 6.75M, False: 9] ------------------ 124| | 125| 6.75M| memset(props, 0, sizeof(*props)); 126| | props->timestamp = INT64_MIN; 127| 6.75M| props->offset = -1; 128| 6.75M|} dav1d_data_props_unref_internal: 130| 9.41k|void dav1d_data_props_unref_internal(Dav1dDataProps *const props) { 131| 9.41k| validate_input(props != NULL); ------------------ | | 59| 9.41k|#define validate_input(x) validate_input_or_ret(x, ) | | ------------------ | | | | 52| 9.41k| if (!(x)) { \ | | | | ------------------ | | | | | Branch (52:9): [True: 0, False: 9.41k] | | | | ------------------ | | | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | | | ------------------ | | | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | | | ------------------ | | | | 54| 0| #x, __func__); \ | | | | 55| 0| debug_abort(); \ | | | | ------------------ | | | | | | 39| 0|#define debug_abort abort | | | | ------------------ | | | | 56| 0| return r; \ | | | | 57| 0| } | | ------------------ ------------------ 132| | 133| 9.41k| struct Dav1dRef *user_data_ref = props->user_data.ref; 134| 9.41k| dav1d_data_props_set_defaults(props); 135| 9.41k| dav1d_ref_dec(&user_data_ref); 136| 9.41k|} dav1d_data_unref_internal: 138| 1.19M|void dav1d_data_unref_internal(Dav1dData *const buf) { 139| 1.19M| validate_input(buf != NULL); ------------------ | | 59| 1.19M|#define validate_input(x) validate_input_or_ret(x, ) | | ------------------ | | | | 52| 1.19M| if (!(x)) { \ | | | | ------------------ | | | | | Branch (52:9): [True: 0, False: 1.19M] | | | | ------------------ | | | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | | | ------------------ | | | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | | | ------------------ | | | | 54| 0| #x, __func__); \ | | | | 55| 0| debug_abort(); \ | | | | ------------------ | | | | | | 39| 0|#define debug_abort abort | | | | ------------------ | | | | 56| 0| return r; \ | | | | 57| 0| } | | ------------------ ------------------ 140| | 141| 1.19M| struct Dav1dRef *user_data_ref = buf->m.user_data.ref; 142| 1.19M| if (buf->ref) { ------------------ | Branch (142:9): [True: 1.18M, False: 9.40k] ------------------ 143| 1.18M| validate_input(buf->data != NULL); ------------------ | | 59| 1.18M|#define validate_input(x) validate_input_or_ret(x, ) | | ------------------ | | | | 52| 1.18M| if (!(x)) { \ | | | | ------------------ | | | | | Branch (52:9): [True: 0, False: 1.18M] | | | | ------------------ | | | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | | | ------------------ | | | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | | | ------------------ | | | | 54| 0| #x, __func__); \ | | | | 55| 0| debug_abort(); \ | | | | ------------------ | | | | | | 39| 0|#define debug_abort abort | | | | ------------------ | | | | 56| 0| return r; \ | | | | 57| 0| } | | ------------------ ------------------ 144| 1.18M| dav1d_ref_dec(&buf->ref); 145| 1.18M| } 146| 1.19M| memset(buf, 0, sizeof(*buf)); 147| 1.19M| dav1d_data_props_set_defaults(&buf->m); 148| 1.19M| dav1d_ref_dec(&user_data_ref); 149| 1.19M|} dav1d_decode_tile_sbrow: 2597| 4.12M|int dav1d_decode_tile_sbrow(Dav1dTaskContext *const t) { 2598| 4.12M| const Dav1dFrameContext *const f = t->f; 2599| 4.12M| const enum BlockLevel root_bl = f->seq_hdr->sb128 ? BL_128X128 : BL_64X64; ------------------ | Branch (2599:37): [True: 2.10M, False: 2.02M] ------------------ 2600| 4.12M| Dav1dTileState *const ts = t->ts; 2601| 4.12M| const Dav1dContext *const c = f->c; 2602| 4.12M| const int sb_step = f->sb_step; 2603| 4.12M| const int tile_row = ts->tiling.row, tile_col = ts->tiling.col; 2604| 4.12M| const int col_sb_start = f->frame_hdr->tiling.col_start_sb[tile_col]; 2605| 4.12M| const int col_sb128_start = col_sb_start >> !f->seq_hdr->sb128; 2606| | 2607| 4.12M| if (IS_INTER_OR_SWITCH(f->frame_hdr) || f->frame_hdr->allow_intrabc) { ------------------ | | 36| 8.25M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 3.05M, False: 1.06M] | | ------------------ ------------------ | Branch (2607:45): [True: 958k, False: 110k] ------------------ 2608| 4.01M| dav1d_refmvs_tile_sbrow_init(&t->rt, &f->rf, ts->tiling.col_start, 2609| 4.01M| ts->tiling.col_end, ts->tiling.row_start, 2610| 4.01M| ts->tiling.row_end, t->by >> f->sb_shift, 2611| 4.01M| ts->tiling.row, t->frame_thread.pass); 2612| 4.01M| } 2613| | 2614| 4.12M| if (IS_INTER_OR_SWITCH(f->frame_hdr) && c->n_fc > 1) { ------------------ | | 36| 8.25M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 3.05M, False: 1.07M] | | ------------------ ------------------ | Branch (2614:45): [True: 3.05M, False: 18.4E] ------------------ 2615| 3.05M| const int sby = (t->by - ts->tiling.row_start) >> f->sb_shift; 2616| 3.05M| int (*const lowest_px)[2] = ts->lowest_pixel[sby]; 2617| 24.4M| for (int n = 0; n < 7; n++) ------------------ | Branch (2617:25): [True: 21.3M, False: 3.05M] ------------------ 2618| 64.1M| for (int m = 0; m < 2; m++) ------------------ | Branch (2618:29): [True: 42.7M, False: 21.3M] ------------------ 2619| 42.7M| lowest_px[n][m] = INT_MIN; 2620| 3.05M| } 2621| | 2622| 4.12M| reset_context(&t->l, IS_KEY_OR_INTRA(f->frame_hdr), t->frame_thread.pass); ------------------ | | 43| 4.12M| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 4.12M| ((frame_header)->frame_type & 1) | | ------------------ ------------------ 2623| 4.12M| if (t->frame_thread.pass == 2) { ------------------ | Branch (2623:9): [True: 1.98M, False: 2.13M] ------------------ 2624| 18.4E| const int off_2pass = c->n_tc > 1 ? f->sb128w * f->frame_hdr->tiling.rows : 0; ------------------ | Branch (2624:31): [True: 1.98M, False: 18.4E] ------------------ 2625| 1.98M| for (t->bx = ts->tiling.col_start, 2626| 1.98M| t->a = f->a + off_2pass + col_sb128_start + tile_row * f->sb128w; 2627| 4.41M| t->bx < ts->tiling.col_end; t->bx += sb_step) ------------------ | Branch (2627:14): [True: 2.42M, False: 1.98M] ------------------ 2628| 2.42M| { 2629| 2.42M| if (atomic_load_explicit(c->flush, memory_order_acquire)) ------------------ | Branch (2629:17): [True: 7, False: 2.42M] ------------------ 2630| 7| return 1; 2631| 2.42M| if (decode_sb(t, root_bl, dav1d_intra_edge_tree[root_bl])) ------------------ | Branch (2631:17): [True: 0, False: 2.42M] ------------------ 2632| 0| return 1; 2633| 2.42M| if (t->bx & 16 || f->seq_hdr->sb128) ------------------ | Branch (2633:17): [True: 653k, False: 1.77M] | Branch (2633:31): [True: 1.09M, False: 675k] ------------------ 2634| 1.74M| t->a++; 2635| 2.42M| } 2636| 1.98M| f->bd_fn.backup_ipred_edge(t); 2637| 1.98M| return 0; 2638| 1.98M| } 2639| | 2640| 2.13M| if (f->c->n_tc > 1 && f->frame_hdr->use_ref_frame_mvs) { ------------------ | Branch (2640:9): [True: 2.13M, False: 576] | Branch (2640:27): [True: 827k, False: 1.30M] ------------------ 2641| 827k| f->c->refmvs_dsp.load_tmvs(&f->rf, ts->tiling.row, 2642| 827k| ts->tiling.col_start >> 1, ts->tiling.col_end >> 1, 2643| 827k| t->by >> 1, (t->by + sb_step) >> 1); 2644| 827k| } 2645| 2.13M| memset(t->pal_sz_uv[1], 0, sizeof(*t->pal_sz_uv)); 2646| 2.13M| const int sb128y = t->by >> 5; 2647| 2.13M| for (t->bx = ts->tiling.col_start, t->a = f->a + col_sb128_start + tile_row * f->sb128w, 2648| 2.13M| t->lf_mask = f->lf.mask + sb128y * f->sb128w + col_sb128_start; 2649| 5.23M| t->bx < ts->tiling.col_end; t->bx += sb_step) ------------------ | Branch (2649:10): [True: 3.17M, False: 2.06M] ------------------ 2650| 3.17M| { 2651| 3.17M| if (atomic_load_explicit(c->flush, memory_order_acquire)) ------------------ | Branch (2651:13): [True: 377, False: 3.17M] ------------------ 2652| 377| return 1; 2653| 3.17M| if (root_bl == BL_128X128) { ------------------ | Branch (2653:13): [True: 1.53M, False: 1.63M] ------------------ 2654| 1.53M| t->cur_sb_cdef_idx_ptr = t->lf_mask->cdef_idx; 2655| 1.53M| t->cur_sb_cdef_idx_ptr[0] = -1; 2656| 1.53M| t->cur_sb_cdef_idx_ptr[1] = -1; 2657| 1.53M| t->cur_sb_cdef_idx_ptr[2] = -1; 2658| 1.53M| t->cur_sb_cdef_idx_ptr[3] = -1; 2659| 1.63M| } else { 2660| 1.63M| t->cur_sb_cdef_idx_ptr = 2661| 1.63M| &t->lf_mask->cdef_idx[((t->bx & 16) >> 4) + 2662| 1.63M| ((t->by & 16) >> 3)]; 2663| 1.63M| t->cur_sb_cdef_idx_ptr[0] = -1; 2664| 1.63M| } 2665| | // Restoration filter 2666| 12.6M| for (int p = 0; p < 3; p++) { ------------------ | Branch (2666:25): [True: 9.50M, False: 3.17M] ------------------ 2667| 9.50M| if (!((f->lf.restore_planes >> p) & 1U)) ------------------ | Branch (2667:17): [True: 8.80M, False: 694k] ------------------ 2668| 8.80M| continue; 2669| | 2670| 694k| const int ss_ver = p && f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (2670:32): [True: 206k, False: 488k] | Branch (2670:37): [True: 150k, False: 55.7k] ------------------ 2671| 694k| const int ss_hor = p && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (2671:32): [True: 206k, False: 488k] | Branch (2671:37): [True: 162k, False: 43.9k] ------------------ 2672| 694k| const int unit_size_log2 = f->frame_hdr->restoration.unit_size[!!p]; 2673| 694k| const int y = t->by * 4 >> ss_ver; 2674| 694k| const int h = (f->cur.p.h + ss_ver) >> ss_ver; 2675| | 2676| 694k| const int unit_size = 1 << unit_size_log2; 2677| 694k| const unsigned mask = unit_size - 1; 2678| 694k| if (y & mask) continue; ------------------ | Branch (2678:17): [True: 248k, False: 446k] ------------------ 2679| 446k| const int half_unit = unit_size >> 1; 2680| | // Round half up at frame boundaries, if there's more than one 2681| | // restoration unit 2682| 446k| if (y && y + half_unit > h) continue; ------------------ | Branch (2682:17): [True: 252k, False: 193k] | Branch (2682:22): [True: 11.9k, False: 240k] ------------------ 2683| | 2684| 434k| const enum Dav1dRestorationType frame_type = f->frame_hdr->restoration.type[p]; 2685| | 2686| 434k| if (f->frame_hdr->width[0] != f->frame_hdr->width[1]) { ------------------ | Branch (2686:17): [True: 45.2k, False: 389k] ------------------ 2687| 45.2k| const int w = (f->sr_cur.p.p.w + ss_hor) >> ss_hor; 2688| 45.2k| const int n_units = imax(1, (w + half_unit) >> unit_size_log2); 2689| | 2690| 45.2k| const int d = f->frame_hdr->super_res.width_scale_denominator; 2691| 45.2k| const int rnd = unit_size * 8 - 1, shift = unit_size_log2 + 3; 2692| 45.2k| const int x0 = ((4 * t->bx * d >> ss_hor) + rnd) >> shift; 2693| 45.2k| const int x1 = ((4 * (t->bx + sb_step) * d >> ss_hor) + rnd) >> shift; 2694| | 2695| 88.5k| for (int x = x0; x < imin(x1, n_units); x++) { ------------------ | Branch (2695:34): [True: 43.2k, False: 45.2k] ------------------ 2696| 43.2k| const int px_x = x << (unit_size_log2 + ss_hor); 2697| 43.2k| const int sb_idx = (t->by >> 5) * f->sr_sb128w + (px_x >> 7); 2698| 43.2k| const int unit_idx = ((t->by & 16) >> 3) + ((px_x & 64) >> 6); 2699| 43.2k| Av1RestorationUnit *const lr = &f->lf.lr_mask[sb_idx].lr[p][unit_idx]; 2700| | 2701| 43.2k| read_restoration_info(t, lr, p, frame_type); 2702| 43.2k| } 2703| 389k| } else { 2704| 389k| const int x = 4 * t->bx >> ss_hor; 2705| 389k| if (x & mask) continue; ------------------ | Branch (2705:21): [True: 171k, False: 217k] ------------------ 2706| 217k| const int w = (f->cur.p.w + ss_hor) >> ss_hor; 2707| | // Round half up at frame boundaries, if there's more than one 2708| | // restoration unit 2709| 217k| if (x && x + half_unit > w) continue; ------------------ | Branch (2709:21): [True: 99.1k, False: 118k] | Branch (2709:26): [True: 1.87k, False: 97.3k] ------------------ 2710| 215k| const int sb_idx = (t->by >> 5) * f->sr_sb128w + (t->bx >> 5); 2711| 215k| const int unit_idx = ((t->by & 16) >> 3) + ((t->bx & 16) >> 4); 2712| 215k| Av1RestorationUnit *const lr = &f->lf.lr_mask[sb_idx].lr[p][unit_idx]; 2713| | 2714| 215k| read_restoration_info(t, lr, p, frame_type); 2715| 215k| } 2716| 434k| } 2717| 3.17M| if (decode_sb(t, root_bl, dav1d_intra_edge_tree[root_bl])) ------------------ | Branch (2717:13): [True: 70.9k, False: 3.10M] ------------------ 2718| 70.9k| return 1; 2719| 3.10M| if (t->bx & 16 || f->seq_hdr->sb128) { ------------------ | Branch (2719:13): [True: 770k, False: 2.32M] | Branch (2719:27): [True: 1.52M, False: 805k] ------------------ 2720| 2.29M| t->a++; 2721| 2.29M| t->lf_mask++; 2722| 2.29M| } 2723| 3.10M| } 2724| | 2725| 2.06M| if (f->seq_hdr->ref_frame_mvs && f->c->n_tc > 1 && IS_INTER_OR_SWITCH(f->frame_hdr)) { ------------------ | | 36| 975k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 832k, False: 142k] | | ------------------ ------------------ | Branch (2725:9): [True: 975k, False: 1.09M] | Branch (2725:38): [True: 975k, False: 18.4E] ------------------ 2726| 832k| dav1d_refmvs_save_tmvs(&f->c->refmvs_dsp, &t->rt, 2727| 832k| ts->tiling.col_start >> 1, ts->tiling.col_end >> 1, 2728| 832k| t->by >> 1, (t->by + sb_step) >> 1); 2729| 832k| } 2730| | 2731| | // backup pre-loopfilter pixels for intra prediction of the next sbrow 2732| 2.06M| if (t->frame_thread.pass != 1) ------------------ | Branch (2732:9): [True: 0, False: 2.06M] ------------------ 2733| 0| f->bd_fn.backup_ipred_edge(t); 2734| | 2735| | // backup t->a/l.tx_lpf_y/uv at tile boundaries to use them to "fix" 2736| | // up the initial value in neighbour tiles when running the loopfilter 2737| 2.06M| int align_h = (f->bh + 31) & ~31; 2738| 2.06M| memcpy(&f->lf.tx_lpf_right_edge[0][align_h * tile_col + t->by], 2739| 2.06M| &t->l.tx_lpf_y[t->by & 16], sb_step); 2740| 2.06M| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2741| 2.06M| align_h >>= ss_ver; 2742| 2.06M| memcpy(&f->lf.tx_lpf_right_edge[1][align_h * tile_col + (t->by >> ss_ver)], 2743| 2.06M| &t->l.tx_lpf_uv[(t->by & 16) >> ss_ver], sb_step >> ss_ver); 2744| | 2745| | // error out on symbol decoder overread 2746| 2.06M| if (ts->msac.cnt <= -15) return 1; ------------------ | Branch (2746:9): [True: 38.1k, False: 2.02M] ------------------ 2747| | 2748| 2.02M| return c->strict_std_compliance && ------------------ | Branch (2748:12): [True: 0, False: 2.02M] ------------------ 2749| 0| (t->by >> f->sb_shift) + 1 >= f->frame_hdr->tiling.row_start_sb[tile_row + 1] && ------------------ | Branch (2749:12): [True: 0, False: 0] ------------------ 2750| 0| check_trailing_bits_after_symbol_coder(&ts->msac); ------------------ | Branch (2750:12): [True: 0, False: 0] ------------------ 2751| 2.06M|} dav1d_decode_frame_init: 2753| 355k|int dav1d_decode_frame_init(Dav1dFrameContext *const f) { 2754| 355k| const Dav1dContext *const c = f->c; 2755| 355k| int retval = DAV1D_ERR(ENOMEM); ------------------ | | 58| 355k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 2756| | 2757| 355k| if (f->sbh > f->lf.start_of_tile_row_sz) { ------------------ | Branch (2757:9): [True: 17.0k, False: 338k] ------------------ 2758| 17.0k| dav1d_free(f->lf.start_of_tile_row); ------------------ | | 135| 17.0k|#define dav1d_free(ptr) free(ptr) ------------------ 2759| 17.0k| f->lf.start_of_tile_row = dav1d_malloc(ALLOC_TILE, f->sbh * sizeof(uint8_t)); ------------------ | | 132| 17.0k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 2760| 17.0k| if (!f->lf.start_of_tile_row) { ------------------ | Branch (2760:13): [True: 0, False: 17.0k] ------------------ 2761| 0| f->lf.start_of_tile_row_sz = 0; 2762| 0| goto error; 2763| 0| } 2764| 17.0k| f->lf.start_of_tile_row_sz = f->sbh; 2765| 17.0k| } 2766| 355k| int sby = 0; 2767| 780k| for (int tile_row = 0; tile_row < f->frame_hdr->tiling.rows; tile_row++) { ------------------ | Branch (2767:28): [True: 424k, False: 355k] ------------------ 2768| 424k| f->lf.start_of_tile_row[sby++] = tile_row; 2769| 10.9M| while (sby < f->frame_hdr->tiling.row_start_sb[tile_row + 1]) ------------------ | Branch (2769:16): [True: 10.5M, False: 424k] ------------------ 2770| 10.5M| f->lf.start_of_tile_row[sby++] = 0; 2771| 424k| } 2772| | 2773| 355k| const int n_ts = f->frame_hdr->tiling.cols * f->frame_hdr->tiling.rows; 2774| 355k| if (n_ts != f->n_ts) { ------------------ | Branch (2774:9): [True: 43.1k, False: 312k] ------------------ 2775| 43.1k| if (c->n_fc > 1) { ------------------ | Branch (2775:13): [True: 43.1k, False: 18.4E] ------------------ 2776| 43.1k| dav1d_free(f->frame_thread.tile_start_off); ------------------ | | 135| 43.1k|#define dav1d_free(ptr) free(ptr) ------------------ 2777| 43.1k| f->frame_thread.tile_start_off = 2778| 43.1k| dav1d_malloc(ALLOC_TILE, sizeof(*f->frame_thread.tile_start_off) * n_ts); ------------------ | | 132| 43.1k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 2779| 43.1k| if (!f->frame_thread.tile_start_off) { ------------------ | Branch (2779:17): [True: 0, False: 43.1k] ------------------ 2780| 0| f->n_ts = 0; 2781| 0| goto error; 2782| 0| } 2783| 43.1k| } 2784| 43.1k| dav1d_free_aligned(f->ts); ------------------ | | 136| 43.1k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 2785| 43.1k| f->ts = dav1d_alloc_aligned(ALLOC_TILE, sizeof(*f->ts) * n_ts, 32); ------------------ | | 134| 43.1k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 2786| 43.1k| if (!f->ts) goto error; ------------------ | Branch (2786:13): [True: 0, False: 43.1k] ------------------ 2787| 43.1k| f->n_ts = n_ts; 2788| 43.1k| } 2789| | 2790| 355k| const int a_sz = f->sb128w * f->frame_hdr->tiling.rows * (1 + (c->n_fc > 1 && c->n_tc > 1)); ------------------ | Branch (2790:68): [True: 355k, False: 143] | Branch (2790:83): [True: 355k, False: 18.4E] ------------------ 2791| 355k| if (a_sz != f->a_sz) { ------------------ | Branch (2791:9): [True: 43.5k, False: 311k] ------------------ 2792| 43.5k| dav1d_free(f->a); ------------------ | | 135| 43.5k|#define dav1d_free(ptr) free(ptr) ------------------ 2793| 43.5k| f->a = dav1d_malloc(ALLOC_TILE, sizeof(*f->a) * a_sz); ------------------ | | 132| 43.5k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 2794| 43.5k| if (!f->a) { ------------------ | Branch (2794:13): [True: 0, False: 43.5k] ------------------ 2795| 0| f->a_sz = 0; 2796| 0| goto error; 2797| 0| } 2798| 43.5k| f->a_sz = a_sz; 2799| 43.5k| } 2800| | 2801| 355k| const int num_sb128 = f->sb128w * f->sb128h; 2802| 355k| const uint8_t *const size_mul = ss_size_mul[f->cur.p.layout]; 2803| 355k| const int hbd = !!f->seq_hdr->hbd; 2804| 355k| if (c->n_fc > 1) { ------------------ | Branch (2804:9): [True: 355k, False: 94] ------------------ 2805| 355k| const unsigned sb_step4 = f->sb_step * 4; 2806| 355k| int tile_idx = 0; 2807| 779k| for (int tile_row = 0; tile_row < f->frame_hdr->tiling.rows; tile_row++) { ------------------ | Branch (2807:32): [True: 424k, False: 355k] ------------------ 2808| 424k| const unsigned row_off = f->frame_hdr->tiling.row_start_sb[tile_row] * 2809| 424k| sb_step4 * f->sb128w * 128; 2810| 424k| const unsigned b_diff = (f->frame_hdr->tiling.row_start_sb[tile_row + 1] - 2811| 424k| f->frame_hdr->tiling.row_start_sb[tile_row]) * sb_step4; 2812| 877k| for (int tile_col = 0; tile_col < f->frame_hdr->tiling.cols; tile_col++) { ------------------ | Branch (2812:36): [True: 453k, False: 424k] ------------------ 2813| 453k| f->frame_thread.tile_start_off[tile_idx++] = row_off + b_diff * 2814| 453k| f->frame_hdr->tiling.col_start_sb[tile_col] * sb_step4; 2815| 453k| } 2816| 424k| } 2817| | 2818| 355k| const int lowest_pixel_mem_sz = f->frame_hdr->tiling.cols * f->sbh; 2819| 355k| if (lowest_pixel_mem_sz != f->tile_thread.lowest_pixel_mem_sz) { ------------------ | Branch (2819:13): [True: 29.4k, False: 325k] ------------------ 2820| 29.4k| dav1d_free(f->tile_thread.lowest_pixel_mem); ------------------ | | 135| 29.4k|#define dav1d_free(ptr) free(ptr) ------------------ 2821| 29.4k| f->tile_thread.lowest_pixel_mem = 2822| 29.4k| dav1d_malloc(ALLOC_TILE, lowest_pixel_mem_sz * ------------------ | | 132| 29.4k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 2823| 29.4k| sizeof(*f->tile_thread.lowest_pixel_mem)); 2824| 29.4k| if (!f->tile_thread.lowest_pixel_mem) { ------------------ | Branch (2824:17): [True: 0, False: 29.4k] ------------------ 2825| 0| f->tile_thread.lowest_pixel_mem_sz = 0; 2826| 0| goto error; 2827| 0| } 2828| 29.4k| f->tile_thread.lowest_pixel_mem_sz = lowest_pixel_mem_sz; 2829| 29.4k| } 2830| 355k| int (*lowest_pixel_ptr)[7][2] = f->tile_thread.lowest_pixel_mem; 2831| 779k| for (int tile_row = 0, tile_row_base = 0; tile_row < f->frame_hdr->tiling.rows; ------------------ | Branch (2831:51): [True: 424k, False: 355k] ------------------ 2832| 424k| tile_row++, tile_row_base += f->frame_hdr->tiling.cols) 2833| 424k| { 2834| 424k| const int tile_row_sb_h = f->frame_hdr->tiling.row_start_sb[tile_row + 1] - 2835| 424k| f->frame_hdr->tiling.row_start_sb[tile_row]; 2836| 877k| for (int tile_col = 0; tile_col < f->frame_hdr->tiling.cols; tile_col++) { ------------------ | Branch (2836:36): [True: 453k, False: 424k] ------------------ 2837| 453k| f->ts[tile_row_base + tile_col].lowest_pixel = lowest_pixel_ptr; 2838| 453k| lowest_pixel_ptr += tile_row_sb_h; 2839| 453k| } 2840| 424k| } 2841| | 2842| 355k| const int cbi_sz = num_sb128 * size_mul[0]; 2843| 355k| if (cbi_sz != f->frame_thread.cbi_sz) { ------------------ | Branch (2843:13): [True: 24.3k, False: 330k] ------------------ 2844| 24.3k| dav1d_free_aligned(f->frame_thread.cbi); ------------------ | | 136| 24.3k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 2845| 24.3k| f->frame_thread.cbi = 2846| 24.3k| dav1d_alloc_aligned(ALLOC_BLOCK, sizeof(*f->frame_thread.cbi) * ------------------ | | 134| 24.3k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 2847| 24.3k| cbi_sz * 32 * 32 / 4, 64); 2848| 24.3k| if (!f->frame_thread.cbi) { ------------------ | Branch (2848:17): [True: 0, False: 24.3k] ------------------ 2849| 0| f->frame_thread.cbi_sz = 0; 2850| 0| goto error; 2851| 0| } 2852| 24.3k| f->frame_thread.cbi_sz = cbi_sz; 2853| 24.3k| } 2854| | 2855| 355k| const int cf_sz = (num_sb128 * size_mul[0]) << hbd; 2856| 355k| if (cf_sz != f->frame_thread.cf_sz) { ------------------ | Branch (2856:13): [True: 25.6k, False: 329k] ------------------ 2857| 25.6k| dav1d_free_aligned(f->frame_thread.cf); ------------------ | | 136| 25.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 2858| 25.6k| f->frame_thread.cf = 2859| 25.6k| dav1d_alloc_aligned(ALLOC_COEF, (size_t)cf_sz * 128 * 128 / 2, 64); ------------------ | | 134| 25.6k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 2860| 25.6k| if (!f->frame_thread.cf) { ------------------ | Branch (2860:17): [True: 0, False: 25.6k] ------------------ 2861| 0| f->frame_thread.cf_sz = 0; 2862| 0| goto error; 2863| 0| } 2864| 25.6k| memset(f->frame_thread.cf, 0, (size_t)cf_sz * 128 * 128 / 2); 2865| 25.6k| f->frame_thread.cf_sz = cf_sz; 2866| 25.6k| } 2867| | 2868| 355k| if (f->frame_hdr->allow_screen_content_tools) { ------------------ | Branch (2868:13): [True: 282k, False: 72.3k] ------------------ 2869| 282k| const int pal_sz = num_sb128 << hbd; 2870| 282k| if (pal_sz != f->frame_thread.pal_sz) { ------------------ | Branch (2870:17): [True: 13.8k, False: 269k] ------------------ 2871| 13.8k| dav1d_free_aligned(f->frame_thread.pal); ------------------ | | 136| 13.8k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 2872| 13.8k| f->frame_thread.pal = 2873| 13.8k| dav1d_alloc_aligned(ALLOC_PAL, sizeof(*f->frame_thread.pal) * ------------------ | | 134| 13.8k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 2874| 13.8k| pal_sz * 16 * 16, 64); 2875| 13.8k| if (!f->frame_thread.pal) { ------------------ | Branch (2875:21): [True: 0, False: 13.8k] ------------------ 2876| 0| f->frame_thread.pal_sz = 0; 2877| 0| goto error; 2878| 0| } 2879| 13.8k| f->frame_thread.pal_sz = pal_sz; 2880| 13.8k| } 2881| | 2882| 282k| const int pal_idx_sz = num_sb128 * size_mul[1]; 2883| 282k| if (pal_idx_sz != f->frame_thread.pal_idx_sz) { ------------------ | Branch (2883:17): [True: 12.9k, False: 269k] ------------------ 2884| 12.9k| dav1d_free_aligned(f->frame_thread.pal_idx); ------------------ | | 136| 12.9k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 2885| 12.9k| f->frame_thread.pal_idx = 2886| 12.9k| dav1d_alloc_aligned(ALLOC_PAL, sizeof(*f->frame_thread.pal_idx) * ------------------ | | 134| 12.9k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 2887| 12.9k| pal_idx_sz * 128 * 128 / 8, 64); 2888| 12.9k| if (!f->frame_thread.pal_idx) { ------------------ | Branch (2888:21): [True: 0, False: 12.9k] ------------------ 2889| 0| f->frame_thread.pal_idx_sz = 0; 2890| 0| goto error; 2891| 0| } 2892| 12.9k| f->frame_thread.pal_idx_sz = pal_idx_sz; 2893| 12.9k| } 2894| 282k| } else if (f->frame_thread.pal) { ------------------ | Branch (2894:20): [True: 1.94k, False: 70.3k] ------------------ 2895| 1.94k| dav1d_freep_aligned(&f->frame_thread.pal); 2896| 1.94k| dav1d_freep_aligned(&f->frame_thread.pal_idx); 2897| 1.94k| f->frame_thread.pal_sz = f->frame_thread.pal_idx_sz = 0; 2898| 1.94k| } 2899| 355k| } 2900| | 2901| | // update allocation of block contexts for above 2902| 355k| ptrdiff_t y_stride = f->cur.stride[0], uv_stride = f->cur.stride[1]; 2903| 355k| const int has_resize = f->frame_hdr->width[0] != f->frame_hdr->width[1]; 2904| 355k| const int need_cdef_lpf_copy = c->n_tc > 1 && has_resize; ------------------ | Branch (2904:36): [True: 355k, False: 147] | Branch (2904:51): [True: 38.8k, False: 316k] ------------------ 2905| 355k| if (y_stride * f->sbh * 4 != f->lf.cdef_buf_plane_sz[0] || ------------------ | Branch (2905:9): [True: 24.3k, False: 330k] ------------------ 2906| 330k| uv_stride * f->sbh * 8 != f->lf.cdef_buf_plane_sz[1] || ------------------ | Branch (2906:9): [True: 346, False: 330k] ------------------ 2907| 330k| need_cdef_lpf_copy != f->lf.need_cdef_lpf_copy || ------------------ | Branch (2907:9): [True: 362, False: 330k] ------------------ 2908| 330k| f->sbh != f->lf.cdef_buf_sbh) ------------------ | Branch (2908:9): [True: 1.17k, False: 329k] ------------------ 2909| 26.0k| { 2910| 26.0k| dav1d_free_aligned(f->lf.cdef_line_buf); ------------------ | | 136| 26.0k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 2911| 26.0k| size_t alloc_sz = 64; 2912| 26.0k| alloc_sz += (size_t)llabs(y_stride) * 4 * f->sbh << need_cdef_lpf_copy; 2913| 26.0k| alloc_sz += (size_t)llabs(uv_stride) * 8 * f->sbh << need_cdef_lpf_copy; 2914| 26.0k| uint8_t *ptr = f->lf.cdef_line_buf = dav1d_alloc_aligned(ALLOC_CDEF, alloc_sz, 32); ------------------ | | 134| 26.0k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 2915| 26.0k| if (!ptr) { ------------------ | Branch (2915:13): [True: 0, False: 26.0k] ------------------ 2916| 0| f->lf.cdef_buf_plane_sz[0] = f->lf.cdef_buf_plane_sz[1] = 0; 2917| 0| goto error; 2918| 0| } 2919| | 2920| 26.0k| ptr += 32; 2921| 26.0k| if (y_stride < 0) { ------------------ | Branch (2921:13): [True: 0, False: 26.0k] ------------------ 2922| 0| f->lf.cdef_line[0][0] = ptr - y_stride * (f->sbh * 4 - 1); 2923| 0| f->lf.cdef_line[1][0] = ptr - y_stride * (f->sbh * 4 - 3); 2924| 26.0k| } else { 2925| 26.0k| f->lf.cdef_line[0][0] = ptr + y_stride * 0; 2926| 26.0k| f->lf.cdef_line[1][0] = ptr + y_stride * 2; 2927| 26.0k| } 2928| 26.0k| ptr += llabs(y_stride) * f->sbh * 4; 2929| 26.0k| if (uv_stride < 0) { ------------------ | Branch (2929:13): [True: 0, False: 26.0k] ------------------ 2930| 0| f->lf.cdef_line[0][1] = ptr - uv_stride * (f->sbh * 8 - 1); 2931| 0| f->lf.cdef_line[0][2] = ptr - uv_stride * (f->sbh * 8 - 3); 2932| 0| f->lf.cdef_line[1][1] = ptr - uv_stride * (f->sbh * 8 - 5); 2933| 0| f->lf.cdef_line[1][2] = ptr - uv_stride * (f->sbh * 8 - 7); 2934| 26.0k| } else { 2935| 26.0k| f->lf.cdef_line[0][1] = ptr + uv_stride * 0; 2936| 26.0k| f->lf.cdef_line[0][2] = ptr + uv_stride * 2; 2937| 26.0k| f->lf.cdef_line[1][1] = ptr + uv_stride * 4; 2938| 26.0k| f->lf.cdef_line[1][2] = ptr + uv_stride * 6; 2939| 26.0k| } 2940| | 2941| 26.0k| if (need_cdef_lpf_copy) { ------------------ | Branch (2941:13): [True: 6.48k, False: 19.5k] ------------------ 2942| 6.48k| ptr += llabs(uv_stride) * f->sbh * 8; 2943| 6.48k| if (y_stride < 0) ------------------ | Branch (2943:17): [True: 0, False: 6.48k] ------------------ 2944| 0| f->lf.cdef_lpf_line[0] = ptr - y_stride * (f->sbh * 4 - 1); 2945| 6.48k| else 2946| 6.48k| f->lf.cdef_lpf_line[0] = ptr; 2947| 6.48k| ptr += llabs(y_stride) * f->sbh * 4; 2948| 6.48k| if (uv_stride < 0) { ------------------ | Branch (2948:17): [True: 0, False: 6.48k] ------------------ 2949| 0| f->lf.cdef_lpf_line[1] = ptr - uv_stride * (f->sbh * 4 - 1); 2950| 0| f->lf.cdef_lpf_line[2] = ptr - uv_stride * (f->sbh * 8 - 1); 2951| 6.48k| } else { 2952| 6.48k| f->lf.cdef_lpf_line[1] = ptr; 2953| 6.48k| f->lf.cdef_lpf_line[2] = ptr + uv_stride * f->sbh * 4; 2954| 6.48k| } 2955| 6.48k| } 2956| | 2957| 26.0k| f->lf.cdef_buf_plane_sz[0] = (int) y_stride * f->sbh * 4; 2958| 26.0k| f->lf.cdef_buf_plane_sz[1] = (int) uv_stride * f->sbh * 8; 2959| 26.0k| f->lf.need_cdef_lpf_copy = need_cdef_lpf_copy; 2960| 26.0k| f->lf.cdef_buf_sbh = f->sbh; 2961| 26.0k| } 2962| | 2963| 355k| const int sb128 = f->seq_hdr->sb128; 2964| 355k| const int num_lines = c->n_tc > 1 ? f->sbh * 4 << sb128 : 12; ------------------ | Branch (2964:27): [True: 355k, False: 212] ------------------ 2965| 355k| y_stride = f->sr_cur.p.stride[0], uv_stride = f->sr_cur.p.stride[1]; 2966| 355k| if (y_stride * num_lines != f->lf.lr_buf_plane_sz[0] || ------------------ | Branch (2966:9): [True: 24.3k, False: 330k] ------------------ 2967| 330k| uv_stride * num_lines * 2 != f->lf.lr_buf_plane_sz[1]) ------------------ | Branch (2967:9): [True: 346, False: 330k] ------------------ 2968| 24.4k| { 2969| 24.4k| dav1d_free_aligned(f->lf.lr_line_buf); ------------------ | | 136| 24.4k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 2970| | // lr simd may overread the input, so slightly over-allocate the lpf buffer 2971| 24.4k| size_t alloc_sz = 128; 2972| 24.4k| alloc_sz += (size_t)llabs(y_stride) * num_lines; 2973| 24.4k| alloc_sz += (size_t)llabs(uv_stride) * num_lines * 2; 2974| 24.4k| uint8_t *ptr = f->lf.lr_line_buf = dav1d_alloc_aligned(ALLOC_LR, alloc_sz, 64); ------------------ | | 134| 24.4k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 2975| 24.4k| if (!ptr) { ------------------ | Branch (2975:13): [True: 0, False: 24.4k] ------------------ 2976| 0| f->lf.lr_buf_plane_sz[0] = f->lf.lr_buf_plane_sz[1] = 0; 2977| 0| goto error; 2978| 0| } 2979| | 2980| 24.4k| ptr += 64; 2981| 24.4k| if (y_stride < 0) ------------------ | Branch (2981:13): [True: 0, False: 24.4k] ------------------ 2982| 0| f->lf.lr_lpf_line[0] = ptr - y_stride * (num_lines - 1); 2983| 24.4k| else 2984| 24.4k| f->lf.lr_lpf_line[0] = ptr; 2985| 24.4k| ptr += llabs(y_stride) * num_lines; 2986| 24.4k| if (uv_stride < 0) { ------------------ | Branch (2986:13): [True: 0, False: 24.4k] ------------------ 2987| 0| f->lf.lr_lpf_line[1] = ptr - uv_stride * (num_lines * 1 - 1); 2988| 0| f->lf.lr_lpf_line[2] = ptr - uv_stride * (num_lines * 2 - 1); 2989| 24.4k| } else { 2990| 24.4k| f->lf.lr_lpf_line[1] = ptr; 2991| 24.4k| f->lf.lr_lpf_line[2] = ptr + uv_stride * num_lines; 2992| 24.4k| } 2993| | 2994| 24.4k| f->lf.lr_buf_plane_sz[0] = (int) y_stride * num_lines; 2995| 24.4k| f->lf.lr_buf_plane_sz[1] = (int) uv_stride * num_lines * 2; 2996| 24.4k| } 2997| | 2998| | // update allocation for loopfilter masks 2999| 355k| if (num_sb128 != f->lf.mask_sz) { ------------------ | Branch (2999:9): [True: 24.0k, False: 331k] ------------------ 3000| 24.0k| dav1d_free(f->lf.mask); ------------------ | | 135| 24.0k|#define dav1d_free(ptr) free(ptr) ------------------ 3001| 24.0k| dav1d_free(f->lf.level); ------------------ | | 135| 24.0k|#define dav1d_free(ptr) free(ptr) ------------------ 3002| 24.0k| f->lf.mask = dav1d_malloc(ALLOC_LF, sizeof(*f->lf.mask) * num_sb128); ------------------ | | 132| 24.0k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 3003| | // over-allocate by 3 bytes since some of the SIMD implementations 3004| | // index this from the level type and can thus over-read by up to 3 3005| 24.0k| f->lf.level = dav1d_malloc(ALLOC_LF, sizeof(*f->lf.level) * num_sb128 * 32 * 32 + 3); ------------------ | | 132| 24.0k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 3006| 24.0k| if (!f->lf.mask || !f->lf.level) { ------------------ | Branch (3006:13): [True: 18.4E, False: 24.0k] | Branch (3006:28): [True: 0, False: 24.0k] ------------------ 3007| 0| f->lf.mask_sz = 0; 3008| 0| goto error; 3009| 0| } 3010| 24.0k| if (c->n_fc > 1) { ------------------ | Branch (3010:13): [True: 24.0k, False: 18.4E] ------------------ 3011| 24.0k| dav1d_free(f->frame_thread.b); ------------------ | | 135| 24.0k|#define dav1d_free(ptr) free(ptr) ------------------ 3012| 24.0k| f->frame_thread.b = dav1d_malloc(ALLOC_BLOCK, sizeof(*f->frame_thread.b) * ------------------ | | 132| 24.0k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 3013| 24.0k| num_sb128 * 32 * 32); 3014| 24.0k| if (!f->frame_thread.b) { ------------------ | Branch (3014:17): [True: 0, False: 24.0k] ------------------ 3015| 0| f->lf.mask_sz = 0; 3016| 0| goto error; 3017| 0| } 3018| 24.0k| } 3019| 24.0k| f->lf.mask_sz = num_sb128; 3020| 24.0k| } 3021| | 3022| 355k| f->sr_sb128w = (f->sr_cur.p.p.w + 127) >> 7; 3023| 355k| const int lr_mask_sz = f->sr_sb128w * f->sb128h; 3024| 355k| if (lr_mask_sz != f->lf.lr_mask_sz) { ------------------ | Branch (3024:9): [True: 22.8k, False: 332k] ------------------ 3025| 22.8k| dav1d_free(f->lf.lr_mask); ------------------ | | 135| 22.8k|#define dav1d_free(ptr) free(ptr) ------------------ 3026| 22.8k| f->lf.lr_mask = dav1d_malloc(ALLOC_LR, sizeof(*f->lf.lr_mask) * lr_mask_sz); ------------------ | | 132| 22.8k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 3027| 22.8k| if (!f->lf.lr_mask) { ------------------ | Branch (3027:13): [True: 0, False: 22.8k] ------------------ 3028| 0| f->lf.lr_mask_sz = 0; 3029| 0| goto error; 3030| 0| } 3031| 22.8k| f->lf.lr_mask_sz = lr_mask_sz; 3032| 22.8k| } 3033| 355k| f->lf.restore_planes = 3034| 355k| ((f->frame_hdr->restoration.type[0] != DAV1D_RESTORATION_NONE) << 0) + 3035| 355k| ((f->frame_hdr->restoration.type[1] != DAV1D_RESTORATION_NONE) << 1) + 3036| 355k| ((f->frame_hdr->restoration.type[2] != DAV1D_RESTORATION_NONE) << 2); 3037| 355k| if (f->frame_hdr->loopfilter.sharpness != f->lf.last_sharpness) { ------------------ | Branch (3037:9): [True: 34.1k, False: 321k] ------------------ 3038| 34.1k| dav1d_calc_eih(&f->lf.lim_lut, f->frame_hdr->loopfilter.sharpness); 3039| 34.1k| f->lf.last_sharpness = f->frame_hdr->loopfilter.sharpness; 3040| 34.1k| } 3041| 355k| dav1d_calc_lf_values(f->lf.lvl, f->frame_hdr, (int8_t[4]) { 0, 0, 0, 0 }); 3042| 355k| memset(f->lf.mask, 0, sizeof(*f->lf.mask) * num_sb128); 3043| | 3044| 355k| const int ipred_edge_sz = f->sbh * f->sb128w << hbd; 3045| 355k| if (ipred_edge_sz != f->ipred_edge_sz) { ------------------ | Branch (3045:9): [True: 24.1k, False: 331k] ------------------ 3046| 24.1k| dav1d_free_aligned(f->ipred_edge[0]); ------------------ | | 136| 24.1k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 3047| 24.1k| uint8_t *ptr = f->ipred_edge[0] = 3048| 24.1k| dav1d_alloc_aligned(ALLOC_IPRED, ipred_edge_sz * 128 * 3, 64); ------------------ | | 134| 24.1k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 3049| 24.1k| if (!ptr) { ------------------ | Branch (3049:13): [True: 0, False: 24.1k] ------------------ 3050| 0| f->ipred_edge_sz = 0; 3051| 0| goto error; 3052| 0| } 3053| 24.1k| f->ipred_edge[1] = ptr + ipred_edge_sz * 128 * 1; 3054| 24.1k| f->ipred_edge[2] = ptr + ipred_edge_sz * 128 * 2; 3055| 24.1k| f->ipred_edge_sz = ipred_edge_sz; 3056| 24.1k| } 3057| | 3058| 355k| const int re_sz = f->sb128h * f->frame_hdr->tiling.cols; 3059| 355k| if (re_sz != f->lf.re_sz) { ------------------ | Branch (3059:9): [True: 28.2k, False: 327k] ------------------ 3060| 28.2k| dav1d_free(f->lf.tx_lpf_right_edge[0]); ------------------ | | 135| 28.2k|#define dav1d_free(ptr) free(ptr) ------------------ 3061| 28.2k| f->lf.tx_lpf_right_edge[0] = dav1d_malloc(ALLOC_LF, re_sz * 32 * 2); ------------------ | | 132| 28.2k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 3062| 28.2k| if (!f->lf.tx_lpf_right_edge[0]) { ------------------ | Branch (3062:13): [True: 0, False: 28.2k] ------------------ 3063| 0| f->lf.re_sz = 0; 3064| 0| goto error; 3065| 0| } 3066| 28.2k| f->lf.tx_lpf_right_edge[1] = f->lf.tx_lpf_right_edge[0] + re_sz * 32; 3067| 28.2k| f->lf.re_sz = re_sz; 3068| 28.2k| } 3069| | 3070| | // init ref mvs 3071| 355k| if (IS_INTER_OR_SWITCH(f->frame_hdr) || f->frame_hdr->allow_intrabc) { ------------------ | | 36| 710k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 98.2k, False: 257k] | | ------------------ ------------------ | Branch (3071:45): [True: 227k, False: 29.2k] ------------------ 3072| 325k| const int ret = 3073| 325k| dav1d_refmvs_init_frame(&f->rf, f->seq_hdr, f->frame_hdr, 3074| 325k| f->refpoc, f->mvs, f->refrefpoc, f->ref_mvs, 3075| 325k| f->c->n_tc, f->c->n_fc); 3076| 325k| if (ret < 0) goto error; ------------------ | Branch (3076:13): [True: 0, False: 325k] ------------------ 3077| 325k| } 3078| | 3079| | // setup dequant tables 3080| 355k| init_quant_tables(f->seq_hdr, f->frame_hdr, f->frame_hdr->quant.yac, f->dq); 3081| 355k| if (f->frame_hdr->quant.qm) ------------------ | Branch (3081:9): [True: 25.8k, False: 329k] ------------------ 3082| 517k| for (int i = 0; i < N_RECT_TX_SIZES; i++) { ------------------ | Branch (3082:25): [True: 491k, False: 25.8k] ------------------ 3083| 491k| f->qm[i][0] = dav1d_qm_tbl[f->frame_hdr->quant.qm_y][0][i]; 3084| 491k| f->qm[i][1] = dav1d_qm_tbl[f->frame_hdr->quant.qm_u][1][i]; 3085| 491k| f->qm[i][2] = dav1d_qm_tbl[f->frame_hdr->quant.qm_v][1][i]; 3086| 491k| } 3087| 329k| else 3088| 329k| memset(f->qm, 0, sizeof(f->qm)); 3089| | 3090| | // setup jnt_comp weights 3091| 355k| if (f->frame_hdr->switchable_comp_refs) { ------------------ | Branch (3091:9): [True: 52.7k, False: 302k] ------------------ 3092| 421k| for (int i = 0; i < 7; i++) { ------------------ | Branch (3092:25): [True: 368k, False: 52.7k] ------------------ 3093| 368k| const unsigned ref0poc = f->refp[i].p.frame_hdr->frame_offset; 3094| | 3095| 1.46M| for (int j = i + 1; j < 7; j++) { ------------------ | Branch (3095:33): [True: 1.09M, False: 368k] ------------------ 3096| 1.09M| const unsigned ref1poc = f->refp[j].p.frame_hdr->frame_offset; 3097| | 3098| 1.09M| const unsigned d1 = 3099| 1.09M| imin(abs(get_poc_diff(f->seq_hdr->order_hint_n_bits, ref0poc, 3100| 1.09M| f->cur.frame_hdr->frame_offset)), 31); 3101| 1.09M| const unsigned d0 = 3102| 1.09M| imin(abs(get_poc_diff(f->seq_hdr->order_hint_n_bits, ref1poc, 3103| 1.09M| f->cur.frame_hdr->frame_offset)), 31); 3104| 1.09M| const int order = d0 <= d1; 3105| | 3106| 1.09M| static const uint8_t quant_dist_weight[3][2] = { 3107| 1.09M| { 2, 3 }, { 2, 5 }, { 2, 7 } 3108| 1.09M| }; 3109| 1.09M| static const uint8_t quant_dist_lookup_table[4][2] = { 3110| 1.09M| { 9, 7 }, { 11, 5 }, { 12, 4 }, { 13, 3 } 3111| 1.09M| }; 3112| | 3113| 1.09M| int k; 3114| 2.56M| for (k = 0; k < 3; k++) { ------------------ | Branch (3114:29): [True: 2.18M, False: 377k] ------------------ 3115| 2.18M| const int c0 = quant_dist_weight[k][order]; 3116| 2.18M| const int c1 = quant_dist_weight[k][!order]; 3117| 2.18M| const int d0_c0 = d0 * c0; 3118| 2.18M| const int d1_c1 = d1 * c1; 3119| 2.18M| if ((d0 > d1 && d0_c0 < d1_c1) || (d0 <= d1 && d0_c0 > d1_c1)) break; ------------------ | Branch (3119:26): [True: 760k, False: 1.42M] | Branch (3119:37): [True: 192k, False: 567k] | Branch (3119:56): [True: 1.43M, False: 559k] | Branch (3119:68): [True: 529k, False: 904k] ------------------ 3120| 2.18M| } 3121| | 3122| 1.09M| f->jnt_weights[i][j] = quant_dist_lookup_table[k][order]; 3123| 1.09M| } 3124| 368k| } 3125| 52.7k| } 3126| | 3127| | /* Init loopfilter pointers. Increasing NULL pointers is technically UB, 3128| | * so just point the chroma pointers in 4:0:0 to the luma plane here to 3129| | * avoid having additional in-loop branches in various places. We never 3130| | * dereference those pointers so it doesn't really matter what they 3131| | * point at, as long as the pointers are valid. */ 3132| 355k| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400; 3133| 355k| f->lf.p[0] = f->cur.data[0]; 3134| 355k| f->lf.p[1] = f->cur.data[has_chroma ? 1 : 0]; ------------------ | Branch (3134:30): [True: 285k, False: 70.1k] ------------------ 3135| 355k| f->lf.p[2] = f->cur.data[has_chroma ? 2 : 0]; ------------------ | Branch (3135:30): [True: 285k, False: 70.1k] ------------------ 3136| 355k| f->lf.sr_p[0] = f->sr_cur.p.data[0]; 3137| 355k| f->lf.sr_p[1] = f->sr_cur.p.data[has_chroma ? 1 : 0]; ------------------ | Branch (3137:38): [True: 285k, False: 70.1k] ------------------ 3138| 355k| f->lf.sr_p[2] = f->sr_cur.p.data[has_chroma ? 2 : 0]; ------------------ | Branch (3138:38): [True: 285k, False: 70.1k] ------------------ 3139| | 3140| 355k| retval = 0; 3141| 355k|error: 3142| 354k| return retval; 3143| 355k|} dav1d_decode_frame_init_cdf: 3145| 302k|int dav1d_decode_frame_init_cdf(Dav1dFrameContext *const f) { 3146| 302k| const Dav1dContext *const c = f->c; 3147| 302k| int retval = DAV1D_ERR(EINVAL); ------------------ | | 58| 302k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3148| | 3149| 302k| if (f->frame_hdr->refresh_context) ------------------ | Branch (3149:9): [True: 21.0k, False: 281k] ------------------ 3150| 21.0k| dav1d_cdf_thread_copy(f->out_cdf.data.cdf, &f->in_cdf); 3151| | 3152| | // parse individual tiles per tile group 3153| 302k| int tile_row = 0, tile_col = 0; 3154| 302k| f->task_thread.update_set = 0; 3155| 598k| for (int i = 0; i < f->n_tile_data; i++) { ------------------ | Branch (3155:21): [True: 302k, False: 295k] ------------------ 3156| 302k| const uint8_t *data = f->tile[i].data.data; 3157| 302k| size_t size = f->tile[i].data.sz; 3158| | 3159| 618k| for (int j = f->tile[i].start; j <= f->tile[i].end; j++) { ------------------ | Branch (3159:40): [True: 323k, False: 295k] ------------------ 3160| 323k| size_t tile_sz; 3161| 323k| if (j == f->tile[i].end) { ------------------ | Branch (3161:17): [True: 295k, False: 27.7k] ------------------ 3162| 295k| tile_sz = size; 3163| 295k| } else { 3164| 27.7k| if (f->frame_hdr->tiling.n_bytes > size) goto error; ------------------ | Branch (3164:21): [True: 6.06k, False: 21.7k] ------------------ 3165| 21.7k| tile_sz = 0; 3166| 50.3k| for (unsigned k = 0; k < f->frame_hdr->tiling.n_bytes; k++) ------------------ | Branch (3166:38): [True: 28.6k, False: 21.7k] ------------------ 3167| 28.6k| tile_sz |= (unsigned)*data++ << (k * 8); 3168| 21.7k| tile_sz++; 3169| 21.7k| size -= f->frame_hdr->tiling.n_bytes; 3170| 21.7k| if (tile_sz > size) goto error; ------------------ | Branch (3170:21): [True: 1.43k, False: 20.2k] ------------------ 3171| 21.7k| } 3172| | 3173| 315k| setup_tile(&f->ts[j], f, data, tile_sz, tile_row, tile_col++, 3174| 315k| c->n_fc > 1 ? f->frame_thread.tile_start_off[j] : 0); ------------------ | Branch (3174:24): [True: 315k, False: 20] ------------------ 3175| | 3176| 315k| if (tile_col == f->frame_hdr->tiling.cols) { ------------------ | Branch (3176:17): [True: 306k, False: 9.06k] ------------------ 3177| 306k| tile_col = 0; 3178| 306k| tile_row++; 3179| 306k| } 3180| 315k| if (j == f->frame_hdr->tiling.update && f->frame_hdr->refresh_context) ------------------ | Branch (3180:17): [True: 294k, False: 20.8k] | Branch (3180:53): [True: 20.7k, False: 274k] ------------------ 3181| 20.7k| f->task_thread.update_set = 1; 3182| 315k| data += tile_sz; 3183| 315k| size -= tile_sz; 3184| 315k| } 3185| 302k| } 3186| | 3187| 295k| if (c->n_tc > 1) { ------------------ | Branch (3187:9): [True: 294k, False: 475] ------------------ 3188| 294k| const int uses_2pass = c->n_fc > 1; 3189| 3.34M| for (int n = 0; n < f->sb128w * f->frame_hdr->tiling.rows * (1 + uses_2pass); n++) ------------------ | Branch (3189:25): [True: 3.05M, False: 294k] ------------------ 3190| 3.05M| reset_context(&f->a[n], IS_KEY_OR_INTRA(f->frame_hdr), ------------------ | | 43| 3.05M| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 3.05M| ((frame_header)->frame_type & 1) | | ------------------ ------------------ 3191| 18.4E| uses_2pass ? 1 + (n >= f->sb128w * f->frame_hdr->tiling.rows) : 0); ------------------ | Branch (3191:27): [True: 3.05M, False: 18.4E] ------------------ 3192| 294k| } 3193| | 3194| 295k| retval = 0; 3195| 301k|error: 3196| 301k| return retval; 3197| 295k|} dav1d_decode_frame_exit: 3245| 391k|void dav1d_decode_frame_exit(Dav1dFrameContext *const f, int retval) { 3246| 391k| const Dav1dContext *const c = f->c; 3247| | 3248| 391k| if (f->sr_cur.p.data[0]) ------------------ | Branch (3248:9): [True: 356k, False: 35.5k] ------------------ 3249| 391k| atomic_init(&f->task_thread.error, 0); 3250| | 3251| 391k| if (c->n_fc > 1 && retval && f->frame_thread.cf) { ------------------ | Branch (3251:9): [True: 391k, False: 0] | Branch (3251:24): [True: 220k, False: 171k] | Branch (3251:34): [True: 199k, False: 20.9k] ------------------ 3252| 199k| memset(f->frame_thread.cf, 0, 3253| 199k| (size_t)f->frame_thread.cf_sz * 128 * 128 / 2); 3254| 199k| } 3255| 3.13M| for (int i = 0; i < 7; i++) { ------------------ | Branch (3255:21): [True: 2.74M, False: 391k] ------------------ 3256| 2.74M| if (f->refp[i].p.frame_hdr) { ------------------ | Branch (3256:13): [True: 688k, False: 2.05M] ------------------ 3257| 688k| if (!retval && c->n_fc > 1 && c->strict_std_compliance && ------------------ | Branch (3257:17): [True: 76.4k, False: 611k] | Branch (3257:28): [True: 76.4k, False: 0] | Branch (3257:43): [True: 0, False: 76.4k] ------------------ 3258| 688k| atomic_load(&f->refp[i].progress[1]) == FRAME_ERROR) ------------------ | | 35| 0|#define FRAME_ERROR (UINT_MAX - 1) ------------------ | Branch (3258:17): [True: 0, False: 0] ------------------ 3259| 0| { 3260| 0| retval = DAV1D_ERR(EINVAL); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3261| 0| atomic_store(&f->task_thread.error, 1); 3262| 0| atomic_store(&f->sr_cur.progress[1], FRAME_ERROR); 3263| 0| } 3264| 688k| dav1d_thread_picture_unref(&f->refp[i]); 3265| 688k| } 3266| 2.74M| dav1d_ref_dec(&f->ref_mvs_ref[i]); 3267| 2.74M| } 3268| | 3269| 391k| dav1d_picture_unref_internal(&f->cur); 3270| 391k| dav1d_thread_picture_unref(&f->sr_cur); 3271| 391k| dav1d_cdf_thread_unref(&f->in_cdf); 3272| 391k| if (f->frame_hdr && f->frame_hdr->refresh_context) { ------------------ | Branch (3272:9): [True: 371k, False: 20.5k] | Branch (3272:25): [True: 59.1k, False: 312k] ------------------ 3273| 59.1k| if (f->out_cdf.progress) ------------------ | Branch (3273:13): [True: 55.3k, False: 3.80k] ------------------ 3274| 59.1k| atomic_store(f->out_cdf.progress, retval == 0 ? 1 : TILE_ERROR); ------------------ | Branch (3274:13): [True: 9.81k, False: 45.5k] ------------------ 3275| 59.1k| dav1d_cdf_thread_unref(&f->out_cdf); 3276| 59.1k| } 3277| 391k| dav1d_ref_dec(&f->cur_segmap_ref); 3278| 391k| dav1d_ref_dec(&f->prev_segmap_ref); 3279| 391k| dav1d_ref_dec(&f->mvs_ref); 3280| 391k| dav1d_ref_dec(&f->seq_hdr_ref); 3281| 391k| dav1d_ref_dec(&f->frame_hdr_ref); 3282| | 3283| 748k| for (int i = 0; i < f->n_tile_data; i++) ------------------ | Branch (3283:21): [True: 356k, False: 391k] ------------------ 3284| 356k| dav1d_data_unref_internal(&f->tile[i].data); 3285| 391k| f->task_thread.retval = retval; 3286| 391k|} dav1d_submit_frame: 3330| 359k|int dav1d_submit_frame(Dav1dContext *const c) { 3331| 359k| Dav1dFrameContext *f; 3332| 359k| int res = -1; 3333| | 3334| | // wait for c->out_delayed[next] and move into c->out if visible 3335| 359k| Dav1dThreadPicture *out_delayed; 3336| 359k| if (c->n_fc > 1) { ------------------ | Branch (3336:9): [True: 359k, False: 0] ------------------ 3337| 359k| pthread_mutex_lock(&c->task_thread.lock); 3338| 359k| const unsigned next = c->frame_thread.next++; 3339| 359k| if (c->frame_thread.next == c->n_fc) ------------------ | Branch (3339:13): [True: 86.9k, False: 272k] ------------------ 3340| 86.9k| c->frame_thread.next = 0; 3341| | 3342| 359k| f = &c->fc[next]; 3343| 477k| while (f->n_tile_data > 0) ------------------ | Branch (3343:16): [True: 118k, False: 359k] ------------------ 3344| 118k| pthread_cond_wait(&f->task_thread.cond, 3345| 118k| &c->task_thread.lock); 3346| 359k| out_delayed = &c->frame_thread.out_delayed[next]; 3347| 359k| if (out_delayed->p.data[0] || atomic_load(&f->task_thread.error)) { ------------------ | Branch (3347:13): [True: 339k, False: 19.6k] | Branch (3347:39): [True: 2.57k, False: 17.0k] ------------------ 3348| 342k| unsigned first = atomic_load(&c->task_thread.first); 3349| 342k| if (first + 1U < c->n_fc) ------------------ | Branch (3349:17): [True: 257k, False: 85.0k] ------------------ 3350| 342k| atomic_fetch_add(&c->task_thread.first, 1U); 3351| 85.0k| else 3352| 342k| atomic_store(&c->task_thread.first, 0); 3353| 342k| atomic_compare_exchange_strong(&c->task_thread.reset_task_cur, 3354| 342k| &first, UINT_MAX); 3355| 342k| if (c->task_thread.cur && c->task_thread.cur < c->n_fc) ------------------ | Branch (3355:17): [True: 339k, False: 3.07k] | Branch (3355:39): [True: 221k, False: 117k] ------------------ 3356| 221k| c->task_thread.cur--; 3357| 342k| } 3358| 359k| const int error = f->task_thread.retval; 3359| 359k| if (error) { ------------------ | Branch (3359:13): [True: 174k, False: 184k] ------------------ 3360| 174k| f->task_thread.retval = 0; 3361| 174k| c->cached_error = error; 3362| 174k| dav1d_data_props_copy(&c->cached_error_props, &out_delayed->p.m); 3363| 174k| dav1d_thread_picture_unref(out_delayed); 3364| 184k| } else if (out_delayed->p.data[0]) { ------------------ | Branch (3364:20): [True: 165k, False: 19.6k] ------------------ 3365| 165k| const unsigned progress = atomic_load_explicit(&out_delayed->progress[1], 3366| 165k| memory_order_relaxed); 3367| 165k| if ((out_delayed->visible || c->output_invisible_frames) && ------------------ | Branch (3367:18): [True: 159k, False: 5.75k] | Branch (3367:42): [True: 0, False: 5.75k] ------------------ 3368| 159k| progress != FRAME_ERROR) ------------------ | | 35| 159k|#define FRAME_ERROR (UINT_MAX - 1) ------------------ | Branch (3368:17): [True: 155k, False: 4.31k] ------------------ 3369| 155k| { 3370| 155k| dav1d_thread_picture_ref(&c->out, out_delayed); 3371| 155k| c->event_flags |= dav1d_picture_get_event_flags(out_delayed); 3372| 155k| } 3373| 165k| dav1d_thread_picture_unref(out_delayed); 3374| 165k| } 3375| 359k| } else { 3376| 0| f = c->fc; 3377| 0| } 3378| | 3379| 359k| f->seq_hdr = c->seq_hdr; 3380| 359k| f->seq_hdr_ref = c->seq_hdr_ref; 3381| 359k| dav1d_ref_inc(f->seq_hdr_ref); 3382| 359k| f->frame_hdr = c->frame_hdr; 3383| 359k| f->frame_hdr_ref = c->frame_hdr_ref; 3384| 359k| c->frame_hdr = NULL; 3385| 359k| c->frame_hdr_ref = NULL; 3386| 359k| f->dsp = &c->dsp[f->seq_hdr->hbd]; 3387| | 3388| 359k| const int bpc = 8 + 2 * f->seq_hdr->hbd; 3389| | 3390| 359k| if (!f->dsp->ipred.intra_pred[DC_PRED]) { ------------------ | Branch (3390:9): [True: 8.57k, False: 350k] ------------------ 3391| 8.57k| Dav1dDSPContext *const dsp = &c->dsp[f->seq_hdr->hbd]; 3392| | 3393| 8.57k| switch (bpc) { 3394| 0|#define assign_bitdepth_case(bd) \ 3395| 0| dav1d_cdef_dsp_init_##bd##bpc(&dsp->cdef); \ 3396| 0| dav1d_intra_pred_dsp_init_##bd##bpc(&dsp->ipred); \ 3397| 0| dav1d_itx_dsp_init_##bd##bpc(&dsp->itx, bpc); \ 3398| 0| dav1d_loop_filter_dsp_init_##bd##bpc(&dsp->lf); \ 3399| 0| dav1d_loop_restoration_dsp_init_##bd##bpc(&dsp->lr, bpc); \ 3400| 0| dav1d_mc_dsp_init_##bd##bpc(&dsp->mc); \ 3401| 0| dav1d_film_grain_dsp_init_##bd##bpc(&dsp->fg); \ 3402| 0| break 3403| 0|#if CONFIG_8BPC 3404| 3.46k| case 8: ------------------ | Branch (3404:9): [True: 3.46k, False: 5.11k] ------------------ 3405| 3.46k| assign_bitdepth_case(8); ------------------ | | 3395| 3.46k| dav1d_cdef_dsp_init_##bd##bpc(&dsp->cdef); \ | | 3396| 3.46k| dav1d_intra_pred_dsp_init_##bd##bpc(&dsp->ipred); \ | | 3397| 3.46k| dav1d_itx_dsp_init_##bd##bpc(&dsp->itx, bpc); \ | | 3398| 3.46k| dav1d_loop_filter_dsp_init_##bd##bpc(&dsp->lf); \ | | 3399| 3.46k| dav1d_loop_restoration_dsp_init_##bd##bpc(&dsp->lr, bpc); \ | | 3400| 3.46k| dav1d_mc_dsp_init_##bd##bpc(&dsp->mc); \ | | 3401| 3.46k| dav1d_film_grain_dsp_init_##bd##bpc(&dsp->fg); \ | | 3402| 3.46k| break ------------------ 3406| 0|#endif 3407| 0|#if CONFIG_16BPC 3408| 2.11k| case 10: ------------------ | Branch (3408:9): [True: 2.11k, False: 6.46k] ------------------ 3409| 5.11k| case 12: ------------------ | Branch (3409:9): [True: 2.99k, False: 5.58k] ------------------ 3410| 5.11k| assign_bitdepth_case(16); ------------------ | | 3395| 5.11k| dav1d_cdef_dsp_init_##bd##bpc(&dsp->cdef); \ | | 3396| 5.11k| dav1d_intra_pred_dsp_init_##bd##bpc(&dsp->ipred); \ | | 3397| 5.11k| dav1d_itx_dsp_init_##bd##bpc(&dsp->itx, bpc); \ | | 3398| 5.11k| dav1d_loop_filter_dsp_init_##bd##bpc(&dsp->lf); \ | | 3399| 5.11k| dav1d_loop_restoration_dsp_init_##bd##bpc(&dsp->lr, bpc); \ | | 3400| 5.11k| dav1d_mc_dsp_init_##bd##bpc(&dsp->mc); \ | | 3401| 5.11k| dav1d_film_grain_dsp_init_##bd##bpc(&dsp->fg); \ | | 3402| 5.11k| break ------------------ 3411| 0|#endif 3412| 0|#undef assign_bitdepth_case 3413| 0| default: ------------------ | Branch (3413:9): [True: 0, False: 8.57k] ------------------ 3414| 0| dav1d_log(c, "Compiled without support for %d-bit decoding\n", ------------------ | | 44| 0|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 0] | | ------------------ ------------------ 3415| 0| 8 + 2 * f->seq_hdr->hbd); 3416| 0| res = DAV1D_ERR(ENOPROTOOPT); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3417| 0| goto error; 3418| 8.57k| } 3419| 8.57k| } 3420| | 3421| 359k|#define assign_bitdepth_case(bd) \ 3422| 359k| f->bd_fn.recon_b_inter = dav1d_recon_b_inter_##bd##bpc; \ 3423| 359k| f->bd_fn.recon_b_intra = dav1d_recon_b_intra_##bd##bpc; \ 3424| 359k| f->bd_fn.filter_sbrow = dav1d_filter_sbrow_##bd##bpc; \ 3425| 359k| f->bd_fn.filter_sbrow_deblock_cols = dav1d_filter_sbrow_deblock_cols_##bd##bpc; \ 3426| 359k| f->bd_fn.filter_sbrow_deblock_rows = dav1d_filter_sbrow_deblock_rows_##bd##bpc; \ 3427| 359k| f->bd_fn.filter_sbrow_cdef = dav1d_filter_sbrow_cdef_##bd##bpc; \ 3428| 359k| f->bd_fn.filter_sbrow_resize = dav1d_filter_sbrow_resize_##bd##bpc; \ 3429| 359k| f->bd_fn.filter_sbrow_lr = dav1d_filter_sbrow_lr_##bd##bpc; \ 3430| 359k| f->bd_fn.backup_ipred_edge = dav1d_backup_ipred_edge_##bd##bpc; \ 3431| 359k| f->bd_fn.read_coef_blocks = dav1d_read_coef_blocks_##bd##bpc; \ 3432| 359k| f->bd_fn.copy_pal_block_y = dav1d_copy_pal_block_y_##bd##bpc; \ 3433| 359k| f->bd_fn.copy_pal_block_uv = dav1d_copy_pal_block_uv_##bd##bpc; \ 3434| 359k| f->bd_fn.read_pal_plane = dav1d_read_pal_plane_##bd##bpc; \ 3435| 359k| f->bd_fn.read_pal_uv = dav1d_read_pal_uv_##bd##bpc 3436| 359k| if (!f->seq_hdr->hbd) { ------------------ | Branch (3436:9): [True: 148k, False: 210k] ------------------ 3437| 148k|#if CONFIG_8BPC 3438| 148k| assign_bitdepth_case(8); ------------------ | | 3422| 148k| f->bd_fn.recon_b_inter = dav1d_recon_b_inter_##bd##bpc; \ | | 3423| 148k| f->bd_fn.recon_b_intra = dav1d_recon_b_intra_##bd##bpc; \ | | 3424| 148k| f->bd_fn.filter_sbrow = dav1d_filter_sbrow_##bd##bpc; \ | | 3425| 148k| f->bd_fn.filter_sbrow_deblock_cols = dav1d_filter_sbrow_deblock_cols_##bd##bpc; \ | | 3426| 148k| f->bd_fn.filter_sbrow_deblock_rows = dav1d_filter_sbrow_deblock_rows_##bd##bpc; \ | | 3427| 148k| f->bd_fn.filter_sbrow_cdef = dav1d_filter_sbrow_cdef_##bd##bpc; \ | | 3428| 148k| f->bd_fn.filter_sbrow_resize = dav1d_filter_sbrow_resize_##bd##bpc; \ | | 3429| 148k| f->bd_fn.filter_sbrow_lr = dav1d_filter_sbrow_lr_##bd##bpc; \ | | 3430| 148k| f->bd_fn.backup_ipred_edge = dav1d_backup_ipred_edge_##bd##bpc; \ | | 3431| 148k| f->bd_fn.read_coef_blocks = dav1d_read_coef_blocks_##bd##bpc; \ | | 3432| 148k| f->bd_fn.copy_pal_block_y = dav1d_copy_pal_block_y_##bd##bpc; \ | | 3433| 148k| f->bd_fn.copy_pal_block_uv = dav1d_copy_pal_block_uv_##bd##bpc; \ | | 3434| 148k| f->bd_fn.read_pal_plane = dav1d_read_pal_plane_##bd##bpc; \ | | 3435| 148k| f->bd_fn.read_pal_uv = dav1d_read_pal_uv_##bd##bpc ------------------ 3439| 148k|#endif 3440| 210k| } else { 3441| 210k|#if CONFIG_16BPC 3442| 210k| assign_bitdepth_case(16); ------------------ | | 3422| 210k| f->bd_fn.recon_b_inter = dav1d_recon_b_inter_##bd##bpc; \ | | 3423| 210k| f->bd_fn.recon_b_intra = dav1d_recon_b_intra_##bd##bpc; \ | | 3424| 210k| f->bd_fn.filter_sbrow = dav1d_filter_sbrow_##bd##bpc; \ | | 3425| 210k| f->bd_fn.filter_sbrow_deblock_cols = dav1d_filter_sbrow_deblock_cols_##bd##bpc; \ | | 3426| 210k| f->bd_fn.filter_sbrow_deblock_rows = dav1d_filter_sbrow_deblock_rows_##bd##bpc; \ | | 3427| 210k| f->bd_fn.filter_sbrow_cdef = dav1d_filter_sbrow_cdef_##bd##bpc; \ | | 3428| 210k| f->bd_fn.filter_sbrow_resize = dav1d_filter_sbrow_resize_##bd##bpc; \ | | 3429| 210k| f->bd_fn.filter_sbrow_lr = dav1d_filter_sbrow_lr_##bd##bpc; \ | | 3430| 210k| f->bd_fn.backup_ipred_edge = dav1d_backup_ipred_edge_##bd##bpc; \ | | 3431| 210k| f->bd_fn.read_coef_blocks = dav1d_read_coef_blocks_##bd##bpc; \ | | 3432| 210k| f->bd_fn.copy_pal_block_y = dav1d_copy_pal_block_y_##bd##bpc; \ | | 3433| 210k| f->bd_fn.copy_pal_block_uv = dav1d_copy_pal_block_uv_##bd##bpc; \ | | 3434| 210k| f->bd_fn.read_pal_plane = dav1d_read_pal_plane_##bd##bpc; \ | | 3435| 210k| f->bd_fn.read_pal_uv = dav1d_read_pal_uv_##bd##bpc ------------------ 3443| 210k|#endif 3444| 210k| } 3445| 359k|#undef assign_bitdepth_case 3446| | 3447| 359k| int ref_coded_width[7]; 3448| 359k| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { ------------------ | | 36| 359k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 101k, False: 258k] | | ------------------ ------------------ 3449| 101k| if (f->frame_hdr->primary_ref_frame != DAV1D_PRIMARY_REF_NONE) { ------------------ | | 45| 101k|#define DAV1D_PRIMARY_REF_NONE 7 ------------------ | Branch (3449:13): [True: 80.1k, False: 21.1k] ------------------ 3450| 80.1k| const int pri_ref = f->frame_hdr->refidx[f->frame_hdr->primary_ref_frame]; 3451| 80.1k| if (!c->refs[pri_ref].p.p.data[0]) { ------------------ | Branch (3451:17): [True: 343, False: 79.8k] ------------------ 3452| 343| res = DAV1D_ERR(EINVAL); ------------------ | | 58| 343|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3453| 343| goto error; 3454| 343| } 3455| 80.1k| } 3456| 790k| for (int i = 0; i < 7; i++) { ------------------ | Branch (3456:25): [True: 692k, False: 98.3k] ------------------ 3457| 692k| const int refidx = f->frame_hdr->refidx[i]; 3458| 692k| if (!c->refs[refidx].p.p.data[0] || ------------------ | Branch (3458:17): [True: 233, False: 691k] ------------------ 3459| 691k| f->frame_hdr->width[0] * 2 < c->refs[refidx].p.p.p.w || ------------------ | Branch (3459:17): [True: 548, False: 691k] ------------------ 3460| 691k| f->frame_hdr->height * 2 < c->refs[refidx].p.p.p.h || ------------------ | Branch (3460:17): [True: 1.25k, False: 690k] ------------------ 3461| 690k| f->frame_hdr->width[0] > c->refs[refidx].p.p.p.w * 16 || ------------------ | Branch (3461:17): [True: 347, False: 689k] ------------------ 3462| 689k| f->frame_hdr->height > c->refs[refidx].p.p.p.h * 16 || ------------------ | Branch (3462:17): [True: 284, False: 689k] ------------------ 3463| 689k| f->seq_hdr->layout != c->refs[refidx].p.p.p.layout || ------------------ | Branch (3463:17): [True: 0, False: 689k] ------------------ 3464| 689k| bpc != c->refs[refidx].p.p.p.bpc) ------------------ | Branch (3464:17): [True: 0, False: 689k] ------------------ 3465| 2.66k| { 3466| 4.00k| for (int j = 0; j < i; j++) ------------------ | Branch (3466:33): [True: 1.33k, False: 2.66k] ------------------ 3467| 1.33k| dav1d_thread_picture_unref(&f->refp[j]); 3468| 2.66k| res = DAV1D_ERR(EINVAL); ------------------ | | 58| 2.66k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3469| 2.66k| goto error; 3470| 2.66k| } 3471| 689k| dav1d_thread_picture_ref(&f->refp[i], &c->refs[refidx].p); 3472| 689k| ref_coded_width[i] = c->refs[refidx].p.p.frame_hdr->width[0]; 3473| 689k| if (f->frame_hdr->width[0] != c->refs[refidx].p.p.p.w || ------------------ | Branch (3473:17): [True: 209k, False: 480k] ------------------ 3474| 480k| f->frame_hdr->height != c->refs[refidx].p.p.p.h) ------------------ | Branch (3474:17): [True: 2.79k, False: 477k] ------------------ 3475| 211k| { 3476| 211k|#define scale_fac(ref_sz, this_sz) \ 3477| 211k| ((((ref_sz) << 14) + ((this_sz) >> 1)) / (this_sz)) 3478| 211k| f->svc[i][0].scale = scale_fac(c->refs[refidx].p.p.p.w, ------------------ | | 3477| 211k| ((((ref_sz) << 14) + ((this_sz) >> 1)) / (this_sz)) ------------------ 3479| 211k| f->frame_hdr->width[0]); 3480| 211k| f->svc[i][1].scale = scale_fac(c->refs[refidx].p.p.p.h, ------------------ | | 3477| 211k| ((((ref_sz) << 14) + ((this_sz) >> 1)) / (this_sz)) ------------------ 3481| 211k| f->frame_hdr->height); 3482| 211k| f->svc[i][0].step = (f->svc[i][0].scale + 8) >> 4; 3483| 211k| f->svc[i][1].step = (f->svc[i][1].scale + 8) >> 4; 3484| 477k| } else { 3485| 477k| f->svc[i][0].scale = f->svc[i][1].scale = 0; 3486| 477k| } 3487| 689k| f->gmv_warp_allowed[i] = f->frame_hdr->gmv[i].type > DAV1D_WM_TYPE_TRANSLATION && ------------------ | Branch (3487:38): [True: 30.3k, False: 659k] ------------------ 3488| 30.3k| !f->frame_hdr->force_integer_mv && ------------------ | Branch (3488:38): [True: 27.4k, False: 2.95k] ------------------ 3489| 27.4k| !dav1d_get_shear_params(&f->frame_hdr->gmv[i]) && ------------------ | Branch (3489:38): [True: 25.8k, False: 1.59k] ------------------ 3490| 25.8k| !f->svc[i][0].scale; ------------------ | Branch (3490:38): [True: 12.1k, False: 13.6k] ------------------ 3491| 689k| } 3492| 100k| } 3493| | 3494| | // setup entropy 3495| 356k| if (f->frame_hdr->primary_ref_frame == DAV1D_PRIMARY_REF_NONE) { ------------------ | | 45| 356k|#define DAV1D_PRIMARY_REF_NONE 7 ------------------ | Branch (3495:9): [True: 278k, False: 77.3k] ------------------ 3496| 278k| dav1d_cdf_thread_init_static(&f->in_cdf, f->frame_hdr->quant.yac); 3497| 278k| } else { 3498| 77.3k| const int pri_ref = f->frame_hdr->refidx[f->frame_hdr->primary_ref_frame]; 3499| 77.3k| dav1d_cdf_thread_ref(&f->in_cdf, &c->cdf[pri_ref]); 3500| 77.3k| } 3501| 356k| if (f->frame_hdr->refresh_context) { ------------------ | Branch (3501:9): [True: 55.3k, False: 301k] ------------------ 3502| 55.3k| res = dav1d_cdf_thread_alloc(c, &f->out_cdf, c->n_fc > 1); 3503| 55.3k| if (res < 0) goto error; ------------------ | Branch (3503:13): [True: 0, False: 55.3k] ------------------ 3504| 55.3k| } 3505| | 3506| | // FIXME qsort so tiles are in order (for frame threading) 3507| 356k| if (f->n_tile_data_alloc < c->n_tile_data) { ------------------ | Branch (3507:9): [True: 16.9k, False: 339k] ------------------ 3508| 16.9k| dav1d_free(f->tile); ------------------ | | 135| 16.9k|#define dav1d_free(ptr) free(ptr) ------------------ 3509| 16.9k| assert(c->n_tile_data < INT_MAX / (int)sizeof(*f->tile)); ------------------ | Branch (3509:9): [True: 16.9k, False: 0] ------------------ 3510| 16.9k| f->tile = dav1d_malloc(ALLOC_TILE, c->n_tile_data * sizeof(*f->tile)); ------------------ | | 132| 16.9k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 3511| 16.9k| if (!f->tile) { ------------------ | Branch (3511:13): [True: 0, False: 16.9k] ------------------ 3512| 0| f->n_tile_data_alloc = f->n_tile_data = 0; 3513| 0| res = DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3514| 0| goto error; 3515| 0| } 3516| 16.9k| f->n_tile_data_alloc = c->n_tile_data; 3517| 16.9k| } 3518| 356k| memcpy(f->tile, c->tile, c->n_tile_data * sizeof(*f->tile)); 3519| 356k| memset(c->tile, 0, c->n_tile_data * sizeof(*c->tile)); 3520| 356k| f->n_tile_data = c->n_tile_data; 3521| 356k| c->n_tile_data = 0; 3522| | 3523| | // allocate frame 3524| 356k| res = dav1d_thread_picture_alloc(c, f, bpc); 3525| 356k| if (res < 0) goto error; ------------------ | Branch (3525:9): [True: 0, False: 356k] ------------------ 3526| | 3527| 356k| if (f->frame_hdr->width[0] != f->frame_hdr->width[1]) { ------------------ | Branch (3527:9): [True: 39.1k, False: 317k] ------------------ 3528| 39.1k| res = dav1d_picture_alloc_copy(c, &f->cur, f->frame_hdr->width[0], &f->sr_cur.p); 3529| 39.1k| if (res < 0) goto error; ------------------ | Branch (3529:13): [True: 0, False: 39.1k] ------------------ 3530| 317k| } else { 3531| 317k| dav1d_picture_ref(&f->cur, &f->sr_cur.p); 3532| 317k| } 3533| | 3534| 356k| if (f->frame_hdr->width[0] != f->frame_hdr->width[1]) { ------------------ | Branch (3534:9): [True: 39.1k, False: 317k] ------------------ 3535| 39.1k| f->resize_step[0] = scale_fac(f->cur.p.w, f->sr_cur.p.p.w); ------------------ | | 3477| 39.1k| ((((ref_sz) << 14) + ((this_sz) >> 1)) / (this_sz)) ------------------ 3536| 39.1k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 3537| 39.1k| const int in_cw = (f->cur.p.w + ss_hor) >> ss_hor; 3538| 39.1k| const int out_cw = (f->sr_cur.p.p.w + ss_hor) >> ss_hor; 3539| 39.1k| f->resize_step[1] = scale_fac(in_cw, out_cw); ------------------ | | 3477| 39.1k| ((((ref_sz) << 14) + ((this_sz) >> 1)) / (this_sz)) ------------------ 3540| 39.1k|#undef scale_fac 3541| 39.1k| f->resize_start[0] = get_upscale_x0(f->cur.p.w, f->sr_cur.p.p.w, f->resize_step[0]); 3542| 39.1k| f->resize_start[1] = get_upscale_x0(in_cw, out_cw, f->resize_step[1]); 3543| 39.1k| } 3544| | 3545| | // move f->cur into output queue 3546| 356k| if (c->n_fc == 1) { ------------------ | Branch (3546:9): [True: 0, False: 356k] ------------------ 3547| 0| if (f->frame_hdr->show_frame || c->output_invisible_frames) { ------------------ | Branch (3547:13): [True: 0, False: 0] | Branch (3547:41): [True: 0, False: 0] ------------------ 3548| 0| dav1d_thread_picture_ref(&c->out, &f->sr_cur); 3549| 0| c->event_flags |= dav1d_picture_get_event_flags(&f->sr_cur); 3550| 0| } 3551| 356k| } else { 3552| 356k| dav1d_thread_picture_ref(out_delayed, &f->sr_cur); 3553| 356k| } 3554| | 3555| 356k| f->w4 = (f->frame_hdr->width[0] + 3) >> 2; 3556| 356k| f->h4 = (f->frame_hdr->height + 3) >> 2; 3557| 356k| f->bw = ((f->frame_hdr->width[0] + 7) >> 3) << 1; 3558| 356k| f->bh = ((f->frame_hdr->height + 7) >> 3) << 1; 3559| 356k| f->sb128w = (f->bw + 31) >> 5; 3560| 356k| f->sb128h = (f->bh + 31) >> 5; 3561| 356k| f->sb_shift = 4 + f->seq_hdr->sb128; 3562| 356k| f->sb_step = 16 << f->seq_hdr->sb128; 3563| 356k| f->sbh = (f->bh + f->sb_step - 1) >> f->sb_shift; 3564| 356k| f->b4_stride = (f->bw + 31) & ~31; 3565| 356k| f->bitdepth_max = (1 << f->cur.p.bpc) - 1; 3566| 356k| atomic_init(&f->task_thread.error, 0); 3567| 356k| const int uses_2pass = c->n_fc > 1; 3568| 356k| const int cols = f->frame_hdr->tiling.cols; 3569| 356k| const int rows = f->frame_hdr->tiling.rows; 3570| 356k| atomic_store(&f->task_thread.task_counter, 3571| 356k| (cols * rows + f->sbh) << uses_2pass); 3572| | 3573| | // ref_mvs 3574| 356k| if (IS_INTER_OR_SWITCH(f->frame_hdr) || f->frame_hdr->allow_intrabc) { ------------------ | | 36| 712k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 98.3k, False: 258k] | | ------------------ ------------------ | Branch (3574:45): [True: 228k, False: 29.5k] ------------------ 3575| 326k| f->mvs_ref = dav1d_ref_create_using_pool(c->refmvs_pool, 3576| 326k| sizeof(*f->mvs) * f->sb128h * 16 * (f->b4_stride >> 1)); 3577| 326k| if (!f->mvs_ref) { ------------------ | Branch (3577:13): [True: 0, False: 326k] ------------------ 3578| 0| res = DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3579| 0| goto error; 3580| 0| } 3581| 326k| f->mvs = f->mvs_ref->data; 3582| 326k| if (!f->frame_hdr->allow_intrabc) { ------------------ | Branch (3582:13): [True: 98.3k, False: 228k] ------------------ 3583| 786k| for (int i = 0; i < 7; i++) ------------------ | Branch (3583:29): [True: 688k, False: 98.3k] ------------------ 3584| 688k| f->refpoc[i] = f->refp[i].p.frame_hdr->frame_offset; 3585| 228k| } else { 3586| 228k| memset(f->refpoc, 0, sizeof(f->refpoc)); 3587| 228k| } 3588| 326k| if (f->frame_hdr->use_ref_frame_mvs) { ------------------ | Branch (3588:13): [True: 75.8k, False: 250k] ------------------ 3589| 606k| for (int i = 0; i < 7; i++) { ------------------ | Branch (3589:29): [True: 530k, False: 75.8k] ------------------ 3590| 530k| const int refidx = f->frame_hdr->refidx[i]; 3591| 530k| const int ref_w = ((ref_coded_width[i] + 7) >> 3) << 1; 3592| 530k| const int ref_h = ((f->refp[i].p.p.h + 7) >> 3) << 1; 3593| 530k| if (c->refs[refidx].refmvs != NULL && ------------------ | Branch (3593:21): [True: 434k, False: 95.9k] ------------------ 3594| 434k| ref_w == f->bw && ref_h == f->bh) ------------------ | Branch (3594:21): [True: 428k, False: 6.29k] | Branch (3594:39): [True: 428k, False: 516] ------------------ 3595| 428k| { 3596| 428k| f->ref_mvs_ref[i] = c->refs[refidx].refmvs; 3597| 428k| dav1d_ref_inc(f->ref_mvs_ref[i]); 3598| 428k| f->ref_mvs[i] = c->refs[refidx].refmvs->data; 3599| 428k| } else { 3600| 102k| f->ref_mvs[i] = NULL; 3601| 102k| f->ref_mvs_ref[i] = NULL; 3602| 102k| } 3603| 530k| memcpy(f->refrefpoc[i], c->refs[refidx].refpoc, 3604| 530k| sizeof(*f->refrefpoc)); 3605| 530k| } 3606| 250k| } else { 3607| 250k| memset(f->ref_mvs_ref, 0, sizeof(f->ref_mvs_ref)); 3608| 250k| } 3609| 326k| } else { 3610| 29.5k| f->mvs_ref = NULL; 3611| 29.5k| memset(f->ref_mvs_ref, 0, sizeof(f->ref_mvs_ref)); 3612| 29.5k| } 3613| | 3614| | // segmap 3615| 356k| if (f->frame_hdr->segmentation.enabled) { ------------------ | Branch (3615:9): [True: 19.6k, False: 336k] ------------------ 3616| | // By default, the previous segmentation map is not initialised. 3617| 19.6k| f->prev_segmap_ref = NULL; 3618| 19.6k| f->prev_segmap = NULL; 3619| | 3620| | // We might need a previous frame's segmentation map. This 3621| | // happens if there is either no update or a temporal update. 3622| 19.6k| if (f->frame_hdr->segmentation.temporal || !f->frame_hdr->segmentation.update_map) { ------------------ | Branch (3622:13): [True: 8.32k, False: 11.2k] | Branch (3622:52): [True: 6.79k, False: 4.50k] ------------------ 3623| 15.1k| const int pri_ref = f->frame_hdr->primary_ref_frame; 3624| 15.1k| assert(pri_ref != DAV1D_PRIMARY_REF_NONE); ------------------ | Branch (3624:13): [True: 15.1k, False: 0] ------------------ 3625| 15.1k| const int ref_w = ((ref_coded_width[pri_ref] + 7) >> 3) << 1; 3626| 15.1k| const int ref_h = ((f->refp[pri_ref].p.p.h + 7) >> 3) << 1; 3627| 15.1k| if (ref_w == f->bw && ref_h == f->bh) { ------------------ | Branch (3627:17): [True: 13.1k, False: 1.92k] | Branch (3627:35): [True: 12.0k, False: 1.17k] ------------------ 3628| 12.0k| f->prev_segmap_ref = c->refs[f->frame_hdr->refidx[pri_ref]].segmap; 3629| 12.0k| if (f->prev_segmap_ref) { ------------------ | Branch (3629:21): [True: 10.4k, False: 1.54k] ------------------ 3630| 10.4k| dav1d_ref_inc(f->prev_segmap_ref); 3631| 10.4k| f->prev_segmap = f->prev_segmap_ref->data; 3632| 10.4k| } 3633| 12.0k| } 3634| 15.1k| } 3635| | 3636| 19.6k| if (f->frame_hdr->segmentation.update_map) { ------------------ | Branch (3636:13): [True: 12.8k, False: 6.79k] ------------------ 3637| | // We're updating an existing map, but need somewhere to 3638| | // put the new values. Allocate them here (the data 3639| | // actually gets set elsewhere) 3640| 12.8k| f->cur_segmap_ref = dav1d_ref_create_using_pool(c->segmap_pool, 3641| 12.8k| sizeof(*f->cur_segmap) * f->b4_stride * 32 * f->sb128h); 3642| 12.8k| if (!f->cur_segmap_ref) { ------------------ | Branch (3642:17): [True: 0, False: 12.8k] ------------------ 3643| 0| dav1d_ref_dec(&f->prev_segmap_ref); 3644| 0| res = DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3645| 0| goto error; 3646| 0| } 3647| 12.8k| f->cur_segmap = f->cur_segmap_ref->data; 3648| 12.8k| } else if (f->prev_segmap_ref) { ------------------ | Branch (3648:20): [True: 4.62k, False: 2.17k] ------------------ 3649| | // We're not updating an existing map, and we have a valid 3650| | // reference. Use that. 3651| 4.62k| f->cur_segmap_ref = f->prev_segmap_ref; 3652| 4.62k| dav1d_ref_inc(f->cur_segmap_ref); 3653| 4.62k| f->cur_segmap = f->prev_segmap_ref->data; 3654| 4.62k| } else { 3655| | // We need to make a new map. Allocate one here and zero it out. 3656| 2.17k| const size_t segmap_size = sizeof(*f->cur_segmap) * f->b4_stride * 32 * f->sb128h; 3657| 2.17k| f->cur_segmap_ref = dav1d_ref_create_using_pool(c->segmap_pool, segmap_size); 3658| 2.17k| if (!f->cur_segmap_ref) { ------------------ | Branch (3658:17): [True: 0, False: 2.17k] ------------------ 3659| 0| res = DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 3660| 0| goto error; 3661| 0| } 3662| 2.17k| f->cur_segmap = f->cur_segmap_ref->data; 3663| 2.17k| memset(f->cur_segmap, 0, segmap_size); 3664| 2.17k| } 3665| 336k| } else { 3666| 336k| f->cur_segmap = NULL; 3667| 336k| f->cur_segmap_ref = NULL; 3668| 336k| f->prev_segmap_ref = NULL; 3669| 336k| } 3670| | 3671| | // update references etc. 3672| 356k| const unsigned refresh_frame_flags = f->frame_hdr->refresh_frame_flags; 3673| 3.20M| for (int i = 0; i < 8; i++) { ------------------ | Branch (3673:21): [True: 2.85M, False: 356k] ------------------ 3674| 2.85M| if (refresh_frame_flags & (1 << i)) { ------------------ | Branch (3674:13): [True: 2.40M, False: 444k] ------------------ 3675| 2.40M| if (c->refs[i].p.p.frame_hdr) ------------------ | Branch (3675:17): [True: 2.30M, False: 106k] ------------------ 3676| 2.30M| dav1d_thread_picture_unref(&c->refs[i].p); 3677| 2.40M| dav1d_thread_picture_ref(&c->refs[i].p, &f->sr_cur); 3678| | 3679| 2.40M| dav1d_cdf_thread_unref(&c->cdf[i]); 3680| 2.40M| if (f->frame_hdr->refresh_context) { ------------------ | Branch (3680:17): [True: 195k, False: 2.21M] ------------------ 3681| 195k| dav1d_cdf_thread_ref(&c->cdf[i], &f->out_cdf); 3682| 2.21M| } else { 3683| 2.21M| dav1d_cdf_thread_ref(&c->cdf[i], &f->in_cdf); 3684| 2.21M| } 3685| | 3686| 2.40M| dav1d_ref_dec(&c->refs[i].segmap); 3687| 2.40M| c->refs[i].segmap = f->cur_segmap_ref; 3688| 2.40M| if (f->cur_segmap_ref) ------------------ | Branch (3688:17): [True: 101k, False: 2.30M] ------------------ 3689| 101k| dav1d_ref_inc(f->cur_segmap_ref); 3690| 2.40M| dav1d_ref_dec(&c->refs[i].refmvs); 3691| 2.40M| if (!f->frame_hdr->allow_intrabc) { ------------------ | Branch (3691:17): [True: 581k, False: 1.82M] ------------------ 3692| 581k| c->refs[i].refmvs = f->mvs_ref; 3693| 581k| if (f->mvs_ref) ------------------ | Branch (3693:21): [True: 353k, False: 228k] ------------------ 3694| 353k| dav1d_ref_inc(f->mvs_ref); 3695| 581k| } 3696| 2.40M| memcpy(c->refs[i].refpoc, f->refpoc, sizeof(f->refpoc)); 3697| 2.40M| } 3698| 2.85M| } 3699| | 3700| 356k| if (c->n_fc == 1) { ------------------ | Branch (3700:9): [True: 0, False: 356k] ------------------ 3701| 0| if ((res = dav1d_decode_frame(f)) < 0) { ------------------ | Branch (3701:13): [True: 0, False: 0] ------------------ 3702| 0| dav1d_thread_picture_unref(&c->out); 3703| 0| for (int i = 0; i < 8; i++) { ------------------ | Branch (3703:29): [True: 0, False: 0] ------------------ 3704| 0| if (refresh_frame_flags & (1 << i)) { ------------------ | Branch (3704:21): [True: 0, False: 0] ------------------ 3705| 0| if (c->refs[i].p.p.frame_hdr) ------------------ | Branch (3705:25): [True: 0, False: 0] ------------------ 3706| 0| dav1d_thread_picture_unref(&c->refs[i].p); 3707| 0| dav1d_cdf_thread_unref(&c->cdf[i]); 3708| 0| dav1d_ref_dec(&c->refs[i].segmap); 3709| 0| dav1d_ref_dec(&c->refs[i].refmvs); 3710| 0| } 3711| 0| } 3712| 0| goto error; 3713| 0| } 3714| 356k| } else { 3715| 356k| dav1d_task_frame_init(f); 3716| 356k| pthread_mutex_unlock(&c->task_thread.lock); 3717| 356k| } 3718| | 3719| 356k| return 0; 3720| 3.01k|error: 3721| 3.01k| atomic_init(&f->task_thread.error, 1); 3722| 3.01k| dav1d_cdf_thread_unref(&f->in_cdf); 3723| 3.01k| if (f->frame_hdr->refresh_context) ------------------ | Branch (3723:9): [True: 2.49k, False: 512] ------------------ 3724| 2.49k| dav1d_cdf_thread_unref(&f->out_cdf); 3725| 24.0k| for (int i = 0; i < 7; i++) { ------------------ | Branch (3725:21): [True: 21.0k, False: 3.01k] ------------------ 3726| 21.0k| if (f->refp[i].p.frame_hdr) ------------------ | Branch (3726:13): [True: 0, False: 21.0k] ------------------ 3727| 0| dav1d_thread_picture_unref(&f->refp[i]); 3728| 21.0k| dav1d_ref_dec(&f->ref_mvs_ref[i]); 3729| 21.0k| } 3730| 3.01k| if (c->n_fc == 1) ------------------ | Branch (3730:9): [True: 0, False: 3.01k] ------------------ 3731| 0| dav1d_thread_picture_unref(&c->out); 3732| 3.01k| else 3733| 3.01k| dav1d_thread_picture_unref(out_delayed); 3734| 3.01k| dav1d_picture_unref_internal(&f->cur); 3735| 3.01k| dav1d_thread_picture_unref(&f->sr_cur); 3736| 3.01k| dav1d_ref_dec(&f->mvs_ref); 3737| 3.01k| dav1d_ref_dec(&f->seq_hdr_ref); 3738| 3.01k| dav1d_ref_dec(&f->frame_hdr_ref); 3739| 3.01k| dav1d_data_props_copy(&c->cached_error_props, &c->in.m); 3740| | 3741| 3.01k| for (int i = 0; i < f->n_tile_data; i++) ------------------ | Branch (3741:21): [True: 0, False: 3.01k] ------------------ 3742| 0| dav1d_data_unref_internal(&f->tile[i].data); 3743| 3.01k| f->n_tile_data = 0; 3744| | 3745| 3.01k| if (c->n_fc > 1) ------------------ | Branch (3745:9): [True: 3.01k, False: 0] ------------------ 3746| 3.01k| pthread_mutex_unlock(&c->task_thread.lock); 3747| | 3748| 3.01k| return res; 3749| 356k|} decode.c:reset_context: 2392| 7.17M|static void reset_context(BlockContext *const ctx, const int keyframe, const int pass) { 2393| 7.17M| memset(ctx->intra, keyframe, sizeof(ctx->intra)); 2394| 7.17M| memset(ctx->uvmode, DC_PRED, sizeof(ctx->uvmode)); 2395| 7.17M| if (keyframe) ------------------ | Branch (2395:9): [True: 2.58M, False: 4.59M] ------------------ 2396| 2.58M| memset(ctx->mode, DC_PRED, sizeof(ctx->mode)); 2397| | 2398| 7.17M| if (pass == 2) return; ------------------ | Branch (2398:9): [True: 3.53M, False: 3.64M] ------------------ 2399| | 2400| 3.64M| memset(ctx->partition, 0, sizeof(ctx->partition)); 2401| 3.64M| memset(ctx->skip, 0, sizeof(ctx->skip)); 2402| 3.64M| memset(ctx->skip_mode, 0, sizeof(ctx->skip_mode)); 2403| 3.64M| memset(ctx->tx_lpf_y, 2, sizeof(ctx->tx_lpf_y)); 2404| 3.64M| memset(ctx->tx_lpf_uv, 1, sizeof(ctx->tx_lpf_uv)); 2405| 3.64M| memset(ctx->tx_intra, -1, sizeof(ctx->tx_intra)); 2406| 3.64M| memset(ctx->tx, TX_64X64, sizeof(ctx->tx)); 2407| 3.64M| if (!keyframe) { ------------------ | Branch (2407:9): [True: 2.31M, False: 1.32M] ------------------ 2408| 2.31M| memset(ctx->ref, -1, sizeof(ctx->ref)); 2409| 2.31M| memset(ctx->comp_type, 0, sizeof(ctx->comp_type)); 2410| 2.31M| memset(ctx->mode, NEARESTMV, sizeof(ctx->mode)); 2411| 2.31M| } 2412| 3.64M| memset(ctx->lcoef, 0x40, sizeof(ctx->lcoef)); 2413| 3.64M| memset(ctx->ccoef, 0x40, sizeof(ctx->ccoef)); 2414| 3.64M| memset(ctx->filter, DAV1D_N_SWITCHABLE_FILTERS, sizeof(ctx->filter)); 2415| 3.64M| memset(ctx->seg_pred, 0, sizeof(ctx->seg_pred)); 2416| 3.64M| memset(ctx->pal_sz, 0, sizeof(ctx->pal_sz)); 2417| 3.64M|} decode.c:decode_sb: 2121| 13.1M|{ 2122| 13.1M| const Dav1dFrameContext *const f = t->f; 2123| 13.1M| Dav1dTileState *const ts = t->ts; 2124| 13.1M| const int hsz = 16 >> bl; 2125| 13.1M| const int have_h_split = f->bw > t->bx + hsz; 2126| 13.1M| const int have_v_split = f->bh > t->by + hsz; 2127| | 2128| 13.1M| if (!have_h_split && !have_v_split) { ------------------ | Branch (2128:9): [True: 899k, False: 12.2M] | Branch (2128:26): [True: 314k, False: 585k] ------------------ 2129| 314k| assert(bl < BL_8X8); ------------------ | Branch (2129:9): [True: 314k, False: 0] ------------------ 2130| 314k| return decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 0)); ------------------ | | 51| 314k| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ 2131| 314k| } 2132| | 2133| 12.8M| uint16_t *pc; 2134| 12.8M| enum BlockPartition bp; 2135| 12.8M| int ctx, bx8, by8; 2136| 12.8M| if (t->frame_thread.pass != 2) { ------------------ | Branch (2136:9): [True: 8.89M, False: 3.91M] ------------------ 2137| 8.89M| if (0 && bl == BL_64X64) ------------------ | Branch (2137:13): [Folded, False: 8.89M] | Branch (2137:18): [True: 0, False: 0] ------------------ 2138| 0| printf("poc=%d,y=%d,x=%d,bl=%d,r=%d\n", 2139| 0| f->frame_hdr->frame_offset, t->by, t->bx, bl, ts->msac.rng); 2140| 8.89M| bx8 = (t->bx & 31) >> 1; 2141| 8.89M| by8 = (t->by & 31) >> 1; 2142| 8.89M| ctx = get_partition_ctx(t->a, &t->l, bl, by8, bx8); 2143| 8.89M| pc = ts->cdf.m.partition[bl][ctx]; 2144| 8.89M| } 2145| | 2146| 12.8M| if (have_h_split && have_v_split) { ------------------ | Branch (2146:9): [True: 12.2M, False: 582k] | Branch (2146:25): [True: 11.8M, False: 416k] ------------------ 2147| 11.8M| if (t->frame_thread.pass == 2) { ------------------ | Branch (2147:13): [True: 3.66M, False: 8.13M] ------------------ 2148| 3.66M| const Av1Block *const b = &f->frame_thread.b[t->by * f->b4_stride + t->bx]; 2149| 3.66M| bp = b->bl == bl ? b->bp : PARTITION_SPLIT; ------------------ | Branch (2149:18): [True: 3.32M, False: 343k] ------------------ 2150| 8.13M| } else { 2151| 8.13M| bp = dav1d_msac_decode_symbol_adapt16(&ts->msac, pc, ------------------ | | 57| 8.13M|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 2152| 8.13M| dav1d_partition_type_count[bl]); 2153| 8.13M| if (f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I422 && ------------------ | Branch (2153:17): [True: 72.4k, False: 8.06M] ------------------ 2154| 72.4k| (bp == PARTITION_V || bp == PARTITION_V4 || ------------------ | Branch (2154:18): [True: 470, False: 72.0k] | Branch (2154:39): [True: 1.37k, False: 70.6k] ------------------ 2155| 70.6k| bp == PARTITION_T_LEFT_SPLIT || bp == PARTITION_T_RIGHT_SPLIT)) ------------------ | Branch (2155:18): [True: 374, False: 70.2k] | Branch (2155:50): [True: 55.7k, False: 14.5k] ------------------ 2156| 57.9k| { 2157| 57.9k| return 1; 2158| 57.9k| } 2159| 8.08M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 8.08M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 8.08M] | | ------------------ | | 35| 8.08M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 8.08M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2160| 0| printf("poc=%d,y=%d,x=%d,bl=%d,ctx=%d,bp=%d: r=%d\n", 2161| 0| f->frame_hdr->frame_offset, t->by, t->bx, bl, ctx, bp, 2162| 0| ts->msac.rng); 2163| 8.08M| } 2164| 11.7M| const uint8_t *const b = dav1d_block_sizes[bl][bp]; 2165| | 2166| 11.7M| switch (bp) { 2167| 7.01M| case PARTITION_NONE: ------------------ | Branch (2167:9): [True: 7.01M, False: 4.73M] ------------------ 2168| 7.01M| if (decode_b(t, bl, b[0], PARTITION_NONE, node->o)) ------------------ | Branch (2168:17): [True: 1.95k, False: 7.01M] ------------------ 2169| 1.95k| return -1; 2170| 7.01M| break; 2171| 7.01M| case PARTITION_H: ------------------ | Branch (2171:9): [True: 902k, False: 10.8M] ------------------ 2172| 902k| if (decode_b(t, bl, b[0], PARTITION_H, node->h[0])) ------------------ | Branch (2172:17): [True: 553, False: 901k] ------------------ 2173| 553| return -1; 2174| 901k| t->by += hsz; 2175| 901k| if (decode_b(t, bl, b[0], PARTITION_H, node->h[1])) ------------------ | Branch (2175:17): [True: 444, False: 901k] ------------------ 2176| 444| return -1; 2177| 901k| t->by -= hsz; 2178| 901k| break; 2179| 617k| case PARTITION_V: ------------------ | Branch (2179:9): [True: 617k, False: 11.1M] ------------------ 2180| 617k| if (decode_b(t, bl, b[0], PARTITION_V, node->v[0])) ------------------ | Branch (2180:17): [True: 1.65k, False: 616k] ------------------ 2181| 1.65k| return -1; 2182| 616k| t->bx += hsz; 2183| 616k| if (decode_b(t, bl, b[0], PARTITION_V, node->v[1])) ------------------ | Branch (2183:17): [True: 274, False: 616k] ------------------ 2184| 274| return -1; 2185| 616k| t->bx -= hsz; 2186| 616k| break; 2187| 1.88M| case PARTITION_SPLIT: ------------------ | Branch (2187:9): [True: 1.88M, False: 9.86M] ------------------ 2188| 1.88M| if (bl == BL_8X8) { ------------------ | Branch (2188:17): [True: 286k, False: 1.59M] ------------------ 2189| 286k| const EdgeTip *const tip = (const EdgeTip *) node; 2190| 286k| assert(hsz == 1); ------------------ | Branch (2190:17): [True: 286k, False: 1] ------------------ 2191| 286k| if (decode_b(t, bl, BS_4x4, PARTITION_SPLIT, EDGE_ALL_TR_AND_BL)) ------------------ | Branch (2191:21): [True: 292, False: 286k] ------------------ 2192| 292| return -1; 2193| 286k| const enum Filter2d tl_filter = t->tl_4x4_filter; 2194| 286k| t->bx++; 2195| 286k| if (decode_b(t, bl, BS_4x4, PARTITION_SPLIT, tip->split[0])) ------------------ | Branch (2195:21): [True: 382, False: 285k] ------------------ 2196| 382| return -1; 2197| 285k| t->bx--; 2198| 285k| t->by++; 2199| 285k| if (decode_b(t, bl, BS_4x4, PARTITION_SPLIT, tip->split[1])) ------------------ | Branch (2199:21): [True: 162, False: 285k] ------------------ 2200| 162| return -1; 2201| 285k| t->bx++; 2202| 285k| t->tl_4x4_filter = tl_filter; 2203| 285k| if (decode_b(t, bl, BS_4x4, PARTITION_SPLIT, tip->split[2])) ------------------ | Branch (2203:21): [True: 74, False: 285k] ------------------ 2204| 74| return -1; 2205| 285k| t->bx--; 2206| 285k| t->by--; 2207| 285k|#if ARCH_X86_64 2208| 285k| if (t->frame_thread.pass) { ------------------ | Branch (2208:21): [True: 285k, False: 18.4E] ------------------ 2209| | /* In 8-bit mode with 2-pass decoding the coefficient buffer 2210| | * can end up misaligned due to skips here. Work around 2211| | * the issue by explicitly realigning the buffer. */ 2212| 285k| const int p = t->frame_thread.pass & 1; 2213| 285k| ts->frame_thread[p].cf = 2214| 285k| (void*)(((uintptr_t)ts->frame_thread[p].cf + 63) & ~63); 2215| 285k| } 2216| 285k|#endif 2217| 1.59M| } else { 2218| 1.59M| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 0))) ------------------ | | 51| 1.59M| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2218:21): [True: 2.95k, False: 1.59M] ------------------ 2219| 2.95k| return 1; 2220| 1.59M| t->bx += hsz; 2221| 1.59M| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 1))) ------------------ | | 51| 1.59M| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2221:21): [True: 2.87k, False: 1.59M] ------------------ 2222| 2.87k| return 1; 2223| 1.59M| t->bx -= hsz; 2224| 1.59M| t->by += hsz; 2225| 1.59M| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 2))) ------------------ | | 51| 1.59M| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2225:21): [True: 2.20k, False: 1.58M] ------------------ 2226| 2.20k| return 1; 2227| 1.58M| t->bx += hsz; 2228| 1.58M| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 3))) ------------------ | | 51| 1.58M| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2228:21): [True: 3.14k, False: 1.58M] ------------------ 2229| 3.14k| return 1; 2230| 1.58M| t->bx -= hsz; 2231| 1.58M| t->by -= hsz; 2232| 1.58M| } 2233| 1.87M| break; 2234| 1.87M| case PARTITION_T_TOP_SPLIT: { ------------------ | Branch (2234:9): [True: 145k, False: 11.6M] ------------------ 2235| 145k| if (decode_b(t, bl, b[0], PARTITION_T_TOP_SPLIT, EDGE_ALL_TR_AND_BL)) ------------------ | Branch (2235:17): [True: 167, False: 145k] ------------------ 2236| 167| return -1; 2237| 145k| t->bx += hsz; 2238| 145k| if (decode_b(t, bl, b[0], PARTITION_T_TOP_SPLIT, node->v[1])) ------------------ | Branch (2238:17): [True: 238, False: 145k] ------------------ 2239| 238| return -1; 2240| 145k| t->bx -= hsz; 2241| 145k| t->by += hsz; 2242| 145k| if (decode_b(t, bl, b[1], PARTITION_T_TOP_SPLIT, node->h[1])) ------------------ | Branch (2242:17): [True: 436, False: 145k] ------------------ 2243| 436| return -1; 2244| 145k| t->by -= hsz; 2245| 145k| break; 2246| 145k| } 2247| 154k| case PARTITION_T_BOTTOM_SPLIT: { ------------------ | Branch (2247:9): [True: 154k, False: 11.5M] ------------------ 2248| 154k| if (decode_b(t, bl, b[0], PARTITION_T_BOTTOM_SPLIT, node->h[0])) ------------------ | Branch (2248:17): [True: 74, False: 154k] ------------------ 2249| 74| return -1; 2250| 154k| t->by += hsz; 2251| 154k| if (decode_b(t, bl, b[1], PARTITION_T_BOTTOM_SPLIT, node->v[0])) ------------------ | Branch (2251:17): [True: 79, False: 154k] ------------------ 2252| 79| return -1; 2253| 154k| t->bx += hsz; 2254| 154k| if (decode_b(t, bl, b[1], PARTITION_T_BOTTOM_SPLIT, 0)) ------------------ | Branch (2254:17): [True: 97, False: 154k] ------------------ 2255| 97| return -1; 2256| 154k| t->bx -= hsz; 2257| 154k| t->by -= hsz; 2258| 154k| break; 2259| 154k| } 2260| 100k| case PARTITION_T_LEFT_SPLIT: { ------------------ | Branch (2260:9): [True: 100k, False: 11.6M] ------------------ 2261| 100k| if (decode_b(t, bl, b[0], PARTITION_T_LEFT_SPLIT, EDGE_ALL_TR_AND_BL)) ------------------ | Branch (2261:17): [True: 498, False: 99.6k] ------------------ 2262| 498| return -1; 2263| 99.6k| t->by += hsz; 2264| 99.6k| if (decode_b(t, bl, b[0], PARTITION_T_LEFT_SPLIT, node->h[1])) ------------------ | Branch (2264:17): [True: 227, False: 99.4k] ------------------ 2265| 227| return -1; 2266| 99.4k| t->by -= hsz; 2267| 99.4k| t->bx += hsz; 2268| 99.4k| if (decode_b(t, bl, b[1], PARTITION_T_LEFT_SPLIT, node->v[1])) ------------------ | Branch (2268:17): [True: 318, False: 99.1k] ------------------ 2269| 318| return -1; 2270| 99.1k| t->bx -= hsz; 2271| 99.1k| break; 2272| 99.4k| } 2273| 281k| case PARTITION_T_RIGHT_SPLIT: { ------------------ | Branch (2273:9): [True: 281k, False: 11.4M] ------------------ 2274| 281k| if (decode_b(t, bl, b[0], PARTITION_T_RIGHT_SPLIT, node->v[0])) ------------------ | Branch (2274:17): [True: 588, False: 280k] ------------------ 2275| 588| return -1; 2276| 280k| t->bx += hsz; 2277| 280k| if (decode_b(t, bl, b[1], PARTITION_T_RIGHT_SPLIT, node->h[0])) ------------------ | Branch (2277:17): [True: 341, False: 280k] ------------------ 2278| 341| return -1; 2279| 280k| t->by += hsz; 2280| 280k| if (decode_b(t, bl, b[1], PARTITION_T_RIGHT_SPLIT, 0)) ------------------ | Branch (2280:17): [True: 210, False: 279k] ------------------ 2281| 210| return -1; 2282| 279k| t->by -= hsz; 2283| 279k| t->bx -= hsz; 2284| 279k| break; 2285| 280k| } 2286| 352k| case PARTITION_H4: { ------------------ | Branch (2286:9): [True: 352k, False: 11.3M] ------------------ 2287| 352k| const EdgeBranch *const branch = (const EdgeBranch *) node; 2288| 352k| if (decode_b(t, bl, b[0], PARTITION_H4, node->h[0])) ------------------ | Branch (2288:17): [True: 345, False: 352k] ------------------ 2289| 345| return -1; 2290| 352k| t->by += hsz >> 1; 2291| 352k| if (decode_b(t, bl, b[0], PARTITION_H4, branch->h4)) ------------------ | Branch (2291:17): [True: 206, False: 352k] ------------------ 2292| 206| return -1; 2293| 352k| t->by += hsz >> 1; 2294| 352k| if (decode_b(t, bl, b[0], PARTITION_H4, EDGE_ALL_LEFT_HAS_BOTTOM)) ------------------ | Branch (2294:17): [True: 341, False: 351k] ------------------ 2295| 341| return -1; 2296| 351k| t->by += hsz >> 1; 2297| 351k| if (t->by < f->bh) ------------------ | Branch (2297:17): [True: 343k, False: 8.73k] ------------------ 2298| 343k| if (decode_b(t, bl, b[0], PARTITION_H4, node->h[1])) ------------------ | Branch (2298:21): [True: 185, False: 342k] ------------------ 2299| 185| return -1; 2300| 351k| t->by -= hsz * 3 >> 1; 2301| 351k| break; 2302| 351k| } 2303| 317k| case PARTITION_V4: { ------------------ | Branch (2303:9): [True: 317k, False: 11.4M] ------------------ 2304| 317k| const EdgeBranch *const branch = (const EdgeBranch *) node; 2305| 317k| if (decode_b(t, bl, b[0], PARTITION_V4, node->v[0])) ------------------ | Branch (2305:17): [True: 407, False: 317k] ------------------ 2306| 407| return -1; 2307| 317k| t->bx += hsz >> 1; 2308| 317k| if (decode_b(t, bl, b[0], PARTITION_V4, branch->v4)) ------------------ | Branch (2308:17): [True: 797, False: 316k] ------------------ 2309| 797| return -1; 2310| 316k| t->bx += hsz >> 1; 2311| 316k| if (decode_b(t, bl, b[0], PARTITION_V4, EDGE_ALL_TOP_HAS_RIGHT)) ------------------ | Branch (2311:17): [True: 86, False: 316k] ------------------ 2312| 86| return -1; 2313| 316k| t->bx += hsz >> 1; 2314| 316k| if (t->bx < f->bw) ------------------ | Branch (2314:17): [True: 305k, False: 10.3k] ------------------ 2315| 305k| if (decode_b(t, bl, b[0], PARTITION_V4, node->v[1])) ------------------ | Branch (2315:21): [True: 523, False: 305k] ------------------ 2316| 523| return -1; 2317| 315k| t->bx -= hsz * 3 >> 1; 2318| 315k| break; 2319| 316k| } 2320| 0| default: assert(0); ------------------ | Branch (2320:9): [True: 0, False: 11.7M] | Branch (2320:18): [Folded, False: 0] ------------------ 2321| 11.7M| } 2322| 11.7M| } else if (have_h_split) { ------------------ | Branch (2322:16): [True: 418k, False: 580k] ------------------ 2323| 418k| unsigned is_split; 2324| 418k| if (t->frame_thread.pass == 2) { ------------------ | Branch (2324:13): [True: 56.2k, False: 362k] ------------------ 2325| 56.2k| const Av1Block *const b = &f->frame_thread.b[t->by * f->b4_stride + t->bx]; 2326| 56.2k| is_split = b->bl != bl; 2327| 362k| } else { 2328| 362k| is_split = dav1d_msac_decode_bool(&ts->msac, ------------------ | | 54| 362k|#define dav1d_msac_decode_bool dav1d_msac_decode_bool_sse2 ------------------ 2329| 362k| gather_top_partition_prob(pc, bl)); 2330| 362k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 362k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 362k] | | ------------------ | | 35| 362k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 362k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2331| 0| printf("poc=%d,y=%d,x=%d,bl=%d,ctx=%d,bp=%d: r=%d\n", 2332| 0| f->frame_hdr->frame_offset, t->by, t->bx, bl, ctx, 2333| 0| is_split ? PARTITION_SPLIT : PARTITION_H, ts->msac.rng); ------------------ | Branch (2333:24): [True: 0, False: 0] ------------------ 2334| 362k| } 2335| | 2336| 418k| assert(bl < BL_8X8); ------------------ | Branch (2336:9): [True: 418k, False: 18.4E] ------------------ 2337| 418k| if (is_split) { ------------------ | Branch (2337:13): [True: 266k, False: 152k] ------------------ 2338| 266k| bp = PARTITION_SPLIT; 2339| 266k| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 0))) return 1; ------------------ | | 51| 266k| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2339:17): [True: 2.39k, False: 264k] ------------------ 2340| 264k| t->bx += hsz; 2341| 264k| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 1))) return 1; ------------------ | | 51| 264k| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2341:17): [True: 1.63k, False: 262k] ------------------ 2342| 262k| t->bx -= hsz; 2343| 262k| } else { 2344| 152k| bp = PARTITION_H; 2345| 152k| if (decode_b(t, bl, dav1d_block_sizes[bl][PARTITION_H][0], ------------------ | Branch (2345:17): [True: 317, False: 152k] ------------------ 2346| 152k| PARTITION_H, node->h[0])) 2347| 317| return -1; 2348| 152k| } 2349| 580k| } else { 2350| 580k| assert(have_v_split); ------------------ | Branch (2350:9): [True: 585k, False: 18.4E] ------------------ 2351| 585k| unsigned is_split; 2352| 585k| if (t->frame_thread.pass == 2) { ------------------ | Branch (2352:13): [True: 214k, False: 371k] ------------------ 2353| 214k| const Av1Block *const b = &f->frame_thread.b[t->by * f->b4_stride + t->bx]; 2354| 214k| is_split = b->bl != bl; 2355| 371k| } else { 2356| 371k| is_split = dav1d_msac_decode_bool(&ts->msac, ------------------ | | 54| 371k|#define dav1d_msac_decode_bool dav1d_msac_decode_bool_sse2 ------------------ 2357| 371k| gather_left_partition_prob(pc, bl)); 2358| 371k| if (f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I422 && !is_split) ------------------ | Branch (2358:17): [True: 60.1k, False: 311k] | Branch (2358:63): [True: 620, False: 59.5k] ------------------ 2359| 620| return 1; 2360| 370k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 370k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 370k] | | ------------------ | | 35| 370k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 370k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2361| 0| printf("poc=%d,y=%d,x=%d,bl=%d,ctx=%d,bp=%d: r=%d\n", 2362| 0| f->frame_hdr->frame_offset, t->by, t->bx, bl, ctx, 2363| 0| is_split ? PARTITION_SPLIT : PARTITION_V, ts->msac.rng); ------------------ | Branch (2363:24): [True: 0, False: 0] ------------------ 2364| 370k| } 2365| | 2366| 585k| assert(bl < BL_8X8); ------------------ | Branch (2366:9): [True: 585k, False: 18.4E] ------------------ 2367| 585k| if (is_split) { ------------------ | Branch (2367:13): [True: 208k, False: 377k] ------------------ 2368| 208k| bp = PARTITION_SPLIT; 2369| 208k| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 0))) return 1; ------------------ | | 51| 208k| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2369:17): [True: 65.5k, False: 142k] ------------------ 2370| 142k| t->by += hsz; 2371| 142k| if (decode_sb(t, bl + 1, INTRA_EDGE_SPLIT(node, 2))) return 1; ------------------ | | 51| 142k| ((const EdgeNode*)((uintptr_t)(n) + ((const EdgeBranch*)(n))->split_offset[i])) ------------------ | Branch (2371:17): [True: 8.77k, False: 133k] ------------------ 2372| 133k| t->by -= hsz; 2373| 377k| } else { 2374| 377k| bp = PARTITION_V; 2375| 377k| if (decode_b(t, bl, dav1d_block_sizes[bl][PARTITION_V][0], ------------------ | Branch (2375:17): [True: 132, False: 377k] ------------------ 2376| 377k| PARTITION_V, node->v[0])) 2377| 132| return -1; 2378| 377k| } 2379| 585k| } 2380| | 2381| 12.6M| if (t->frame_thread.pass != 2 && (bp != PARTITION_SPLIT || bl == BL_8X8)) { ------------------ | Branch (2381:9): [True: 8.74M, False: 3.89M] | Branch (2381:39): [True: 6.93M, False: 1.80M] | Branch (2381:64): [True: 235k, False: 1.57M] ------------------ 2382| 7.17M|#define set_ctx(rep_macro) \ 2383| 7.17M| rep_macro(t->a->partition, bx8, dav1d_al_part_ctx[0][bl][bp]); \ 2384| 7.17M| rep_macro(t->l.partition, by8, dav1d_al_part_ctx[1][bl][bp]) 2385| 7.17M| case_set_upto16(ulog2(hsz)); ------------------ | | 80| 7.17M| switch (var) { \ | | 81| 1.37M| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 2383| 1.37M| rep_macro(t->a->partition, bx8, dav1d_al_part_ctx[0][bl][bp]); \ | | | | ------------------ | | | | | | 81| 1.37M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 1.37M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 2384| 1.37M| rep_macro(t->l.partition, by8, dav1d_al_part_ctx[1][bl][bp]) | | | | ------------------ | | | | | | 81| 1.37M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 1.37M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (81:5): [True: 1.37M, False: 5.80M] | | ------------------ | | 82| 1.82M| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 2383| 1.82M| rep_macro(t->a->partition, bx8, dav1d_al_part_ctx[0][bl][bp]); \ | | | | ------------------ | | | | | | 82| 1.82M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.82M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 2384| 1.82M| rep_macro(t->l.partition, by8, dav1d_al_part_ctx[1][bl][bp]) | | | | ------------------ | | | | | | 82| 1.82M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.82M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (82:5): [True: 1.82M, False: 5.34M] | | ------------------ | | 83| 937k| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 2383| 937k| rep_macro(t->a->partition, bx8, dav1d_al_part_ctx[0][bl][bp]); \ | | | | ------------------ | | | | | | 83| 937k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 937k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 2384| 937k| rep_macro(t->l.partition, by8, dav1d_al_part_ctx[1][bl][bp]) | | | | ------------------ | | | | | | 83| 937k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 937k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (83:5): [True: 937k, False: 6.23M] | | ------------------ | | 84| 1.81M| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 2383| 1.81M| rep_macro(t->a->partition, bx8, dav1d_al_part_ctx[0][bl][bp]); \ | | | | ------------------ | | | | | | 84| 1.81M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.81M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 2384| 1.81M| rep_macro(t->l.partition, by8, dav1d_al_part_ctx[1][bl][bp]) | | | | ------------------ | | | | | | 84| 1.81M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.81M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (84:5): [True: 1.81M, False: 5.35M] | | ------------------ | | 85| 1.22M| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 2383| 1.22M| rep_macro(t->a->partition, bx8, dav1d_al_part_ctx[0][bl][bp]); \ | | | | ------------------ | | | | | | 85| 1.22M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.22M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.22M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.22M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.22M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 2384| 1.22M| rep_macro(t->l.partition, by8, dav1d_al_part_ctx[1][bl][bp]) | | | | ------------------ | | | | | | 85| 1.22M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.22M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.22M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.22M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.22M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (85:5): [True: 1.22M, False: 5.95M] | | ------------------ | | 86| 0| default: assert(0); \ | | ------------------ | | | Branch (86:5): [True: 0, False: 7.17M] | | ------------------ | | 87| 7.17M| } ------------------ | Branch (2385:9): [Folded, False: 0] ------------------ 2386| 7.17M|#undef set_ctx 2387| 7.17M| } 2388| | 2389| 12.6M| return 0; 2390| 12.6M|} decode.c:decode_b: 687| 16.3M| const enum EdgeFlags intra_edge_flags) { 688| 16.3M| Dav1dTileState *const ts = t->ts; 689| 16.3M| const Dav1dFrameContext *const f = t->f; 690| 16.3M| Av1Block b_mem, *const b = t->frame_thread.pass ? ------------------ | Branch (690:32): [True: 16.3M, False: 18.4E] ------------------ 691| 18.4E| &f->frame_thread.b[t->by * f->b4_stride + t->bx] : &b_mem; 692| 16.3M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 693| 16.3M| const int bx4 = t->bx & 31, by4 = t->by & 31; 694| 16.3M| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 695| 16.3M| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 696| 16.3M| const int cbx4 = bx4 >> ss_hor, cby4 = by4 >> ss_ver; 697| 16.3M| const int bw4 = b_dim[0], bh4 = b_dim[1]; 698| 16.3M| const int w4 = imin(bw4, f->bw - t->bx), h4 = imin(bh4, f->bh - t->by); 699| 16.3M| const int cbw4 = (bw4 + ss_hor) >> ss_hor, cbh4 = (bh4 + ss_ver) >> ss_ver; 700| 16.3M| const int have_left = t->bx > ts->tiling.col_start; 701| 16.3M| const int have_top = t->by > ts->tiling.row_start; 702| 16.3M| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 && ------------------ | Branch (702:28): [True: 10.6M, False: 5.73M] ------------------ 703| 10.6M| (bw4 > ss_hor || t->bx & 1) && ------------------ | Branch (703:29): [True: 9.69M, False: 946k] | Branch (703:45): [True: 472k, False: 473k] ------------------ 704| 10.1M| (bh4 > ss_ver || t->by & 1); ------------------ | Branch (704:29): [True: 9.29M, False: 881k] | Branch (704:45): [True: 440k, False: 441k] ------------------ 705| | 706| 16.3M| if (t->frame_thread.pass == 2) { ------------------ | Branch (706:9): [True: 4.56M, False: 11.8M] ------------------ 707| 4.56M| if (b->intra) { ------------------ | Branch (707:13): [True: 1.73M, False: 2.82M] ------------------ 708| 1.73M| f->bd_fn.recon_b_intra(t, bs, intra_edge_flags, b); 709| | 710| 1.73M| const enum IntraPredMode y_mode_nofilt = 711| 1.73M| b->y_mode == FILTER_PRED ? DC_PRED : b->y_mode; ------------------ | Branch (711:17): [True: 249k, False: 1.48M] ------------------ 712| 1.73M|#define set_ctx(rep_macro) \ 713| 1.73M| rep_macro(edge->mode, off, y_mode_nofilt); \ 714| 1.73M| rep_macro(edge->intra, off, 1) 715| 1.73M| BlockContext *edge = t->a; 716| 5.20M| for (int i = 0, off = bx4; i < 2; i++, off = by4, edge = &t->l) { ------------------ | Branch (716:40): [True: 3.47M, False: 1.73M] ------------------ 717| 3.47M| case_set(b_dim[2 + i]); ------------------ | | 70| 3.47M| switch (var) { \ | | 71| 519k| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 713| 519k| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 71| 519k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 519k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 714| 519k| rep_macro(edge->intra, off, 1) | | | | ------------------ | | | | | | 71| 519k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 519k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (71:5): [True: 519k, False: 2.95M] | | ------------------ | | 72| 712k| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 713| 712k| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 72| 712k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 712k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 714| 712k| rep_macro(edge->intra, off, 1) | | | | ------------------ | | | | | | 72| 712k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 712k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (72:5): [True: 712k, False: 2.75M] | | ------------------ | | 73| 586k| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 713| 586k| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 73| 586k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 586k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 714| 586k| rep_macro(edge->intra, off, 1) | | | | ------------------ | | | | | | 73| 586k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 586k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (73:5): [True: 586k, False: 2.88M] | | ------------------ | | 74| 406k| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 713| 406k| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 74| 406k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 406k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 714| 406k| rep_macro(edge->intra, off, 1) | | | | ------------------ | | | | | | 74| 406k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 406k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (74:5): [True: 406k, False: 3.06M] | | ------------------ | | 75| 810k| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 713| 810k| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 75| 810k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 810k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 810k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 810k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 810k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 714| 810k| rep_macro(edge->intra, off, 1) | | | | ------------------ | | | | | | 75| 810k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 810k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 810k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 810k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 810k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (75:5): [True: 810k, False: 2.65M] | | ------------------ | | 76| 435k| case 5: set_ctx(set_ctx32); break; \ | | ------------------ | | | | 713| 435k| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 76| 435k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 435k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 435k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 435k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 435k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 714| 435k| rep_macro(edge->intra, off, 1) | | | | ------------------ | | | | | | 76| 435k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 435k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 435k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 435k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 435k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (76:5): [True: 435k, False: 3.03M] | | ------------------ | | 77| 0| default: assert(0); \ | | ------------------ | | | Branch (77:5): [True: 0, False: 3.47M] | | ------------------ | | 78| 3.47M| } ------------------ | Branch (717:17): [Folded, False: 0] ------------------ 718| 3.47M| } 719| 1.73M|#undef set_ctx 720| 1.73M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { ------------------ | | 36| 1.73M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 269k, False: 1.46M] | | ------------------ ------------------ 721| 269k| refmvs_block *const r = &t->rt.r[(t->by & 31) + 5 + bh4 - 1][t->bx]; 722| 1.38M| for (int x = 0; x < bw4; x++) { ------------------ | Branch (722:33): [True: 1.11M, False: 269k] ------------------ 723| 1.11M| r[x].ref.ref[0] = 0; 724| 1.11M| r[x].bs = bs; 725| 1.11M| } 726| 269k| refmvs_block *const *rr = &t->rt.r[(t->by & 31) + 5]; 727| 1.10M| for (int y = 0; y < bh4 - 1; y++) { ------------------ | Branch (727:33): [True: 830k, False: 269k] ------------------ 728| 830k| rr[y][t->bx + bw4 - 1].ref.ref[0] = 0; 729| 830k| rr[y][t->bx + bw4 - 1].bs = bs; 730| 830k| } 731| 269k| } 732| | 733| 1.73M| if (has_chroma) { ------------------ | Branch (733:17): [True: 976k, False: 758k] ------------------ 734| 976k| uint8_t uv_mode = b->uv_mode; 735| 976k| dav1d_memset_pow2[ulog2(cbw4)](&t->a->uvmode[cbx4], uv_mode); 736| 976k| dav1d_memset_pow2[ulog2(cbh4)](&t->l.uvmode[cby4], uv_mode); 737| 976k| } 738| 2.82M| } else { 739| 2.82M| if (IS_INTER_OR_SWITCH(f->frame_hdr) /* not intrabc */ && ------------------ | | 36| 5.65M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 2.81M, False: 17.5k] | | ------------------ ------------------ 740| 2.81M| b->comp_type == COMP_INTER_NONE && b->motion_mode == MM_WARP) ------------------ | Branch (740:17): [True: 2.67M, False: 140k] | Branch (740:52): [True: 134k, False: 2.53M] ------------------ 741| 134k| { 742| 134k| if (b->matrix[0] == INT16_MIN) { ------------------ | Branch (742:21): [True: 10.4k, False: 124k] ------------------ 743| 10.4k| t->warpmv.type = DAV1D_WM_TYPE_IDENTITY; 744| 124k| } else { 745| 124k| t->warpmv.type = DAV1D_WM_TYPE_AFFINE; 746| 124k| t->warpmv.matrix[2] = b->matrix[0] + 0x10000; 747| 124k| t->warpmv.matrix[3] = b->matrix[1]; 748| 124k| t->warpmv.matrix[4] = b->matrix[2]; 749| 124k| t->warpmv.matrix[5] = b->matrix[3] + 0x10000; 750| 124k| dav1d_set_affine_mv2d(bw4, bh4, b->mv2d, &t->warpmv, 751| 124k| t->bx, t->by); 752| 124k| dav1d_get_shear_params(&t->warpmv); 753| 124k|#define signabs(v) v < 0 ? '-' : ' ', abs(v) 754| 124k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 124k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 124k] | | ------------------ | | 35| 124k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 124k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 755| 0| printf("[ %c%x %c%x %c%x\n %c%x %c%x %c%x ]\n" 756| 0| "alpha=%c%x, beta=%c%x, gamma=%c%x, delta=%c%x, mv=y:%d,x:%d\n", 757| 0| signabs(t->warpmv.matrix[0]), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 758| 0| signabs(t->warpmv.matrix[1]), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 759| 0| signabs(t->warpmv.matrix[2]), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 760| 0| signabs(t->warpmv.matrix[3]), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 761| 0| signabs(t->warpmv.matrix[4]), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 762| 0| signabs(t->warpmv.matrix[5]), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 763| 0| signabs(t->warpmv.u.p.alpha), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 764| 0| signabs(t->warpmv.u.p.beta), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 765| 0| signabs(t->warpmv.u.p.gamma), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 766| 0| signabs(t->warpmv.u.p.delta), ------------------ | | 753| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (753:20): [True: 0, False: 0] | | ------------------ ------------------ 767| 0| b->mv2d.y, b->mv2d.x); 768| 124k|#undef signabs 769| 124k| } 770| 134k| } 771| 2.82M| if (f->bd_fn.recon_b_inter(t, bs, b)) return -1; ------------------ | Branch (771:17): [True: 0, False: 2.82M] ------------------ 772| | 773| 2.82M| const uint8_t *const filter = dav1d_filter_dir[b->filter2d]; 774| 2.82M| BlockContext *edge = t->a; 775| 8.48M| for (int i = 0, off = bx4; i < 2; i++, off = by4, edge = &t->l) { ------------------ | Branch (775:40): [True: 5.66M, False: 2.82M] ------------------ 776| 5.66M|#define set_ctx(rep_macro) \ 777| 5.66M| rep_macro(edge->filter[0], off, filter[0]); \ 778| 5.66M| rep_macro(edge->filter[1], off, filter[1]); \ 779| 5.66M| rep_macro(edge->intra, off, 0) 780| 5.66M| case_set(b_dim[2 + i]); ------------------ | | 70| 5.66M| switch (var) { \ | | 71| 398k| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 777| 398k| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 71| 398k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 398k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 778| 398k| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 71| 398k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 398k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 779| 398k| rep_macro(edge->intra, off, 0) | | | | ------------------ | | | | | | 71| 398k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 398k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (71:5): [True: 398k, False: 5.26M] | | ------------------ | | 72| 831k| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 777| 831k| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 72| 831k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 831k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 778| 831k| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 72| 831k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 831k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 779| 831k| rep_macro(edge->intra, off, 0) | | | | ------------------ | | | | | | 72| 831k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 831k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (72:5): [True: 831k, False: 4.82M] | | ------------------ | | 73| 679k| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 777| 679k| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 73| 679k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 679k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 778| 679k| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 73| 679k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 679k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 779| 679k| rep_macro(edge->intra, off, 0) | | | | ------------------ | | | | | | 73| 679k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 679k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (73:5): [True: 679k, False: 4.98M] | | ------------------ | | 74| 257k| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 777| 257k| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 74| 257k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 257k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 778| 257k| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 74| 257k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 257k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 779| 257k| rep_macro(edge->intra, off, 0) | | | | ------------------ | | | | | | 74| 257k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 257k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (74:5): [True: 257k, False: 5.40M] | | ------------------ | | 75| 2.11M| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 777| 2.11M| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 75| 2.11M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.11M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.11M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.11M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.11M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 778| 2.11M| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 75| 2.11M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.11M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.11M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.11M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.11M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 779| 2.11M| rep_macro(edge->intra, off, 0) | | | | ------------------ | | | | | | 75| 2.11M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.11M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.11M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.11M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.11M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (75:5): [True: 2.11M, False: 3.54M] | | ------------------ | | 76| 1.37M| case 5: set_ctx(set_ctx32); break; \ | | ------------------ | | | | 777| 1.37M| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 76| 1.37M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.37M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.37M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.37M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.37M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 778| 1.37M| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 76| 1.37M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.37M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.37M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.37M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.37M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 779| 1.37M| rep_macro(edge->intra, off, 0) | | | | ------------------ | | | | | | 76| 1.37M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.37M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.37M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.37M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.37M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (76:5): [True: 1.37M, False: 4.28M] | | ------------------ | | 77| 0| default: assert(0); \ | | ------------------ | | | Branch (77:5): [True: 0, False: 5.66M] | | ------------------ | | 78| 5.66M| } ------------------ | Branch (780:17): [Folded, False: 0] ------------------ 781| 5.66M|#undef set_ctx 782| 5.66M| } 783| | 784| 2.82M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { ------------------ | | 36| 2.82M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 2.81M, False: 17.9k] | | ------------------ ------------------ 785| 2.81M| refmvs_block *const r = &t->rt.r[(t->by & 31) + 5 + bh4 - 1][t->bx]; 786| 2.81M| const int ref1 = b->ref[0] + 1; 787| 2.81M| const union mv mv1 = b->mv[0]; 788| 44.7M| for (int x = 0; x < bw4; x++) { ------------------ | Branch (788:33): [True: 41.9M, False: 2.81M] ------------------ 789| 41.9M| r[x].ref.ref[0] = ref1; 790| 41.9M| r[x].mv.mv[0] = mv1; 791| 41.9M| r[x].bs = bs; 792| 41.9M| } 793| 2.81M| refmvs_block *const *rr = &t->rt.r[(t->by & 31) + 5]; 794| 42.2M| for (int y = 0; y < bh4 - 1; y++) { ------------------ | Branch (794:33): [True: 39.4M, False: 2.81M] ------------------ 795| 39.4M| rr[y][t->bx + bw4 - 1].ref.ref[0] = ref1; 796| 39.4M| rr[y][t->bx + bw4 - 1].mv.mv[0] = mv1; 797| 39.4M| rr[y][t->bx + bw4 - 1].bs = bs; 798| 39.4M| } 799| 2.81M| } 800| | 801| 2.82M| if (has_chroma) { ------------------ | Branch (801:17): [True: 924k, False: 1.90M] ------------------ 802| 924k| dav1d_memset_pow2[ulog2(cbw4)](&t->a->uvmode[cbx4], DC_PRED); 803| 924k| dav1d_memset_pow2[ulog2(cbh4)](&t->l.uvmode[cby4], DC_PRED); 804| 924k| } 805| 2.82M| } 806| 4.56M| return 0; 807| 4.56M| } 808| | 809| 11.8M| const int cw4 = (w4 + ss_hor) >> ss_hor, ch4 = (h4 + ss_ver) >> ss_ver; 810| | 811| 11.8M| b->bl = bl; 812| 11.8M| b->bp = bp; 813| 11.8M| b->bs = bs; 814| | 815| 11.8M| const Dav1dSegmentationData *seg = NULL; 816| | 817| | // segment_id (if seg_feature for skip/ref/gmv is enabled) 818| 11.8M| int seg_pred = 0; 819| 11.8M| if (f->frame_hdr->segmentation.enabled) { ------------------ | Branch (819:9): [True: 3.46M, False: 8.34M] ------------------ 820| 3.46M| if (!f->frame_hdr->segmentation.update_map) { ------------------ | Branch (820:13): [True: 906k, False: 2.55M] ------------------ 821| 906k| if (f->prev_segmap) { ------------------ | Branch (821:17): [True: 695k, False: 210k] ------------------ 822| 695k| unsigned seg_id = get_prev_frame_segid(f, t->by, t->bx, w4, h4, 823| 695k| f->prev_segmap, 824| 695k| f->b4_stride); 825| 695k| if (seg_id >= 8) return -1; ------------------ | Branch (825:21): [True: 0, False: 695k] ------------------ 826| 695k| b->seg_id = seg_id; 827| 695k| } else { 828| 210k| b->seg_id = 0; 829| 210k| } 830| 906k| seg = &f->frame_hdr->segmentation.seg_data.d[b->seg_id]; 831| 2.55M| } else if (f->frame_hdr->segmentation.seg_data.preskip) { ------------------ | Branch (831:20): [True: 2.15M, False: 408k] ------------------ 832| 2.15M| if (f->frame_hdr->segmentation.temporal && ------------------ | Branch (832:17): [True: 794k, False: 1.35M] ------------------ 833| 794k| (seg_pred = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 794k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (833:17): [True: 90.5k, False: 703k] ------------------ 834| 794k| ts->cdf.m.seg_pred[t->a->seg_pred[bx4] + 835| 794k| t->l.seg_pred[by4]]))) 836| 90.5k| { 837| | // temporal predicted seg_id 838| 90.5k| if (f->prev_segmap) { ------------------ | Branch (838:21): [True: 65.9k, False: 24.5k] ------------------ 839| 65.9k| unsigned seg_id = get_prev_frame_segid(f, t->by, t->bx, 840| 65.9k| w4, h4, 841| 65.9k| f->prev_segmap, 842| 65.9k| f->b4_stride); 843| 65.9k| if (seg_id >= 8) return -1; ------------------ | Branch (843:25): [True: 0, False: 65.9k] ------------------ 844| 65.9k| b->seg_id = seg_id; 845| 65.9k| } else { 846| 24.5k| b->seg_id = 0; 847| 24.5k| } 848| 2.06M| } else { 849| 2.06M| int seg_ctx; 850| 2.06M| const unsigned pred_seg_id = 851| 2.06M| get_cur_frame_segid(t->by, t->bx, have_top, have_left, 852| 2.06M| &seg_ctx, f->cur_segmap, f->b4_stride); 853| 2.06M| const unsigned diff = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 2.06M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 854| 2.06M| ts->cdf.m.seg_id[seg_ctx], 855| 2.06M| DAV1D_MAX_SEGMENTS - 1); ------------------ | | 43| 2.06M|#define DAV1D_MAX_SEGMENTS 8 ------------------ 856| 2.06M| const unsigned last_active_seg_id = 857| 2.06M| f->frame_hdr->segmentation.seg_data.last_active_segid; 858| 2.06M| b->seg_id = neg_deinterleave(diff, pred_seg_id, 859| 2.06M| last_active_seg_id + 1); 860| 2.06M| if (b->seg_id > last_active_seg_id) b->seg_id = 0; // error? ------------------ | Branch (860:21): [True: 106k, False: 1.95M] ------------------ 861| 2.06M| if (b->seg_id >= DAV1D_MAX_SEGMENTS) b->seg_id = 0; // error? ------------------ | | 43| 2.06M|#define DAV1D_MAX_SEGMENTS 8 ------------------ | Branch (861:21): [True: 0, False: 2.06M] ------------------ 862| 2.06M| } 863| | 864| 2.15M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 2.15M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 2.15M] | | ------------------ | | 35| 2.15M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 2.15M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 865| 0| printf("Post-segid[preskip;%d]: r=%d\n", 866| 0| b->seg_id, ts->msac.rng); 867| | 868| 2.15M| seg = &f->frame_hdr->segmentation.seg_data.d[b->seg_id]; 869| 2.15M| } 870| 8.34M| } else { 871| 8.34M| b->seg_id = 0; 872| 8.34M| } 873| | 874| | // skip_mode 875| 11.8M| if ((!seg || (!seg->globalmv && seg->ref == -1 && !seg->skip)) && ------------------ | Branch (875:10): [True: 8.75M, False: 3.05M] | Branch (875:19): [True: 654k, False: 2.40M] | Branch (875:37): [True: 356k, False: 297k] | Branch (875:55): [True: 293k, False: 62.8k] ------------------ 876| 9.01M| f->frame_hdr->skip_mode_enabled && imin(bw4, bh4) > 1) ------------------ | Branch (876:9): [True: 242k, False: 8.77M] | Branch (876:44): [True: 207k, False: 35.3k] ------------------ 877| 207k| { 878| 207k| const int smctx = t->a->skip_mode[bx4] + t->l.skip_mode[by4]; 879| 207k| b->skip_mode = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 207k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 880| 207k| ts->cdf.m.skip_mode[smctx]); 881| 207k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 207k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 207k] | | ------------------ | | 35| 207k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 207k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 882| 0| printf("Post-skipmode[%d]: r=%d\n", b->skip_mode, ts->msac.rng); 883| 11.6M| } else { 884| 11.6M| b->skip_mode = 0; 885| 11.6M| } 886| | 887| | // skip 888| 11.8M| if (b->skip_mode || (seg && seg->skip)) { ------------------ | Branch (888:9): [True: 61.4k, False: 11.7M] | Branch (888:26): [True: 3.05M, False: 8.69M] | Branch (888:33): [True: 2.54M, False: 507k] ------------------ 889| 2.57M| b->skip = 1; 890| 9.24M| } else { 891| 9.24M| const int sctx = t->a->skip[bx4] + t->l.skip[by4]; 892| 9.24M| b->skip = dav1d_msac_decode_bool_adapt(&ts->msac, ts->cdf.m.skip[sctx]); ------------------ | | 52| 9.24M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 893| 9.24M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 9.24M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 9.24M] | | ------------------ | | 35| 9.24M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 9.24M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 894| 0| printf("Post-skip[%d]: r=%d\n", b->skip, ts->msac.rng); 895| 9.24M| } 896| | 897| | // segment_id 898| 11.8M| if (f->frame_hdr->segmentation.enabled && ------------------ | Branch (898:9): [True: 3.46M, False: 8.34M] ------------------ 899| 3.46M| f->frame_hdr->segmentation.update_map && ------------------ | Branch (899:9): [True: 2.56M, False: 903k] ------------------ 900| 2.56M| !f->frame_hdr->segmentation.seg_data.preskip) ------------------ | Branch (900:9): [True: 408k, False: 2.15M] ------------------ 901| 408k| { 902| 408k| if (!b->skip && f->frame_hdr->segmentation.temporal && ------------------ | Branch (902:13): [True: 265k, False: 143k] | Branch (902:25): [True: 15.3k, False: 250k] ------------------ 903| 15.3k| (seg_pred = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 15.3k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (903:13): [True: 7.21k, False: 8.08k] ------------------ 904| 15.3k| ts->cdf.m.seg_pred[t->a->seg_pred[bx4] + 905| 15.3k| t->l.seg_pred[by4]]))) 906| 7.21k| { 907| | // temporal predicted seg_id 908| 7.21k| if (f->prev_segmap) { ------------------ | Branch (908:17): [True: 1.39k, False: 5.82k] ------------------ 909| 1.39k| unsigned seg_id = get_prev_frame_segid(f, t->by, t->bx, w4, h4, 910| 1.39k| f->prev_segmap, 911| 1.39k| f->b4_stride); 912| 1.39k| if (seg_id >= 8) return -1; ------------------ | Branch (912:21): [True: 0, False: 1.39k] ------------------ 913| 1.39k| b->seg_id = seg_id; 914| 5.82k| } else { 915| 5.82k| b->seg_id = 0; 916| 5.82k| } 917| 401k| } else { 918| 401k| int seg_ctx; 919| 401k| const unsigned pred_seg_id = 920| 401k| get_cur_frame_segid(t->by, t->bx, have_top, have_left, 921| 401k| &seg_ctx, f->cur_segmap, f->b4_stride); 922| 401k| if (b->skip) { ------------------ | Branch (922:17): [True: 145k, False: 255k] ------------------ 923| 145k| b->seg_id = pred_seg_id; 924| 255k| } else { 925| 255k| const unsigned diff = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 255k|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 926| 255k| ts->cdf.m.seg_id[seg_ctx], 927| 255k| DAV1D_MAX_SEGMENTS - 1); ------------------ | | 43| 255k|#define DAV1D_MAX_SEGMENTS 8 ------------------ 928| 255k| const unsigned last_active_seg_id = 929| 255k| f->frame_hdr->segmentation.seg_data.last_active_segid; 930| 255k| b->seg_id = neg_deinterleave(diff, pred_seg_id, 931| 255k| last_active_seg_id + 1); 932| 255k| if (b->seg_id > last_active_seg_id) b->seg_id = 0; // error? ------------------ | Branch (932:21): [True: 9.37k, False: 246k] ------------------ 933| 255k| } 934| 401k| if (b->seg_id >= DAV1D_MAX_SEGMENTS) b->seg_id = 0; // error? ------------------ | | 43| 401k|#define DAV1D_MAX_SEGMENTS 8 ------------------ | Branch (934:17): [True: 1.56k, False: 399k] ------------------ 935| 401k| } 936| | 937| 408k| seg = &f->frame_hdr->segmentation.seg_data.d[b->seg_id]; 938| | 939| 408k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 408k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 408k] | | ------------------ | | 35| 408k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 408k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 940| 0| printf("Post-segid[postskip;%d]: r=%d\n", 941| 0| b->seg_id, ts->msac.rng); 942| 408k| } 943| | 944| | // cdef index 945| 11.8M| if (!b->skip) { ------------------ | Branch (945:9): [True: 6.59M, False: 5.22M] ------------------ 946| 6.59M| const int idx = f->seq_hdr->sb128 ? ((t->bx & 16) >> 4) + ------------------ | Branch (946:25): [True: 5.06M, False: 1.52M] ------------------ 947| 5.06M| ((t->by & 16) >> 3) : 0; 948| 6.59M| if (t->cur_sb_cdef_idx_ptr[idx] == -1) { ------------------ | Branch (948:13): [True: 1.04M, False: 5.54M] ------------------ 949| 1.04M| const int v = dav1d_msac_decode_bools(&ts->msac, 950| 1.04M| f->frame_hdr->cdef.n_bits); 951| 1.04M| t->cur_sb_cdef_idx_ptr[idx] = v; 952| 1.04M| if (bw4 > 16) t->cur_sb_cdef_idx_ptr[idx + 1] = v; ------------------ | Branch (952:17): [True: 191k, False: 853k] ------------------ 953| 1.04M| if (bh4 > 16) t->cur_sb_cdef_idx_ptr[idx + 2] = v; ------------------ | Branch (953:17): [True: 192k, False: 852k] ------------------ 954| 1.04M| if (bw4 == 32 && bh4 == 32) t->cur_sb_cdef_idx_ptr[idx + 3] = v; ------------------ | Branch (954:17): [True: 191k, False: 853k] | Branch (954:30): [True: 178k, False: 12.8k] ------------------ 955| | 956| 1.04M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.04M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.04M] | | ------------------ | | 35| 1.04M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.04M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 957| 0| printf("Post-cdef_idx[%d]: r=%d\n", 958| 0| *t->cur_sb_cdef_idx_ptr, ts->msac.rng); 959| 1.04M| } 960| 6.59M| } 961| | 962| | // delta-q/lf 963| 11.8M| if (!((t->bx | t->by) & (31 >> !f->seq_hdr->sb128))) { ------------------ | Branch (963:9): [True: 3.11M, False: 8.70M] ------------------ 964| 3.11M| const int prev_qidx = ts->last_qidx; 965| 3.11M| const int have_delta_q = f->frame_hdr->delta.q.present && ------------------ | Branch (965:34): [True: 1.26M, False: 1.84M] ------------------ 966| 1.26M| (bs != (f->seq_hdr->sb128 ? BS_128x128 : BS_64x64) || !b->skip); ------------------ | Branch (966:14): [True: 148k, False: 1.11M] | Branch (966:21): [True: 118k, False: 1.14M] | Branch (966:67): [True: 23.9k, False: 1.09M] ------------------ 967| | 968| 3.11M| uint32_t prev_delta_lf = ts->last_delta_lf.u32; 969| | 970| 3.11M| if (have_delta_q) { ------------------ | Branch (970:13): [True: 172k, False: 2.93M] ------------------ 971| 172k| int delta_q = dav1d_msac_decode_symbol_adapt4(&ts->msac, ------------------ | | 47| 172k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 972| 172k| ts->cdf.m.delta_q, 3); 973| 172k| if (delta_q == 3) { ------------------ | Branch (973:17): [True: 30.7k, False: 141k] ------------------ 974| 30.7k| const int n_bits = 1 + dav1d_msac_decode_bools(&ts->msac, 3); 975| 30.7k| delta_q = dav1d_msac_decode_bools(&ts->msac, n_bits) + 976| 30.7k| 1 + (1 << n_bits); 977| 30.7k| } 978| 172k| if (delta_q) { ------------------ | Branch (978:17): [True: 51.1k, False: 121k] ------------------ 979| 51.1k| if (dav1d_msac_decode_bool_equi(&ts->msac)) delta_q = -delta_q; ------------------ | | 53| 51.1k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (979:21): [True: 38.7k, False: 12.4k] ------------------ 980| 51.1k| delta_q *= 1 << f->frame_hdr->delta.q.res_log2; 981| 51.1k| } 982| 172k| ts->last_qidx = iclip(ts->last_qidx + delta_q, 1, 255); 983| 172k| if (have_delta_q && DEBUG_BLOCK_INFO) ------------------ | | 34| 172k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 172k] | | ------------------ | | 35| 172k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 172k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ | Branch (983:17): [True: 172k, False: 18.4E] ------------------ 984| 0| printf("Post-delta_q[%d->%d]: r=%d\n", 985| 0| delta_q, ts->last_qidx, ts->msac.rng); 986| | 987| 172k| if (f->frame_hdr->delta.lf.present) { ------------------ | Branch (987:17): [True: 62.9k, False: 109k] ------------------ 988| 62.9k| const int n_lfs = f->frame_hdr->delta.lf.multi ? ------------------ | Branch (988:35): [True: 46.3k, False: 16.5k] ------------------ 989| 46.3k| f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 ? 4 : 2 : 1; ------------------ | Branch (989:21): [True: 39.1k, False: 7.22k] ------------------ 990| | 991| 250k| for (int i = 0; i < n_lfs; i++) { ------------------ | Branch (991:33): [True: 187k, False: 62.9k] ------------------ 992| 187k| int delta_lf = dav1d_msac_decode_symbol_adapt4(&ts->msac, ------------------ | | 47| 187k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 993| 187k| ts->cdf.m.delta_lf[i + f->frame_hdr->delta.lf.multi], 3); 994| 187k| if (delta_lf == 3) { ------------------ | Branch (994:25): [True: 48.8k, False: 138k] ------------------ 995| 48.8k| const int n_bits = 1 + dav1d_msac_decode_bools(&ts->msac, 3); 996| 48.8k| delta_lf = dav1d_msac_decode_bools(&ts->msac, n_bits) + 997| 48.8k| 1 + (1 << n_bits); 998| 48.8k| } 999| 187k| if (delta_lf) { ------------------ | Branch (999:25): [True: 67.9k, False: 119k] ------------------ 1000| 67.9k| if (dav1d_msac_decode_bool_equi(&ts->msac)) ------------------ | | 53| 67.9k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (1000:29): [True: 58.3k, False: 9.65k] ------------------ 1001| 58.3k| delta_lf = -delta_lf; 1002| 67.9k| delta_lf *= 1 << f->frame_hdr->delta.lf.res_log2; 1003| 67.9k| } 1004| 187k| ts->last_delta_lf.i8[i] = 1005| 187k| iclip(ts->last_delta_lf.i8[i] + delta_lf, -63, 63); 1006| 187k| if (have_delta_q && DEBUG_BLOCK_INFO) ------------------ | | 34| 187k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 187k] | | ------------------ | | 35| 187k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 187k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ | Branch (1006:25): [True: 187k, False: 18.4E] ------------------ 1007| 0| printf("Post-delta_lf[%d:%d]: r=%d\n", i, delta_lf, 1008| 0| ts->msac.rng); 1009| 187k| } 1010| 62.9k| } 1011| 172k| } 1012| 3.11M| if (ts->last_qidx == f->frame_hdr->quant.yac) { ------------------ | Branch (1012:13): [True: 2.97M, False: 130k] ------------------ 1013| | // assign frame-wide q values to this sb 1014| 2.97M| ts->dq = f->dq; 1015| 2.97M| } else if (ts->last_qidx != prev_qidx) { ------------------ | Branch (1015:20): [True: 20.5k, False: 109k] ------------------ 1016| | // find sb-specific quant parameters 1017| 20.5k| init_quant_tables(f->seq_hdr, f->frame_hdr, ts->last_qidx, ts->dqmem); 1018| 20.5k| ts->dq = ts->dqmem; 1019| 20.5k| } 1020| 3.11M| if (!ts->last_delta_lf.u32) { ------------------ | Branch (1020:13): [True: 3.06M, False: 49.3k] ------------------ 1021| | // assign frame-wide lf values to this sb 1022| 3.06M| ts->lflvl = f->lf.lvl; 1023| 3.06M| } else if (ts->last_delta_lf.u32 != prev_delta_lf) { ------------------ | Branch (1023:20): [True: 16.3k, False: 32.9k] ------------------ 1024| | // find sb-specific lf lvl parameters 1025| 16.3k| ts->lflvl = ts->lflvlmem; 1026| 16.3k| dav1d_calc_lf_values(ts->lflvlmem, f->frame_hdr, ts->last_delta_lf.i8); 1027| 16.3k| } 1028| 3.11M| } 1029| | 1030| 11.8M| if (b->skip_mode) { ------------------ | Branch (1030:9): [True: 25.3k, False: 11.7M] ------------------ 1031| 25.3k| b->intra = 0; 1032| 11.7M| } else if (IS_INTER_OR_SWITCH(f->frame_hdr)) { ------------------ | | 36| 11.7M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 4.35M, False: 7.43M] | | ------------------ ------------------ 1033| 4.35M| if (seg && (seg->ref >= 0 || seg->globalmv)) { ------------------ | Branch (1033:13): [True: 2.25M, False: 2.09M] | Branch (1033:21): [True: 263k, False: 1.99M] | Branch (1033:38): [True: 1.70M, False: 282k] ------------------ 1034| 1.97M| b->intra = !seg->ref; 1035| 2.37M| } else { 1036| 2.37M| const int ictx = get_intra_ctx(t->a, &t->l, by4, bx4, 1037| 2.37M| have_top, have_left); 1038| 2.37M| b->intra = !dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 2.37M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1039| 2.37M| ts->cdf.m.intra[ictx]); 1040| 2.37M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 2.37M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 2.37M] | | ------------------ | | 35| 2.37M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 2.37M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1041| 0| printf("Post-intra[%d]: r=%d\n", b->intra, ts->msac.rng); 1042| 2.37M| } 1043| 7.43M| } else if (f->frame_hdr->allow_intrabc) { ------------------ | Branch (1043:16): [True: 5.91M, False: 1.52M] ------------------ 1044| 5.91M| b->intra = !dav1d_msac_decode_bool_adapt(&ts->msac, ts->cdf.m.intrabc); ------------------ | | 52| 5.91M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1045| 5.91M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 5.91M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 5.91M] | | ------------------ | | 35| 5.91M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 5.91M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1046| 0| printf("Post-intrabcflag[%d]: r=%d\n", b->intra, ts->msac.rng); 1047| 5.91M| } else { 1048| 1.52M| b->intra = 1; 1049| 1.52M| } 1050| | 1051| | // intra/inter-specific stuff 1052| 11.8M| if (b->intra) { ------------------ | Branch (1052:9): [True: 6.55M, False: 5.25M] ------------------ 1053| 6.55M| uint16_t *const ymode_cdf = IS_INTER_OR_SWITCH(f->frame_hdr) ? ------------------ | | 36| 6.55M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 442k, False: 6.11M] | | ------------------ ------------------ 1054| 442k| ts->cdf.m.y_mode[dav1d_ymode_size_context[bs]] : 1055| 6.55M| ts->cdf.kfym[dav1d_intra_mode_context[t->a->mode[bx4]]] 1056| 6.11M| [dav1d_intra_mode_context[t->l.mode[by4]]]; 1057| 6.55M| b->y_mode = dav1d_msac_decode_symbol_adapt16(&ts->msac, ymode_cdf, ------------------ | | 57| 6.55M|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 1058| 6.55M| N_INTRA_PRED_MODES - 1); 1059| 6.55M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 6.55M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 6.55M] | | ------------------ | | 35| 6.55M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 6.55M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1060| 0| printf("Post-ymode[%d]: r=%d\n", b->y_mode, ts->msac.rng); 1061| | 1062| | // angle delta 1063| 6.55M| if (b_dim[2] + b_dim[3] >= 2 && b->y_mode >= VERT_PRED && ------------------ | Branch (1063:13): [True: 5.40M, False: 1.14M] | Branch (1063:41): [True: 2.75M, False: 2.65M] ------------------ 1064| 2.75M| b->y_mode <= VERT_LEFT_PRED) ------------------ | Branch (1064:13): [True: 1.50M, False: 1.24M] ------------------ 1065| 1.50M| { 1066| 1.50M| uint16_t *const acdf = ts->cdf.m.angle_delta[b->y_mode - VERT_PRED]; 1067| 1.50M| const int angle = dav1d_msac_decode_symbol_adapt8(&ts->msac, acdf, 6); ------------------ | | 48| 1.50M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 1068| 1.50M| b->y_angle = angle - 3; 1069| 5.04M| } else { 1070| 5.04M| b->y_angle = 0; 1071| 5.04M| } 1072| | 1073| 6.55M| if (has_chroma) { ------------------ | Branch (1073:13): [True: 5.31M, False: 1.24M] ------------------ 1074| 5.31M| const int cfl_allowed = f->frame_hdr->segmentation.lossless[b->seg_id] ? ------------------ | Branch (1074:37): [True: 127k, False: 5.18M] ------------------ 1075| 5.18M| cbw4 == 1 && cbh4 == 1 : !!(cfl_allowed_mask & (1 << bs)); ------------------ | Branch (1075:17): [True: 77.9k, False: 49.1k] | Branch (1075:30): [True: 74.0k, False: 3.95k] ------------------ 1076| 5.31M| uint16_t *const uvmode_cdf = ts->cdf.m.uv_mode[cfl_allowed][b->y_mode]; 1077| 5.31M| b->uv_mode = dav1d_msac_decode_symbol_adapt16(&ts->msac, uvmode_cdf, ------------------ | | 57| 5.31M|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 1078| 5.31M| N_UV_INTRA_PRED_MODES - 1 - !cfl_allowed); 1079| 5.31M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 5.31M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 5.31M] | | ------------------ | | 35| 5.31M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 5.31M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1080| 0| printf("Post-uvmode[%d]: r=%d\n", b->uv_mode, ts->msac.rng); 1081| | 1082| 5.31M| b->uv_angle = 0; 1083| 5.31M| if (b->uv_mode == CFL_PRED) { ------------------ | Branch (1083:17): [True: 1.09M, False: 4.22M] ------------------ 1084| 1.09M|#define SIGN(a) (!!(a) + ((a) > 0)) 1085| 1.09M| const int sign = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 1.09M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 1086| 1.09M| ts->cdf.m.cfl_sign, 7) + 1; 1087| 1.09M| const int sign_u = sign * 0x56 >> 8, sign_v = sign - sign_u * 3; 1088| 1.09M| assert(sign_u == sign / 3); ------------------ | Branch (1088:17): [True: 1.09M, False: 18.4E] ------------------ 1089| 1.09M| if (sign_u) { ------------------ | Branch (1089:21): [True: 1.04M, False: 49.3k] ------------------ 1090| 1.04M| const int ctx = (sign_u == 2) * 3 + sign_v; 1091| 1.04M| b->cfl_alpha[0] = dav1d_msac_decode_symbol_adapt16(&ts->msac, ------------------ | | 57| 1.04M|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 1092| 1.04M| ts->cdf.m.cfl_alpha[ctx], 15) + 1; 1093| 1.04M| if (sign_u == 1) b->cfl_alpha[0] = -b->cfl_alpha[0]; ------------------ | Branch (1093:25): [True: 616k, False: 426k] ------------------ 1094| 1.04M| } else { 1095| 49.3k| b->cfl_alpha[0] = 0; 1096| 49.3k| } 1097| 1.09M| if (sign_v) { ------------------ | Branch (1097:21): [True: 769k, False: 322k] ------------------ 1098| 769k| const int ctx = (sign_v == 2) * 3 + sign_u; 1099| 769k| b->cfl_alpha[1] = dav1d_msac_decode_symbol_adapt16(&ts->msac, ------------------ | | 57| 769k|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 1100| 769k| ts->cdf.m.cfl_alpha[ctx], 15) + 1; 1101| 769k| if (sign_v == 1) b->cfl_alpha[1] = -b->cfl_alpha[1]; ------------------ | Branch (1101:25): [True: 271k, False: 498k] ------------------ 1102| 769k| } else { 1103| 322k| b->cfl_alpha[1] = 0; 1104| 322k| } 1105| 1.09M|#undef SIGN 1106| 1.09M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.09M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.09M] | | ------------------ | | 35| 1.09M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.09M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1107| 0| printf("Post-uvalphas[%d/%d]: r=%d\n", 1108| 0| b->cfl_alpha[0], b->cfl_alpha[1], ts->msac.rng); 1109| 4.22M| } else if (b_dim[2] + b_dim[3] >= 2 && b->uv_mode >= VERT_PRED && ------------------ | Branch (1109:24): [True: 3.69M, False: 526k] | Branch (1109:52): [True: 2.11M, False: 1.57M] ------------------ 1110| 2.11M| b->uv_mode <= VERT_LEFT_PRED) ------------------ | Branch (1110:24): [True: 1.09M, False: 1.01M] ------------------ 1111| 1.09M| { 1112| 1.09M| uint16_t *const acdf = ts->cdf.m.angle_delta[b->uv_mode - VERT_PRED]; 1113| 1.09M| const int angle = dav1d_msac_decode_symbol_adapt8(&ts->msac, acdf, 6); ------------------ | | 48| 1.09M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 1114| 1.09M| b->uv_angle = angle - 3; 1115| 1.09M| } 1116| 5.31M| } 1117| | 1118| 6.55M| b->pal_sz[0] = b->pal_sz[1] = 0; 1119| 6.55M| if (f->frame_hdr->allow_screen_content_tools && ------------------ | Branch (1119:13): [True: 4.82M, False: 1.73M] ------------------ 1120| 4.82M| imax(bw4, bh4) <= 16 && bw4 + bh4 >= 4) ------------------ | Branch (1120:13): [True: 4.48M, False: 339k] | Branch (1120:37): [True: 3.77M, False: 710k] ------------------ 1121| 3.77M| { 1122| 3.77M| const int sz_ctx = b_dim[2] + b_dim[3] - 2; 1123| 3.77M| if (b->y_mode == DC_PRED) { ------------------ | Branch (1123:17): [True: 1.88M, False: 1.88M] ------------------ 1124| 1.88M| const int pal_ctx = (t->a->pal_sz[bx4] > 0) + (t->l.pal_sz[by4] > 0); 1125| 1.88M| const int use_y_pal = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 1.88M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1126| 1.88M| ts->cdf.m.pal_y[sz_ctx][pal_ctx]); 1127| 1.88M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.88M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.88M] | | ------------------ | | 35| 1.88M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.88M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1128| 0| printf("Post-y_pal[%d]: r=%d\n", use_y_pal, ts->msac.rng); 1129| 1.88M| if (use_y_pal) ------------------ | Branch (1129:21): [True: 165k, False: 1.71M] ------------------ 1130| 165k| f->bd_fn.read_pal_plane(t, b, 0, sz_ctx, bx4, by4); 1131| 1.88M| } 1132| | 1133| 3.77M| if (has_chroma && b->uv_mode == DC_PRED) { ------------------ | Branch (1133:17): [True: 3.33M, False: 436k] | Branch (1133:31): [True: 1.08M, False: 2.25M] ------------------ 1134| 1.08M| const int pal_ctx = b->pal_sz[0] > 0; 1135| 1.08M| const int use_uv_pal = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 1.08M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1136| 1.08M| ts->cdf.m.pal_uv[pal_ctx]); 1137| 1.08M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.08M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.08M] | | ------------------ | | 35| 1.08M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.08M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1138| 0| printf("Post-uv_pal[%d]: r=%d\n", use_uv_pal, ts->msac.rng); 1139| 1.08M| if (use_uv_pal) // see aomedia bug 2183 for why we use luma coordinates ------------------ | Branch (1139:21): [True: 33.9k, False: 1.04M] ------------------ 1140| 33.9k| f->bd_fn.read_pal_uv(t, b, sz_ctx, bx4, by4); 1141| 1.08M| } 1142| 3.77M| } 1143| | 1144| 6.55M| if (b->y_mode == DC_PRED && !b->pal_sz[0] && ------------------ | Branch (1144:13): [True: 3.12M, False: 3.42M] | Branch (1144:37): [True: 2.96M, False: 165k] ------------------ 1145| 2.96M| imax(b_dim[2], b_dim[3]) <= 3 && f->seq_hdr->filter_intra) ------------------ | Branch (1145:13): [True: 2.07M, False: 886k] | Branch (1145:46): [True: 1.40M, False: 668k] ------------------ 1146| 1.40M| { 1147| 1.40M| const int is_filter = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 1.40M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1148| 1.40M| ts->cdf.m.use_filter_intra[bs]); 1149| 1.40M| if (is_filter) { ------------------ | Branch (1149:17): [True: 913k, False: 495k] ------------------ 1150| 913k| b->y_mode = FILTER_PRED; 1151| 913k| b->y_angle = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 913k|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 1152| 913k| ts->cdf.m.filter_intra, 4); 1153| 913k| } 1154| 1.40M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.40M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.40M] | | ------------------ | | 35| 1.40M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.40M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1155| 0| printf("Post-filterintramode[%d/%d]: r=%d\n", 1156| 0| b->y_mode, b->y_angle, ts->msac.rng); 1157| 1.40M| } 1158| | 1159| 6.55M| if (b->pal_sz[0]) { ------------------ | Branch (1159:13): [True: 166k, False: 6.39M] ------------------ 1160| 166k| uint8_t *pal_idx; 1161| 166k| if (t->frame_thread.pass) { ------------------ | Branch (1161:17): [True: 165k, False: 6] ------------------ 1162| 165k| const int p = t->frame_thread.pass & 1; 1163| 165k| assert(ts->frame_thread[p].pal_idx); ------------------ | Branch (1163:17): [True: 166k, False: 18.4E] ------------------ 1164| 166k| pal_idx = ts->frame_thread[p].pal_idx; 1165| 166k| ts->frame_thread[p].pal_idx += bw4 * bh4 * 8; 1166| 166k| } else 1167| 6| pal_idx = t->scratch.pal_idx_y; 1168| 166k| read_pal_indices(t, pal_idx, b->pal_sz[0], 0, w4, h4, bw4, bh4); 1169| 166k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 166k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 166k] | | ------------------ | | 35| 166k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 166k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1170| 0| printf("Post-y-pal-indices: r=%d\n", ts->msac.rng); 1171| 166k| } 1172| | 1173| 6.55M| if (has_chroma && b->pal_sz[1]) { ------------------ | Branch (1173:13): [True: 5.32M, False: 1.23M] | Branch (1173:27): [True: 33.9k, False: 5.28M] ------------------ 1174| 33.9k| uint8_t *pal_idx; 1175| 33.9k| if (t->frame_thread.pass) { ------------------ | Branch (1175:17): [True: 33.9k, False: 0] ------------------ 1176| 33.9k| const int p = t->frame_thread.pass & 1; 1177| 33.9k| assert(ts->frame_thread[p].pal_idx); ------------------ | Branch (1177:17): [True: 33.9k, False: 18.4E] ------------------ 1178| 33.9k| pal_idx = ts->frame_thread[p].pal_idx; 1179| 33.9k| ts->frame_thread[p].pal_idx += cbw4 * cbh4 * 8; 1180| 33.9k| } else 1181| 0| pal_idx = t->scratch.pal_idx_uv; 1182| 33.9k| read_pal_indices(t, pal_idx, b->pal_sz[1], 1, cw4, ch4, cbw4, cbh4); 1183| 33.9k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 33.9k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 33.9k] | | ------------------ | | 35| 33.9k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 33.9k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1184| 0| printf("Post-uv-pal-indices: r=%d\n", ts->msac.rng); 1185| 33.9k| } 1186| | 1187| 6.55M| const TxfmInfo *t_dim; 1188| 6.55M| if (f->frame_hdr->segmentation.lossless[b->seg_id]) { ------------------ | Branch (1188:13): [True: 197k, False: 6.35M] ------------------ 1189| 197k| b->tx = b->uvtx = (int) TX_4X4; 1190| 197k| t_dim = &dav1d_txfm_dimensions[TX_4X4]; 1191| 6.35M| } else { 1192| 6.35M| b->tx = dav1d_max_txfm_size_for_bs[bs][0]; 1193| 6.35M| b->uvtx = dav1d_max_txfm_size_for_bs[bs][f->cur.p.layout]; 1194| 6.35M| t_dim = &dav1d_txfm_dimensions[b->tx]; 1195| 6.35M| if (f->frame_hdr->txfm_mode == DAV1D_TX_SWITCHABLE && t_dim->max > TX_4X4) { ------------------ | Branch (1195:17): [True: 1.03M, False: 5.32M] | Branch (1195:67): [True: 933k, False: 97.2k] ------------------ 1196| 933k| const int tctx = get_tx_ctx(t->a, &t->l, t_dim, by4, bx4); 1197| 933k| uint16_t *const tx_cdf = ts->cdf.m.txsz[t_dim->max - 1][tctx]; 1198| 933k| int depth = dav1d_msac_decode_symbol_adapt4(&ts->msac, tx_cdf, ------------------ | | 47| 933k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 1199| 933k| imin(t_dim->max, 2)); 1200| | 1201| 1.36M| while (depth--) { ------------------ | Branch (1201:24): [True: 432k, False: 933k] ------------------ 1202| 432k| b->tx = t_dim->sub; 1203| 432k| t_dim = &dav1d_txfm_dimensions[b->tx]; 1204| 432k| } 1205| 933k| } 1206| 6.35M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 6.35M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 6.35M] | | ------------------ | | 35| 6.35M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 6.35M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1207| 0| printf("Post-tx[%d]: r=%d\n", b->tx, ts->msac.rng); 1208| 6.35M| } 1209| | 1210| | // reconstruction 1211| 6.56M| if (t->frame_thread.pass == 1) { ------------------ | Branch (1211:13): [True: 6.56M, False: 18.4E] ------------------ 1212| 6.56M| f->bd_fn.read_coef_blocks(t, bs, b); 1213| 18.4E| } else { 1214| 18.4E| f->bd_fn.recon_b_intra(t, bs, intra_edge_flags, b); 1215| 18.4E| } 1216| | 1217| 6.55M| if (f->frame_hdr->loopfilter.level_y[0] || ------------------ | Branch (1217:13): [True: 1.30M, False: 5.25M] ------------------ 1218| 5.25M| f->frame_hdr->loopfilter.level_y[1]) ------------------ | Branch (1218:13): [True: 218k, False: 5.03M] ------------------ 1219| 1.53M| { 1220| 1.53M| dav1d_create_lf_mask_intra(t->lf_mask, f->lf.level, f->b4_stride, 1221| 1.53M| (const uint8_t (*)[8][2]) 1222| 1.53M| &ts->lflvl[b->seg_id][0][0][0], 1223| 1.53M| t->bx, t->by, f->w4, f->h4, bs, 1224| 1.53M| b->tx, b->uvtx, f->cur.p.layout, 1225| 1.53M| &t->a->tx_lpf_y[bx4], &t->l.tx_lpf_y[by4], 1226| 1.53M| has_chroma ? &t->a->tx_lpf_uv[cbx4] : NULL, ------------------ | Branch (1226:40): [True: 1.08M, False: 452k] ------------------ 1227| 1.53M| has_chroma ? &t->l.tx_lpf_uv[cby4] : NULL); ------------------ | Branch (1227:40): [True: 1.08M, False: 452k] ------------------ 1228| 1.53M| } 1229| | // update contexts 1230| 6.55M| const enum IntraPredMode y_mode_nofilt = 1231| 6.55M| b->y_mode == FILTER_PRED ? DC_PRED : b->y_mode; ------------------ | Branch (1231:13): [True: 913k, False: 5.64M] ------------------ 1232| 6.55M| BlockContext *edge = t->a; 1233| 19.6M| for (int i = 0, off = bx4; i < 2; i++, off = by4, edge = &t->l) { ------------------ | Branch (1233:36): [True: 13.1M, False: 6.55M] ------------------ 1234| 13.1M| int t_lsz = ((uint8_t *) &t_dim->lw)[i]; // lw then lh 1235| 13.1M|#define set_ctx(rep_macro) \ 1236| 13.1M| rep_macro(edge->tx_intra, off, t_lsz); \ 1237| 13.1M| rep_macro(edge->tx, off, t_lsz); \ 1238| 13.1M| rep_macro(edge->mode, off, y_mode_nofilt); \ 1239| 13.1M| rep_macro(edge->pal_sz, off, b->pal_sz[0]); \ 1240| 13.1M| rep_macro(edge->seg_pred, off, seg_pred); \ 1241| 13.1M| rep_macro(edge->skip_mode, off, 0); \ 1242| 13.1M| rep_macro(edge->intra, off, 1); \ 1243| 13.1M| rep_macro(edge->skip, off, b->skip); \ 1244| | /* see aomedia bug 2183 for why we use luma coordinates here */ \ 1245| 13.1M| rep_macro(t->pal_sz_uv[i], off, (has_chroma ? b->pal_sz[1] : 0)); \ 1246| 13.1M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { \ 1247| 13.1M| rep_macro(edge->comp_type, off, COMP_INTER_NONE); \ 1248| 13.1M| rep_macro(edge->ref[0], off, ((uint8_t) -1)); \ 1249| 13.1M| rep_macro(edge->ref[1], off, ((uint8_t) -1)); \ 1250| 13.1M| rep_macro(edge->filter[0], off, DAV1D_N_SWITCHABLE_FILTERS); \ 1251| 13.1M| rep_macro(edge->filter[1], off, DAV1D_N_SWITCHABLE_FILTERS); \ 1252| 13.1M| } 1253| 13.1M| case_set(b_dim[2 + i]); ------------------ | | 70| 13.1M| switch (var) { \ | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 1236| 2.31M| rep_macro(edge->tx_intra, off, t_lsz); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1237| 2.31M| rep_macro(edge->tx, off, t_lsz); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1238| 2.31M| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1239| 2.31M| rep_macro(edge->pal_sz, off, b->pal_sz[0]); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1240| 2.31M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1241| 2.31M| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1242| 2.31M| rep_macro(edge->intra, off, 1); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1243| 2.31M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 2.31M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1244| 2.31M| /* see aomedia bug 2183 for why we use luma coordinates here */ \ | | | | 1245| 2.31M| rep_macro(t->pal_sz_uv[i], off, (has_chroma ? b->pal_sz[1] : 0)); \ | | | | ------------------ | | | | | | 71| 2.31M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 4.63M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | | | ------------------ | | | | | | | | | Branch (56:43): [True: 1.54M, False: 772k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1246| 2.31M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { \ | | | | ------------------ | | | | | | 36| 2.31M| ((frame_header)->frame_type & 1) | | | | | | ------------------ | | | | | | | Branch (36:5): [True: 198k, False: 2.11M] | | | | | | ------------------ | | | | ------------------ | | | | 1247| 198k| rep_macro(edge->comp_type, off, COMP_INTER_NONE); \ | | | | ------------------ | | | | | | 71| 198k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 198k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1248| 198k| rep_macro(edge->ref[0], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 71| 198k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 198k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1249| 198k| rep_macro(edge->ref[1], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 71| 198k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 198k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1250| 198k| rep_macro(edge->filter[0], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 71| 198k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 198k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1251| 198k| rep_macro(edge->filter[1], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 71| 198k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 198k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1252| 198k| } | | ------------------ | | | Branch (71:5): [True: 2.31M, False: 10.8M] | | ------------------ | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 1236| 3.81M| rep_macro(edge->tx_intra, off, t_lsz); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1237| 3.81M| rep_macro(edge->tx, off, t_lsz); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1238| 3.81M| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1239| 3.81M| rep_macro(edge->pal_sz, off, b->pal_sz[0]); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1240| 3.81M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1241| 3.81M| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1242| 3.81M| rep_macro(edge->intra, off, 1); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1243| 3.81M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 3.81M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1244| 3.81M| /* see aomedia bug 2183 for why we use luma coordinates here */ \ | | | | 1245| 3.81M| rep_macro(t->pal_sz_uv[i], off, (has_chroma ? b->pal_sz[1] : 0)); \ | | | | ------------------ | | | | | | 72| 3.81M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 7.63M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | | | ------------------ | | | | | | | | | Branch (58:45): [True: 3.39M, False: 426k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1246| 3.81M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { \ | | | | ------------------ | | | | | | 36| 3.81M| ((frame_header)->frame_type & 1) | | | | | | ------------------ | | | | | | | Branch (36:5): [True: 312k, False: 3.50M] | | | | | | ------------------ | | | | ------------------ | | | | 1247| 312k| rep_macro(edge->comp_type, off, COMP_INTER_NONE); \ | | | | ------------------ | | | | | | 72| 312k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 312k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1248| 312k| rep_macro(edge->ref[0], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 72| 312k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 312k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1249| 312k| rep_macro(edge->ref[1], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 72| 312k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 312k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1250| 312k| rep_macro(edge->filter[0], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 72| 312k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 312k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1251| 312k| rep_macro(edge->filter[1], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 72| 312k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 312k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1252| 312k| } | | ------------------ | | | Branch (72:5): [True: 3.81M, False: 9.31M] | | ------------------ | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 1236| 3.20M| rep_macro(edge->tx_intra, off, t_lsz); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1237| 3.20M| rep_macro(edge->tx, off, t_lsz); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1238| 3.20M| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1239| 3.20M| rep_macro(edge->pal_sz, off, b->pal_sz[0]); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1240| 3.20M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1241| 3.20M| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1242| 3.20M| rep_macro(edge->intra, off, 1); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1243| 3.20M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 3.20M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1244| 3.20M| /* see aomedia bug 2183 for why we use luma coordinates here */ \ | | | | 1245| 3.20M| rep_macro(t->pal_sz_uv[i], off, (has_chroma ? b->pal_sz[1] : 0)); \ | | | | ------------------ | | | | | | 73| 3.20M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 6.40M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | | | ------------------ | | | | | | | | | Branch (60:45): [True: 2.84M, False: 355k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1246| 3.20M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { \ | | | | ------------------ | | | | | | 36| 3.20M| ((frame_header)->frame_type & 1) | | | | | | ------------------ | | | | | | | Branch (36:5): [True: 231k, False: 2.97M] | | | | | | ------------------ | | | | ------------------ | | | | 1247| 231k| rep_macro(edge->comp_type, off, COMP_INTER_NONE); \ | | | | ------------------ | | | | | | 73| 231k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 231k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1248| 231k| rep_macro(edge->ref[0], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 73| 231k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 231k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1249| 231k| rep_macro(edge->ref[1], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 73| 231k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 231k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1250| 231k| rep_macro(edge->filter[0], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 73| 231k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 231k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1251| 231k| rep_macro(edge->filter[1], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 73| 231k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 231k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1252| 231k| } | | ------------------ | | | Branch (73:5): [True: 3.20M, False: 9.92M] | | ------------------ | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 1236| 1.78M| rep_macro(edge->tx_intra, off, t_lsz); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1237| 1.78M| rep_macro(edge->tx, off, t_lsz); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1238| 1.78M| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1239| 1.78M| rep_macro(edge->pal_sz, off, b->pal_sz[0]); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1240| 1.78M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1241| 1.78M| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1242| 1.78M| rep_macro(edge->intra, off, 1); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1243| 1.78M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.78M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1244| 1.78M| /* see aomedia bug 2183 for why we use luma coordinates here */ \ | | | | 1245| 1.78M| rep_macro(t->pal_sz_uv[i], off, (has_chroma ? b->pal_sz[1] : 0)); \ | | | | ------------------ | | | | | | 74| 1.78M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 3.57M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | | | ------------------ | | | | | | | | | Branch (62:45): [True: 1.65M, False: 134k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1246| 1.78M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { \ | | | | ------------------ | | | | | | 36| 1.78M| ((frame_header)->frame_type & 1) | | | | | | ------------------ | | | | | | | Branch (36:5): [True: 73.3k, False: 1.71M] | | | | | | ------------------ | | | | ------------------ | | | | 1247| 73.3k| rep_macro(edge->comp_type, off, COMP_INTER_NONE); \ | | | | ------------------ | | | | | | 74| 73.3k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 73.3k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1248| 73.3k| rep_macro(edge->ref[0], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 74| 73.3k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 73.3k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1249| 73.3k| rep_macro(edge->ref[1], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 74| 73.3k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 73.3k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1250| 73.3k| rep_macro(edge->filter[0], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 74| 73.3k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 73.3k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1251| 73.3k| rep_macro(edge->filter[1], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 74| 73.3k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 73.3k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1252| 73.3k| } | | ------------------ | | | Branch (74:5): [True: 1.78M, False: 11.3M] | | ------------------ | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 1236| 1.29M| rep_macro(edge->tx_intra, off, t_lsz); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1237| 1.29M| rep_macro(edge->tx, off, t_lsz); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1238| 1.29M| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1239| 1.29M| rep_macro(edge->pal_sz, off, b->pal_sz[0]); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1240| 1.29M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1241| 1.29M| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1242| 1.29M| rep_macro(edge->intra, off, 1); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1243| 1.29M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 1.29M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1244| 1.29M| /* see aomedia bug 2183 for why we use luma coordinates here */ \ | | | | 1245| 1.29M| rep_macro(t->pal_sz_uv[i], off, (has_chroma ? b->pal_sz[1] : 0)); \ | | | | ------------------ | | | | | | 75| 1.29M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 1.29M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.58M| memset(&(var)[off], val, 16); \ | | | | | | | | ------------------ | | | | | | | | | Branch (64:29): [True: 684k, False: 607k] | | | | | | | | ------------------ | | | | | | | | 65| 1.29M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 1.29M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1246| 1.29M| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { \ | | | | ------------------ | | | | | | 36| 1.29M| ((frame_header)->frame_type & 1) | | | | | | ------------------ | | | | | | | Branch (36:5): [True: 38.8k, False: 1.25M] | | | | | | ------------------ | | | | ------------------ | | | | 1247| 38.8k| rep_macro(edge->comp_type, off, COMP_INTER_NONE); \ | | | | ------------------ | | | | | | 75| 38.8k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 38.8k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 38.8k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 38.8k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 38.8k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1248| 38.8k| rep_macro(edge->ref[0], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 75| 38.8k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 38.8k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 38.8k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 38.8k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 38.8k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1249| 38.8k| rep_macro(edge->ref[1], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 75| 38.8k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 38.8k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 38.8k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 38.8k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 38.8k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1250| 38.8k| rep_macro(edge->filter[0], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 75| 38.8k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 38.8k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 38.8k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 38.8k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 38.8k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1251| 38.8k| rep_macro(edge->filter[1], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 75| 38.8k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 38.8k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 38.8k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 38.8k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 38.8k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1252| 38.8k| } | | ------------------ | | | Branch (75:5): [True: 1.29M, False: 11.8M] | | ------------------ | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | ------------------ | | | | 1236| 746k| rep_macro(edge->tx_intra, off, t_lsz); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1237| 746k| rep_macro(edge->tx, off, t_lsz); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1238| 746k| rep_macro(edge->mode, off, y_mode_nofilt); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1239| 746k| rep_macro(edge->pal_sz, off, b->pal_sz[0]); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1240| 746k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1241| 746k| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1242| 746k| rep_macro(edge->intra, off, 1); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1243| 746k| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 746k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1244| 746k| /* see aomedia bug 2183 for why we use luma coordinates here */ \ | | | | 1245| 746k| rep_macro(t->pal_sz_uv[i], off, (has_chroma ? b->pal_sz[1] : 0)); \ | | | | ------------------ | | | | | | 76| 746k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 746k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.49M| memset(&(var)[off], val, 32); \ | | | | | | | | ------------------ | | | | | | | | | Branch (67:29): [True: 544k, False: 202k] | | | | | | | | ------------------ | | | | | | | | 68| 746k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 746k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1246| 746k| if (IS_INTER_OR_SWITCH(f->frame_hdr)) { \ | | | | ------------------ | | | | | | 36| 746k| ((frame_header)->frame_type & 1) | | | | | | ------------------ | | | | | | | Branch (36:5): [True: 30.4k, False: 716k] | | | | | | ------------------ | | | | ------------------ | | | | 1247| 30.4k| rep_macro(edge->comp_type, off, COMP_INTER_NONE); \ | | | | ------------------ | | | | | | 76| 30.4k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 30.4k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 30.4k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 30.4k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 30.4k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1248| 30.4k| rep_macro(edge->ref[0], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 76| 30.4k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 30.4k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 30.4k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 30.4k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 30.4k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1249| 30.4k| rep_macro(edge->ref[1], off, ((uint8_t) -1)); \ | | | | ------------------ | | | | | | 76| 30.4k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 30.4k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 30.4k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 30.4k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 30.4k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1250| 30.4k| rep_macro(edge->filter[0], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 76| 30.4k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 30.4k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 30.4k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 30.4k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 30.4k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1251| 30.4k| rep_macro(edge->filter[1], off, DAV1D_N_SWITCHABLE_FILTERS); \ | | | | ------------------ | | | | | | 76| 30.4k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 30.4k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 30.4k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 30.4k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 30.4k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1252| 30.4k| } | | ------------------ | | | Branch (76:5): [True: 746k, False: 12.3M] | | ------------------ | | 77| 0| default: assert(0); \ | | ------------------ | | | Branch (77:5): [True: 0, False: 13.1M] | | ------------------ | | 78| 13.1M| } ------------------ | Branch (1253:13): [Folded, False: 0] ------------------ 1254| 13.1M|#undef set_ctx 1255| 13.1M| } 1256| 6.55M| if (b->pal_sz[0]) ------------------ | Branch (1256:13): [True: 165k, False: 6.38M] ------------------ 1257| 165k| f->bd_fn.copy_pal_block_y(t, bx4, by4, bw4, bh4); 1258| 6.55M| if (has_chroma) { ------------------ | Branch (1258:13): [True: 5.32M, False: 1.23M] ------------------ 1259| 5.32M| uint8_t uv_mode = b->uv_mode; 1260| 5.32M| dav1d_memset_pow2[ulog2(cbw4)](&t->a->uvmode[cbx4], uv_mode); 1261| 5.32M| dav1d_memset_pow2[ulog2(cbh4)](&t->l.uvmode[cby4], uv_mode); 1262| 5.32M| if (b->pal_sz[1]) ------------------ | Branch (1262:17): [True: 33.9k, False: 5.28M] ------------------ 1263| 33.9k| f->bd_fn.copy_pal_block_uv(t, bx4, by4, bw4, bh4); 1264| 5.32M| } 1265| 6.55M| if (IS_INTER_OR_SWITCH(f->frame_hdr) || f->frame_hdr->allow_intrabc) ------------------ | | 36| 13.1M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 419k, False: 6.13M] | | ------------------ ------------------ | Branch (1265:49): [True: 4.61M, False: 1.52M] ------------------ 1266| 5.05M| splat_intraref(f->c, t, bs, bw4, bh4); 1267| 6.55M| } else if (IS_KEY_OR_INTRA(f->frame_hdr)) { ------------------ | | 43| 5.25M| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 5.25M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (43:5): [True: 1.33M, False: 3.92M] | | ------------------ ------------------ 1268| | // intra block copy 1269| 1.33M| refmvs_candidate mvstack[8]; 1270| 1.33M| int n_mvs, ctx; 1271| 1.33M| dav1d_refmvs_find(&t->rt, mvstack, &n_mvs, &ctx, 1272| 1.33M| (union refmvs_refpair) { .ref = { 0, -1 }}, 1273| 1.33M| bs, intra_edge_flags, t->by, t->bx); 1274| | 1275| 1.33M| if (mvstack[0].mv.mv[0].n) ------------------ | Branch (1275:13): [True: 1.09M, False: 238k] ------------------ 1276| 1.09M| b->mv[0] = mvstack[0].mv.mv[0]; 1277| 238k| else if (mvstack[1].mv.mv[0].n) ------------------ | Branch (1277:18): [True: 0, False: 238k] ------------------ 1278| 0| b->mv[0] = mvstack[1].mv.mv[0]; 1279| 238k| else { 1280| 238k| if (t->by - (16 << f->seq_hdr->sb128) < ts->tiling.row_start) { ------------------ | Branch (1280:17): [True: 236k, False: 1.30k] ------------------ 1281| 236k| b->mv[0].y = 0; 1282| 236k| b->mv[0].x = -(512 << f->seq_hdr->sb128) - 2048; 1283| 236k| } else { 1284| 1.30k| b->mv[0].y = -(512 << f->seq_hdr->sb128); 1285| 1.30k| b->mv[0].x = 0; 1286| 1.30k| } 1287| 238k| } 1288| | 1289| 1.33M| const union mv ref = b->mv[0]; 1290| 1.33M| read_mv_residual(ts, &b->mv[0], -1); 1291| | 1292| | // clip intrabc motion vector to decoded parts of current tile 1293| 1.33M| int border_left = ts->tiling.col_start * 4; 1294| 1.33M| int border_top = ts->tiling.row_start * 4; 1295| 1.33M| if (has_chroma) { ------------------ | Branch (1295:13): [True: 1.24M, False: 89.4k] ------------------ 1296| 1.24M| if (bw4 < 2 && ss_hor) ------------------ | Branch (1296:17): [True: 284k, False: 961k] | Branch (1296:29): [True: 22.2k, False: 262k] ------------------ 1297| 22.2k| border_left += 4; 1298| 1.24M| if (bh4 < 2 && ss_ver) ------------------ | Branch (1298:17): [True: 298k, False: 948k] | Branch (1298:29): [True: 35.4k, False: 262k] ------------------ 1299| 35.4k| border_top += 4; 1300| 1.24M| } 1301| 1.33M| int src_left = t->bx * 4 + (b->mv[0].x >> 3); 1302| 1.33M| int src_top = t->by * 4 + (b->mv[0].y >> 3); 1303| 1.33M| int src_right = src_left + bw4 * 4; 1304| 1.33M| int src_bottom = src_top + bh4 * 4; 1305| 1.33M| const int border_right = ((ts->tiling.col_end + (bw4 - 1)) & ~(bw4 - 1)) * 4; 1306| | 1307| | // check against left or right tile boundary and adjust if necessary 1308| 1.33M| if (src_left < border_left) { ------------------ | Branch (1308:13): [True: 158k, False: 1.17M] ------------------ 1309| 158k| src_right += border_left - src_left; 1310| 158k| src_left += border_left - src_left; 1311| 1.17M| } else if (src_right > border_right) { ------------------ | Branch (1311:20): [True: 289k, False: 888k] ------------------ 1312| 289k| src_left -= src_right - border_right; 1313| 289k| src_right -= src_right - border_right; 1314| 289k| } 1315| | // check against top tile boundary and adjust if necessary 1316| 1.33M| if (src_top < border_top) { ------------------ | Branch (1316:13): [True: 682k, False: 653k] ------------------ 1317| 682k| src_bottom += border_top - src_top; 1318| 682k| src_top += border_top - src_top; 1319| 682k| } 1320| | 1321| 1.33M| const int sbx = (t->bx >> (4 + f->seq_hdr->sb128)) << (6 + f->seq_hdr->sb128); 1322| 1.33M| const int sby = (t->by >> (4 + f->seq_hdr->sb128)) << (6 + f->seq_hdr->sb128); 1323| 1.33M| const int sb_size = 1 << (6 + f->seq_hdr->sb128); 1324| | // check for overlap with current superblock 1325| 1.33M| if (src_bottom > sby && src_right > sbx) { ------------------ | Branch (1325:13): [True: 1.31M, False: 23.9k] | Branch (1325:33): [True: 302k, False: 1.01M] ------------------ 1326| 302k| if (src_top - border_top >= src_bottom - sby) { ------------------ | Branch (1326:17): [True: 1.56k, False: 300k] ------------------ 1327| | // if possible move src up into the previous suberblock row 1328| 1.56k| src_top -= src_bottom - sby; 1329| 1.56k| src_bottom -= src_bottom - sby; 1330| 300k| } else if (src_left - border_left >= src_right - sbx) { ------------------ | Branch (1330:24): [True: 288k, False: 12.3k] ------------------ 1331| | // if possible move src left into the previous suberblock 1332| 288k| src_left -= src_right - sbx; 1333| 288k| src_right -= src_right - sbx; 1334| 288k| } 1335| 302k| } 1336| | // move src up if it is below current superblock row 1337| 1.33M| if (src_bottom > sby + sb_size) { ------------------ | Branch (1337:13): [True: 19.4k, False: 1.31M] ------------------ 1338| 19.4k| src_top -= src_bottom - (sby + sb_size); 1339| 19.4k| src_bottom -= src_bottom - (sby + sb_size); 1340| 19.4k| } 1341| | // error out if mv still overlaps with the current superblock 1342| 1.33M| if (src_bottom > sby && src_right > sbx) ------------------ | Branch (1342:13): [True: 1.31M, False: 25.5k] | Branch (1342:33): [True: 12.3k, False: 1.29M] ------------------ 1343| 12.3k| return -1; 1344| | 1345| 1.32M| b->mv[0].x = (src_left - t->bx * 4) * 8; 1346| 1.32M| b->mv[0].y = (src_top - t->by * 4) * 8; 1347| | 1348| 1.32M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.32M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.32M] | | ------------------ | | 35| 1.32M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.32M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1349| 0| printf("Post-dmv[%d/%d,ref=%d/%d|%d/%d]: r=%d\n", 1350| 0| b->mv[0].y, b->mv[0].x, ref.y, ref.x, 1351| 0| mvstack[0].mv.mv[0].y, mvstack[0].mv.mv[0].x, ts->msac.rng); 1352| 1.32M| read_vartx_tree(t, b, bs, bx4, by4); 1353| | 1354| | // reconstruction 1355| 1.32M| if (t->frame_thread.pass == 1) { ------------------ | Branch (1355:13): [True: 1.32M, False: 18.4E] ------------------ 1356| 1.32M| f->bd_fn.read_coef_blocks(t, bs, b); 1357| 1.32M| b->filter2d = FILTER_2D_BILINEAR; 1358| 18.4E| } else { 1359| 18.4E| if (f->bd_fn.recon_b_inter(t, bs, b)) return -1; ------------------ | Branch (1359:17): [True: 0, False: 18.4E] ------------------ 1360| 18.4E| } 1361| | 1362| 1.32M| splat_intrabc_mv(f->c, t, bs, b, bw4, bh4); 1363| 1.32M| BlockContext *edge = t->a; 1364| 3.97M| for (int i = 0, off = bx4; i < 2; i++, off = by4, edge = &t->l) { ------------------ | Branch (1364:36): [True: 2.64M, False: 1.32M] ------------------ 1365| 2.64M|#define set_ctx(rep_macro) \ 1366| 2.64M| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ 1367| 2.64M| rep_macro(edge->mode, off, DC_PRED); \ 1368| 2.64M| rep_macro(edge->pal_sz, off, 0); \ 1369| | /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ 1370| 2.64M| rep_macro(t->pal_sz_uv[i], off, 0); \ 1371| 2.64M| rep_macro(edge->seg_pred, off, seg_pred); \ 1372| 2.64M| rep_macro(edge->skip_mode, off, 0); \ 1373| 2.64M| rep_macro(edge->intra, off, 0); \ 1374| 2.64M| rep_macro(edge->skip, off, b->skip) 1375| 2.64M| case_set(b_dim[2 + i]); ------------------ | | 70| 2.64M| switch (var) { \ | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 1366| 689k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1367| 689k| rep_macro(edge->mode, off, DC_PRED); \ | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1368| 689k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1369| 689k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1370| 689k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1371| 689k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1372| 689k| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1373| 689k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1374| 689k| rep_macro(edge->skip, off, b->skip) | | | | ------------------ | | | | | | 71| 689k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 689k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (71:5): [True: 689k, False: 1.95M] | | ------------------ | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 1366| 562k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1367| 562k| rep_macro(edge->mode, off, DC_PRED); \ | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1368| 562k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1369| 562k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1370| 562k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1371| 562k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1372| 562k| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1373| 562k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1374| 562k| rep_macro(edge->skip, off, b->skip) | | | | ------------------ | | | | | | 72| 562k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 562k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (72:5): [True: 562k, False: 2.08M] | | ------------------ | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 1366| 517k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1367| 517k| rep_macro(edge->mode, off, DC_PRED); \ | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1368| 517k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1369| 517k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1370| 517k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1371| 517k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1372| 517k| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1373| 517k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1374| 517k| rep_macro(edge->skip, off, b->skip) | | | | ------------------ | | | | | | 73| 517k| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 517k| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (73:5): [True: 517k, False: 2.12M] | | ------------------ | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 1366| 225k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1367| 225k| rep_macro(edge->mode, off, DC_PRED); \ | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1368| 225k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1369| 225k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1370| 225k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1371| 225k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1372| 225k| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1373| 225k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1374| 225k| rep_macro(edge->skip, off, b->skip) | | | | ------------------ | | | | | | 74| 225k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 225k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (74:5): [True: 225k, False: 2.42M] | | ------------------ | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 1366| 555k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1367| 555k| rep_macro(edge->mode, off, DC_PRED); \ | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1368| 555k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1369| 555k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1370| 555k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1371| 555k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1372| 555k| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1373| 555k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1374| 555k| rep_macro(edge->skip, off, b->skip) | | | | ------------------ | | | | | | 75| 555k| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 555k|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 555k| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 555k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 555k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (75:5): [True: 555k, False: 2.09M] | | ------------------ | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | ------------------ | | | | 1366| 99.6k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1367| 99.6k| rep_macro(edge->mode, off, DC_PRED); \ | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1368| 99.6k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1369| 99.6k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1370| 99.6k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1371| 99.6k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1372| 99.6k| rep_macro(edge->skip_mode, off, 0); \ | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1373| 99.6k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1374| 99.6k| rep_macro(edge->skip, off, b->skip) | | | | ------------------ | | | | | | 76| 99.6k| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 99.6k|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 99.6k| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 99.6k| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 99.6k] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (76:5): [True: 99.6k, False: 2.54M] | | ------------------ | | 77| 0| default: assert(0); \ | | ------------------ | | | Branch (77:5): [True: 0, False: 2.64M] | | ------------------ | | 78| 2.64M| } ------------------ | Branch (1375:13): [Folded, False: 0] ------------------ 1376| 2.64M|#undef set_ctx 1377| 2.64M| } 1378| 1.32M| if (has_chroma) { ------------------ | Branch (1378:13): [True: 1.23M, False: 86.5k] ------------------ 1379| 1.23M| dav1d_memset_pow2[ulog2(cbw4)](&t->a->uvmode[cbx4], DC_PRED); 1380| 1.23M| dav1d_memset_pow2[ulog2(cbh4)](&t->l.uvmode[cby4], DC_PRED); 1381| 1.23M| } 1382| 3.92M| } else { 1383| | // inter-specific mode/mv coding 1384| 3.92M| int is_comp, has_subpel_filter; 1385| | 1386| 3.92M| if (b->skip_mode) { ------------------ | Branch (1386:13): [True: 25.3k, False: 3.89M] ------------------ 1387| 25.3k| is_comp = 1; 1388| 3.89M| } else if ((!seg || (seg->ref == -1 && !seg->globalmv && !seg->skip)) && ------------------ | Branch (1388:21): [True: 1.71M, False: 2.18M] | Branch (1388:30): [True: 1.93M, False: 254k] | Branch (1388:48): [True: 220k, False: 1.70M] | Branch (1388:66): [True: 196k, False: 23.9k] ------------------ 1389| 1.92M| f->frame_hdr->switchable_comp_refs && imin(bw4, bh4) > 1) ------------------ | Branch (1389:20): [True: 1.05M, False: 870k] | Branch (1389:58): [True: 786k, False: 265k] ------------------ 1390| 786k| { 1391| 786k| const int ctx = get_comp_ctx(t->a, &t->l, by4, bx4, 1392| 786k| have_top, have_left); 1393| 786k| is_comp = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 786k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1394| 786k| ts->cdf.m.comp[ctx]); 1395| 786k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 786k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 786k] | | ------------------ | | 35| 786k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 786k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1396| 0| printf("Post-compflag[%d]: r=%d\n", is_comp, ts->msac.rng); 1397| 3.10M| } else { 1398| 3.10M| is_comp = 0; 1399| 3.10M| } 1400| | 1401| 3.92M| if (b->skip_mode) { ------------------ | Branch (1401:13): [True: 25.3k, False: 3.89M] ------------------ 1402| 25.3k| b->ref[0] = f->frame_hdr->skip_mode_refs[0]; 1403| 25.3k| b->ref[1] = f->frame_hdr->skip_mode_refs[1]; 1404| 25.3k| b->comp_type = COMP_INTER_AVG; 1405| 25.3k| b->inter_mode = NEARESTMV_NEARESTMV; 1406| 25.3k| b->drl_idx = NEAREST_DRL; 1407| 25.3k| has_subpel_filter = 0; 1408| | 1409| 25.3k| refmvs_candidate mvstack[8]; 1410| 25.3k| int n_mvs, ctx; 1411| 25.3k| dav1d_refmvs_find(&t->rt, mvstack, &n_mvs, &ctx, 1412| 25.3k| (union refmvs_refpair) { .ref = { 1413| 25.3k| b->ref[0] + 1, b->ref[1] + 1 }}, 1414| 25.3k| bs, intra_edge_flags, t->by, t->bx); 1415| | 1416| 25.3k| b->mv[0] = mvstack[0].mv.mv[0]; 1417| 25.3k| b->mv[1] = mvstack[0].mv.mv[1]; 1418| 25.3k| fix_mv_precision(f->frame_hdr, &b->mv[0]); 1419| 25.3k| fix_mv_precision(f->frame_hdr, &b->mv[1]); 1420| 25.3k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 25.3k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 25.3k] | | ------------------ | | 35| 25.3k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 25.3k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1421| 0| printf("Post-skipmodeblock[mv=1:y=%d,x=%d,2:y=%d,x=%d,refs=%d+%d\n", 1422| 0| b->mv[0].y, b->mv[0].x, b->mv[1].y, b->mv[1].x, 1423| 0| b->ref[0], b->ref[1]); 1424| 3.89M| } else if (is_comp) { ------------------ | Branch (1424:20): [True: 372k, False: 3.52M] ------------------ 1425| 372k| const int dir_ctx = get_comp_dir_ctx(t->a, &t->l, by4, bx4, 1426| 372k| have_top, have_left); 1427| 372k| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 372k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1427:17): [True: 302k, False: 69.7k] ------------------ 1428| 372k| ts->cdf.m.comp_dir[dir_ctx])) 1429| 302k| { 1430| | // bidir - first reference (fw) 1431| 302k| const int ctx1 = av1_get_fwd_ref_ctx(t->a, &t->l, by4, bx4, 1432| 302k| have_top, have_left); 1433| 302k| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 302k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1433:21): [True: 99.1k, False: 203k] ------------------ 1434| 302k| ts->cdf.m.comp_fwd_ref[0][ctx1])) 1435| 99.1k| { 1436| 99.1k| const int ctx2 = av1_get_fwd_ref_2_ctx(t->a, &t->l, by4, bx4, 1437| 99.1k| have_top, have_left); 1438| 99.1k| b->ref[0] = 2 + dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 99.1k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1439| 99.1k| ts->cdf.m.comp_fwd_ref[2][ctx2]); 1440| 203k| } else { 1441| 203k| const int ctx2 = av1_get_fwd_ref_1_ctx(t->a, &t->l, by4, bx4, 1442| 203k| have_top, have_left); 1443| 203k| b->ref[0] = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 203k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1444| 203k| ts->cdf.m.comp_fwd_ref[1][ctx2]); 1445| 203k| } 1446| | 1447| | // second reference (bw) 1448| 302k| const int ctx3 = av1_get_bwd_ref_ctx(t->a, &t->l, by4, bx4, 1449| 302k| have_top, have_left); 1450| 302k| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 302k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1450:21): [True: 166k, False: 136k] ------------------ 1451| 302k| ts->cdf.m.comp_bwd_ref[0][ctx3])) 1452| 166k| { 1453| 166k| b->ref[1] = 6; 1454| 166k| } else { 1455| 136k| const int ctx4 = av1_get_bwd_ref_1_ctx(t->a, &t->l, by4, bx4, 1456| 136k| have_top, have_left); 1457| 136k| b->ref[1] = 4 + dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 136k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1458| 136k| ts->cdf.m.comp_bwd_ref[1][ctx4]); 1459| 136k| } 1460| 302k| } else { 1461| | // unidir 1462| 69.7k| const int uctx_p = av1_get_uni_p_ctx(t->a, &t->l, by4, bx4, ------------------ | | 280| 69.7k|#define av1_get_uni_p_ctx av1_get_ref_ctx ------------------ 1463| 69.7k| have_top, have_left); 1464| 69.7k| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 69.7k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1464:21): [True: 17.5k, False: 52.2k] ------------------ 1465| 69.7k| ts->cdf.m.comp_uni_ref[0][uctx_p])) 1466| 17.5k| { 1467| 17.5k| b->ref[0] = 4; 1468| 17.5k| b->ref[1] = 6; 1469| 52.2k| } else { 1470| 52.2k| const int uctx_p1 = av1_get_uni_p1_ctx(t->a, &t->l, by4, bx4, 1471| 52.2k| have_top, have_left); 1472| 52.2k| b->ref[0] = 0; 1473| 52.2k| b->ref[1] = 1 + dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 52.2k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1474| 52.2k| ts->cdf.m.comp_uni_ref[1][uctx_p1]); 1475| 52.2k| if (b->ref[1] == 2) { ------------------ | Branch (1475:25): [True: 32.7k, False: 19.4k] ------------------ 1476| 32.7k| const int uctx_p2 = av1_get_uni_p2_ctx(t->a, &t->l, by4, bx4, ------------------ | | 281| 32.7k|#define av1_get_uni_p2_ctx av1_get_fwd_ref_2_ctx ------------------ 1477| 32.7k| have_top, have_left); 1478| 32.7k| b->ref[1] += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 32.7k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1479| 32.7k| ts->cdf.m.comp_uni_ref[2][uctx_p2]); 1480| 32.7k| } 1481| 52.2k| } 1482| 69.7k| } 1483| 372k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 372k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 372k] | | ------------------ | | 35| 372k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 372k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1484| 0| printf("Post-refs[%d/%d]: r=%d\n", 1485| 0| b->ref[0], b->ref[1], ts->msac.rng); 1486| | 1487| 372k| refmvs_candidate mvstack[8]; 1488| 372k| int n_mvs, ctx; 1489| 372k| dav1d_refmvs_find(&t->rt, mvstack, &n_mvs, &ctx, 1490| 372k| (union refmvs_refpair) { .ref = { 1491| 372k| b->ref[0] + 1, b->ref[1] + 1 }}, 1492| 372k| bs, intra_edge_flags, t->by, t->bx); 1493| | 1494| 372k| b->inter_mode = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 372k|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 1495| 372k| ts->cdf.m.comp_inter_mode[ctx], 1496| 372k| N_COMP_INTER_PRED_MODES - 1); 1497| 372k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 372k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 372k] | | ------------------ | | 35| 372k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 372k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1498| 0| printf("Post-compintermode[%d,ctx=%d,n_mvs=%d]: r=%d\n", 1499| 0| b->inter_mode, ctx, n_mvs, ts->msac.rng); 1500| | 1501| 372k| const uint8_t *const im = dav1d_comp_inter_pred_modes[b->inter_mode]; 1502| 372k| b->drl_idx = NEAREST_DRL; 1503| 372k| if (b->inter_mode == NEWMV_NEWMV) { ------------------ | Branch (1503:17): [True: 76.2k, False: 296k] ------------------ 1504| 76.2k| if (n_mvs > 1) { // NEARER, NEAR or NEARISH ------------------ | Branch (1504:21): [True: 76.2k, False: 2] ------------------ 1505| 76.2k| const int drl_ctx_v1 = get_drl_context(mvstack, 0); 1506| 76.2k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 76.2k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1507| 76.2k| ts->cdf.m.drl_bit[drl_ctx_v1]); 1508| 76.2k| if (b->drl_idx == NEARER_DRL && n_mvs > 2) { ------------------ | Branch (1508:25): [True: 47.9k, False: 28.2k] | Branch (1508:53): [True: 9.04k, False: 38.9k] ------------------ 1509| 9.04k| const int drl_ctx_v2 = get_drl_context(mvstack, 1); 1510| 9.04k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 9.04k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1511| 9.04k| ts->cdf.m.drl_bit[drl_ctx_v2]); 1512| 9.04k| } 1513| 76.2k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 76.2k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 76.2k] | | ------------------ | | 35| 76.2k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 76.2k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1514| 0| printf("Post-drlidx[%d,n_mvs=%d]: r=%d\n", 1515| 0| b->drl_idx, n_mvs, ts->msac.rng); 1516| 76.2k| } 1517| 296k| } else if (im[0] == NEARMV || im[1] == NEARMV) { ------------------ | Branch (1517:24): [True: 66.4k, False: 230k] | Branch (1517:43): [True: 11.6k, False: 218k] ------------------ 1518| 78.3k| b->drl_idx = NEARER_DRL; 1519| 78.3k| if (n_mvs > 2) { // NEAR or NEARISH ------------------ | Branch (1519:21): [True: 17.1k, False: 61.1k] ------------------ 1520| 17.1k| const int drl_ctx_v2 = get_drl_context(mvstack, 1); 1521| 17.1k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 17.1k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1522| 17.1k| ts->cdf.m.drl_bit[drl_ctx_v2]); 1523| 17.1k| if (b->drl_idx == NEAR_DRL && n_mvs > 3) { ------------------ | Branch (1523:25): [True: 7.92k, False: 9.21k] | Branch (1523:51): [True: 5.51k, False: 2.41k] ------------------ 1524| 5.51k| const int drl_ctx_v3 = get_drl_context(mvstack, 2); 1525| 5.51k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 5.51k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1526| 5.51k| ts->cdf.m.drl_bit[drl_ctx_v3]); 1527| 5.51k| } 1528| 17.1k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 17.1k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 17.1k] | | ------------------ | | 35| 17.1k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 17.1k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1529| 0| printf("Post-drlidx[%d,n_mvs=%d]: r=%d\n", 1530| 0| b->drl_idx, n_mvs, ts->msac.rng); 1531| 17.1k| } 1532| 78.3k| } 1533| 372k| assert(b->drl_idx >= NEAREST_DRL && b->drl_idx <= NEARISH_DRL); ------------------ | Branch (1533:13): [True: 372k, False: 288] | Branch (1533:13): [True: 372k, False: 18.4E] ------------------ 1534| | 1535| 372k|#define assign_comp_mv(idx) \ 1536| 372k| switch (im[idx]) { \ 1537| 372k| case NEARMV: \ 1538| 372k| case NEARESTMV: \ 1539| 372k| b->mv[idx] = mvstack[b->drl_idx].mv.mv[idx]; \ 1540| 372k| fix_mv_precision(f->frame_hdr, &b->mv[idx]); \ 1541| 372k| break; \ 1542| 372k| case GLOBALMV: \ 1543| 372k| has_subpel_filter |= \ 1544| 372k| f->frame_hdr->gmv[b->ref[idx]].type == DAV1D_WM_TYPE_TRANSLATION; \ 1545| 372k| b->mv[idx] = get_gmv_2d(&f->frame_hdr->gmv[b->ref[idx]], \ 1546| 372k| t->bx, t->by, bw4, bh4, f->frame_hdr); \ 1547| 372k| break; \ 1548| 372k| case NEWMV: \ 1549| 372k| b->mv[idx] = mvstack[b->drl_idx].mv.mv[idx]; \ 1550| 372k| const int mv_prec = f->frame_hdr->hp - f->frame_hdr->force_integer_mv; \ 1551| 372k| read_mv_residual(ts, &b->mv[idx], mv_prec); \ 1552| 372k| break; \ 1553| 372k| } 1554| 372k| has_subpel_filter = imin(bw4, bh4) == 1 || ------------------ | Branch (1554:33): [True: 110, False: 372k] ------------------ 1555| 372k| b->inter_mode != GLOBALMV_GLOBALMV; ------------------ | Branch (1555:33): [True: 332k, False: 40.2k] ------------------ 1556| 372k| assign_comp_mv(0); ------------------ | | 1536| 372k| switch (im[idx]) { \ | | ------------------ | | | Branch (1536:21): [True: 372k, False: 18.4E] | | ------------------ | | 1537| 66.6k| case NEARMV: \ | | ------------------ | | | Branch (1537:13): [True: 66.6k, False: 305k] | | ------------------ | | 1538| 226k| case NEARESTMV: \ | | ------------------ | | | Branch (1538:13): [True: 159k, False: 212k] | | ------------------ | | 1539| 226k| b->mv[idx] = mvstack[b->drl_idx].mv.mv[idx]; \ | | 1540| 226k| fix_mv_precision(f->frame_hdr, &b->mv[idx]); \ | | 1541| 226k| break; \ | | 1542| 66.6k| case GLOBALMV: \ | | ------------------ | | | Branch (1542:13): [True: 40.2k, False: 332k] | | ------------------ | | 1543| 40.2k| has_subpel_filter |= \ | | 1544| 40.2k| f->frame_hdr->gmv[b->ref[idx]].type == DAV1D_WM_TYPE_TRANSLATION; \ | | 1545| 40.2k| b->mv[idx] = get_gmv_2d(&f->frame_hdr->gmv[b->ref[idx]], \ | | 1546| 40.2k| t->bx, t->by, bw4, bh4, f->frame_hdr); \ | | 1547| 40.2k| break; \ | | 1548| 106k| case NEWMV: \ | | ------------------ | | | Branch (1548:13): [True: 106k, False: 265k] | | ------------------ | | 1549| 106k| b->mv[idx] = mvstack[b->drl_idx].mv.mv[idx]; \ | | 1550| 106k| const int mv_prec = f->frame_hdr->hp - f->frame_hdr->force_integer_mv; \ | | 1551| 106k| read_mv_residual(ts, &b->mv[idx], mv_prec); \ | | 1552| 106k| break; \ | | 1553| 372k| } ------------------ 1557| 372k| assign_comp_mv(1); ------------------ | | 1536| 372k| switch (im[idx]) { \ | | ------------------ | | | Branch (1536:21): [True: 373k, False: 18.4E] | | ------------------ | | 1537| 65.7k| case NEARMV: \ | | ------------------ | | | Branch (1537:13): [True: 65.7k, False: 306k] | | ------------------ | | 1538| 217k| case NEARESTMV: \ | | ------------------ | | | Branch (1538:13): [True: 151k, False: 221k] | | ------------------ | | 1539| 217k| b->mv[idx] = mvstack[b->drl_idx].mv.mv[idx]; \ | | 1540| 217k| fix_mv_precision(f->frame_hdr, &b->mv[idx]); \ | | 1541| 217k| break; \ | | 1542| 65.7k| case GLOBALMV: \ | | ------------------ | | | Branch (1542:13): [True: 40.2k, False: 332k] | | ------------------ | | 1543| 40.2k| has_subpel_filter |= \ | | 1544| 40.2k| f->frame_hdr->gmv[b->ref[idx]].type == DAV1D_WM_TYPE_TRANSLATION; \ | | 1545| 40.2k| b->mv[idx] = get_gmv_2d(&f->frame_hdr->gmv[b->ref[idx]], \ | | 1546| 40.2k| t->bx, t->by, bw4, bh4, f->frame_hdr); \ | | 1547| 40.2k| break; \ | | 1548| 115k| case NEWMV: \ | | ------------------ | | | Branch (1548:13): [True: 115k, False: 256k] | | ------------------ | | 1549| 115k| b->mv[idx] = mvstack[b->drl_idx].mv.mv[idx]; \ | | 1550| 115k| const int mv_prec = f->frame_hdr->hp - f->frame_hdr->force_integer_mv; \ | | 1551| 115k| read_mv_residual(ts, &b->mv[idx], mv_prec); \ | | 1552| 115k| break; \ | | 1553| 372k| } ------------------ 1558| 372k|#undef assign_comp_mv 1559| 372k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 372k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 372k] | | ------------------ | | 35| 372k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 372k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1560| 0| printf("Post-residual_mv[1:y=%d,x=%d,2:y=%d,x=%d]: r=%d\n", 1561| 0| b->mv[0].y, b->mv[0].x, b->mv[1].y, b->mv[1].x, 1562| 0| ts->msac.rng); 1563| | 1564| | // jnt_comp vs. seg vs. wedge 1565| 372k| int is_segwedge = 0; 1566| 372k| if (f->seq_hdr->masked_compound) { ------------------ | Branch (1566:17): [True: 295k, False: 77.3k] ------------------ 1567| 295k| const int mask_ctx = get_mask_comp_ctx(t->a, &t->l, by4, bx4); 1568| | 1569| 295k| is_segwedge = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 295k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1570| 295k| ts->cdf.m.mask_comp[mask_ctx]); 1571| 295k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 295k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 295k] | | ------------------ | | 35| 295k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 295k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1572| 0| printf("Post-segwedge_vs_jntavg[%d,ctx=%d]: r=%d\n", 1573| 0| is_segwedge, mask_ctx, ts->msac.rng); 1574| 295k| } 1575| | 1576| 372k| if (!is_segwedge) { ------------------ | Branch (1576:17): [True: 305k, False: 67.7k] ------------------ 1577| 305k| if (f->seq_hdr->jnt_comp) { ------------------ | Branch (1577:21): [True: 228k, False: 76.4k] ------------------ 1578| 228k| const int jnt_ctx = 1579| 228k| get_jnt_comp_ctx(f->seq_hdr->order_hint_n_bits, 1580| 228k| f->cur.frame_hdr->frame_offset, 1581| 228k| f->refp[b->ref[0]].p.frame_hdr->frame_offset, 1582| 228k| f->refp[b->ref[1]].p.frame_hdr->frame_offset, 1583| 228k| t->a, &t->l, by4, bx4); 1584| 228k| b->comp_type = COMP_INTER_WEIGHTED_AVG + 1585| 228k| dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 228k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1586| 228k| ts->cdf.m.jnt_comp[jnt_ctx]); 1587| 228k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 228k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 228k] | | ------------------ | | 35| 228k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 228k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1588| 0| printf("Post-jnt_comp[%d,ctx=%d[ac:%d,ar:%d,lc:%d,lr:%d]]: r=%d\n", 1589| 0| b->comp_type == COMP_INTER_AVG, 1590| 0| jnt_ctx, t->a->comp_type[bx4], t->a->ref[0][bx4], 1591| 0| t->l.comp_type[by4], t->l.ref[0][by4], 1592| 0| ts->msac.rng); 1593| 228k| } else { 1594| 76.4k| b->comp_type = COMP_INTER_AVG; 1595| 76.4k| } 1596| 305k| } else { 1597| 67.7k| if (wedge_allowed_mask & (1 << bs)) { ------------------ | Branch (1597:21): [True: 50.9k, False: 16.7k] ------------------ 1598| 50.9k| const int ctx = dav1d_wedge_ctx_lut[bs]; 1599| 50.9k| b->comp_type = COMP_INTER_WEDGE - 1600| 50.9k| dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 50.9k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1601| 50.9k| ts->cdf.m.wedge_comp[ctx]); 1602| 50.9k| if (b->comp_type == COMP_INTER_WEDGE) ------------------ | Branch (1602:25): [True: 20.2k, False: 30.6k] ------------------ 1603| 20.2k| b->wedge_idx = dav1d_msac_decode_symbol_adapt16(&ts->msac, ------------------ | | 57| 20.2k|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 1604| 50.9k| ts->cdf.m.wedge_idx[ctx], 15); 1605| 50.9k| } else { 1606| 16.7k| b->comp_type = COMP_INTER_SEG; 1607| 16.7k| } 1608| 67.7k| b->mask_sign = dav1d_msac_decode_bool_equi(&ts->msac); ------------------ | | 53| 67.7k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ 1609| 67.7k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 67.7k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 67.7k] | | ------------------ | | 35| 67.7k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 67.7k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1610| 0| printf("Post-seg/wedge[%d,wedge_idx=%d,sign=%d]: r=%d\n", 1611| 0| b->comp_type == COMP_INTER_WEDGE, 1612| 0| b->wedge_idx, b->mask_sign, ts->msac.rng); 1613| 67.7k| } 1614| 3.52M| } else { 1615| 3.52M| b->comp_type = COMP_INTER_NONE; 1616| | 1617| | // ref 1618| 3.52M| if (seg && seg->ref > 0) { ------------------ | Branch (1618:17): [True: 2.12M, False: 1.39M] | Branch (1618:24): [True: 254k, False: 1.87M] ------------------ 1619| 254k| b->ref[0] = seg->ref - 1; 1620| 3.26M| } else if (seg && (seg->globalmv || seg->skip)) { ------------------ | Branch (1620:24): [True: 1.87M, False: 1.39M] | Branch (1620:32): [True: 1.71M, False: 163k] | Branch (1620:49): [True: 23.9k, False: 139k] ------------------ 1621| 1.73M| b->ref[0] = 0; 1622| 1.73M| } else { 1623| 1.53M| const int ctx1 = av1_get_ref_ctx(t->a, &t->l, by4, bx4, 1624| 1.53M| have_top, have_left); 1625| 1.53M| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 1.53M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1625:21): [True: 524k, False: 1.00M] ------------------ 1626| 1.53M| ts->cdf.m.ref[0][ctx1])) 1627| 524k| { 1628| 524k| const int ctx2 = av1_get_ref_2_ctx(t->a, &t->l, by4, bx4, ------------------ | | 275| 524k|#define av1_get_ref_2_ctx av1_get_bwd_ref_ctx ------------------ 1629| 524k| have_top, have_left); 1630| 524k| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 524k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1630:25): [True: 375k, False: 149k] ------------------ 1631| 524k| ts->cdf.m.ref[1][ctx2])) 1632| 375k| { 1633| 375k| b->ref[0] = 6; 1634| 375k| } else { 1635| 149k| const int ctx3 = av1_get_ref_6_ctx(t->a, &t->l, by4, bx4, ------------------ | | 279| 149k|#define av1_get_ref_6_ctx av1_get_bwd_ref_1_ctx ------------------ 1636| 149k| have_top, have_left); 1637| 149k| b->ref[0] = 4 + dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 149k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1638| 149k| ts->cdf.m.ref[5][ctx3]); 1639| 149k| } 1640| 1.00M| } else { 1641| 1.00M| const int ctx2 = av1_get_ref_3_ctx(t->a, &t->l, by4, bx4, ------------------ | | 276| 1.00M|#define av1_get_ref_3_ctx av1_get_fwd_ref_ctx ------------------ 1642| 1.00M| have_top, have_left); 1643| 1.00M| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 1.00M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1643:25): [True: 146k, False: 861k] ------------------ 1644| 1.00M| ts->cdf.m.ref[2][ctx2])) 1645| 146k| { 1646| 146k| const int ctx3 = av1_get_ref_5_ctx(t->a, &t->l, by4, bx4, ------------------ | | 278| 146k|#define av1_get_ref_5_ctx av1_get_fwd_ref_2_ctx ------------------ 1647| 146k| have_top, have_left); 1648| 146k| b->ref[0] = 2 + dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 146k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1649| 146k| ts->cdf.m.ref[4][ctx3]); 1650| 861k| } else { 1651| 861k| const int ctx3 = av1_get_ref_4_ctx(t->a, &t->l, by4, bx4, ------------------ | | 277| 861k|#define av1_get_ref_4_ctx av1_get_fwd_ref_1_ctx ------------------ 1652| 861k| have_top, have_left); 1653| 861k| b->ref[0] = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 861k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1654| 861k| ts->cdf.m.ref[3][ctx3]); 1655| 861k| } 1656| 1.00M| } 1657| 1.53M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.53M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.53M] | | ------------------ | | 35| 1.53M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.53M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1658| 0| printf("Post-ref[%d]: r=%d\n", b->ref[0], ts->msac.rng); 1659| 1.53M| } 1660| 3.52M| b->ref[1] = -1; 1661| | 1662| 3.52M| refmvs_candidate mvstack[8]; 1663| 3.52M| int n_mvs, ctx; 1664| 3.52M| dav1d_refmvs_find(&t->rt, mvstack, &n_mvs, &ctx, 1665| 3.52M| (union refmvs_refpair) { .ref = { b->ref[0] + 1, -1 }}, 1666| 3.52M| bs, intra_edge_flags, t->by, t->bx); 1667| | 1668| | // mode parsing and mv derivation from ref_mvs 1669| 3.52M| if ((seg && (seg->skip || seg->globalmv)) || ------------------ | Branch (1669:18): [True: 2.13M, False: 1.39M] | Branch (1669:26): [True: 1.97M, False: 154k] | Branch (1669:39): [True: 7.70k, False: 146k] ------------------ 1670| 1.55M| dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 1.55M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1670:17): [True: 1.00M, False: 556k] ------------------ 1671| 1.55M| ts->cdf.m.newmv_mode[ctx & 7])) 1672| 2.98M| { 1673| 2.98M| if ((seg && (seg->skip || seg->globalmv)) || ------------------ | Branch (1673:22): [True: 2.08M, False: 907k] | Branch (1673:30): [True: 1.97M, False: 103k] | Branch (1673:43): [True: 7.71k, False: 95.3k] ------------------ 1674| 1.00M| !dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 1.00M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1674:21): [True: 77.2k, False: 925k] ------------------ 1675| 1.00M| ts->cdf.m.globalmv_mode[(ctx >> 3) & 1])) 1676| 2.06M| { 1677| 2.06M| b->inter_mode = GLOBALMV; 1678| 2.06M| b->mv[0] = get_gmv_2d(&f->frame_hdr->gmv[b->ref[0]], 1679| 2.06M| t->bx, t->by, bw4, bh4, f->frame_hdr); 1680| 2.06M| has_subpel_filter = imin(bw4, bh4) == 1 || ------------------ | Branch (1680:41): [True: 129k, False: 1.93M] ------------------ 1681| 1.93M| f->frame_hdr->gmv[b->ref[0]].type == DAV1D_WM_TYPE_TRANSLATION; ------------------ | Branch (1681:25): [True: 42.2k, False: 1.89M] ------------------ 1682| 2.06M| } else { 1683| 925k| has_subpel_filter = 1; 1684| 925k| if (dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 925k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1684:25): [True: 402k, False: 523k] ------------------ 1685| 925k| ts->cdf.m.refmv_mode[(ctx >> 4) & 15])) 1686| 402k| { // NEAREST, NEARER, NEAR or NEARISH 1687| 402k| b->inter_mode = NEARMV; 1688| 402k| b->drl_idx = NEARER_DRL; 1689| 402k| if (n_mvs > 2) { // NEARER, NEAR or NEARISH ------------------ | Branch (1689:29): [True: 160k, False: 241k] ------------------ 1690| 160k| const int drl_ctx_v2 = get_drl_context(mvstack, 1); 1691| 160k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 160k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1692| 160k| ts->cdf.m.drl_bit[drl_ctx_v2]); 1693| 160k| if (b->drl_idx == NEAR_DRL && n_mvs > 3) { // NEAR or NEARISH ------------------ | Branch (1693:33): [True: 84.9k, False: 75.9k] | Branch (1693:59): [True: 49.3k, False: 35.5k] ------------------ 1694| 49.3k| const int drl_ctx_v3 = 1695| 49.3k| get_drl_context(mvstack, 2); 1696| 49.3k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 49.3k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1697| 49.3k| ts->cdf.m.drl_bit[drl_ctx_v3]); 1698| 49.3k| } 1699| 160k| } 1700| 523k| } else { 1701| 523k| b->inter_mode = NEARESTMV; 1702| 523k| b->drl_idx = NEAREST_DRL; 1703| 523k| } 1704| 925k| assert(b->drl_idx >= NEAREST_DRL && b->drl_idx <= NEARISH_DRL); ------------------ | Branch (1704:21): [True: 926k, False: 18.4E] | Branch (1704:21): [True: 926k, False: 18.4E] ------------------ 1705| 926k| b->mv[0] = mvstack[b->drl_idx].mv.mv[0]; 1706| 926k| if (b->drl_idx < NEAR_DRL) ------------------ | Branch (1706:25): [True: 841k, False: 84.1k] ------------------ 1707| 841k| fix_mv_precision(f->frame_hdr, &b->mv[0]); 1708| 926k| } 1709| | 1710| 2.98M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 2.98M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 2.98M] | | ------------------ | | 35| 2.98M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 2.98M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1711| 0| printf("Post-intermode[%d,drl=%d,mv=y:%d,x:%d,n_mvs=%d]: r=%d\n", 1712| 0| b->inter_mode, b->drl_idx, b->mv[0].y, b->mv[0].x, n_mvs, 1713| 0| ts->msac.rng); 1714| 2.98M| } else { 1715| 535k| has_subpel_filter = 1; 1716| 535k| b->inter_mode = NEWMV; 1717| 535k| b->drl_idx = NEAREST_DRL; 1718| 535k| if (n_mvs > 1) { // NEARER, NEAR or NEARISH ------------------ | Branch (1718:21): [True: 440k, False: 94.4k] ------------------ 1719| 440k| const int drl_ctx_v1 = get_drl_context(mvstack, 0); 1720| 440k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 440k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1721| 440k| ts->cdf.m.drl_bit[drl_ctx_v1]); 1722| 440k| if (b->drl_idx == NEARER_DRL && n_mvs > 2) { // NEAR or NEARISH ------------------ | Branch (1722:25): [True: 149k, False: 290k] | Branch (1722:53): [True: 94.1k, False: 55.5k] ------------------ 1723| 94.1k| const int drl_ctx_v2 = get_drl_context(mvstack, 1); 1724| 94.1k| b->drl_idx += dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 94.1k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1725| 94.1k| ts->cdf.m.drl_bit[drl_ctx_v2]); 1726| 94.1k| } 1727| 440k| } 1728| 535k| assert(b->drl_idx >= NEAREST_DRL && b->drl_idx <= NEARISH_DRL); ------------------ | Branch (1728:17): [True: 556k, False: 18.4E] | Branch (1728:17): [True: 556k, False: 18.4E] ------------------ 1729| 556k| if (n_mvs > 1) { ------------------ | Branch (1729:21): [True: 440k, False: 115k] ------------------ 1730| 440k| b->mv[0] = mvstack[b->drl_idx].mv.mv[0]; 1731| 440k| } else { 1732| 115k| assert(!b->drl_idx); ------------------ | Branch (1732:21): [True: 115k, False: 18.4E] ------------------ 1733| 115k| b->mv[0] = mvstack[0].mv.mv[0]; 1734| 115k| fix_mv_precision(f->frame_hdr, &b->mv[0]); 1735| 115k| } 1736| 556k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 556k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 556k] | | ------------------ | | 35| 556k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 556k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1737| 0| printf("Post-intermode[%d,drl=%d]: r=%d\n", 1738| 0| b->inter_mode, b->drl_idx, ts->msac.rng); 1739| 556k| const int mv_prec = f->frame_hdr->hp - f->frame_hdr->force_integer_mv; 1740| 556k| read_mv_residual(ts, &b->mv[0], mv_prec); 1741| 556k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 556k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 556k] | | ------------------ | | 35| 556k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 556k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1742| 0| printf("Post-residualmv[mv=y:%d,x:%d]: r=%d\n", 1743| 0| b->mv[0].y, b->mv[0].x, ts->msac.rng); 1744| 556k| } 1745| | 1746| | // interintra flags 1747| 3.54M| const int ii_sz_grp = dav1d_ymode_size_context[bs]; 1748| 3.54M| if (f->seq_hdr->inter_intra && ------------------ | Branch (1748:17): [True: 3.10M, False: 436k] ------------------ 1749| 3.10M| interintra_allowed_mask & (1 << bs) && ------------------ | Branch (1749:17): [True: 732k, False: 2.37M] ------------------ 1750| 732k| dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 732k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ | Branch (1750:17): [True: 113k, False: 618k] ------------------ 1751| 732k| ts->cdf.m.interintra[ii_sz_grp])) 1752| 113k| { 1753| 113k| b->interintra_mode = dav1d_msac_decode_symbol_adapt4(&ts->msac, ------------------ | | 47| 113k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 1754| 113k| ts->cdf.m.interintra_mode[ii_sz_grp], 1755| 113k| N_INTER_INTRA_PRED_MODES - 1); 1756| 113k| const int wedge_ctx = dav1d_wedge_ctx_lut[bs]; 1757| 113k| b->interintra_type = INTER_INTRA_BLEND + 1758| 113k| dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 113k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1759| 113k| ts->cdf.m.interintra_wedge[wedge_ctx]); 1760| 113k| if (b->interintra_type == INTER_INTRA_WEDGE) ------------------ | Branch (1760:21): [True: 29.4k, False: 84.4k] ------------------ 1761| 29.4k| b->wedge_idx = dav1d_msac_decode_symbol_adapt16(&ts->msac, ------------------ | | 57| 29.4k|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 1762| 113k| ts->cdf.m.wedge_idx[wedge_ctx], 15); 1763| 3.43M| } else { 1764| 3.43M| b->interintra_type = INTER_INTRA_NONE; 1765| 3.43M| } 1766| 3.54M| if (DEBUG_BLOCK_INFO && f->seq_hdr->inter_intra && ------------------ | | 34| 3.54M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 3.54M] | | ------------------ | | 35| 3.54M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 3.54M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ | Branch (1766:37): [True: 0, False: 0] ------------------ 1767| 0| interintra_allowed_mask & (1 << bs)) ------------------ | Branch (1767:17): [True: 0, False: 0] ------------------ 1768| 0| { 1769| 0| printf("Post-interintra[t=%d,m=%d,w=%d]: r=%d\n", 1770| 0| b->interintra_type, b->interintra_mode, 1771| 0| b->wedge_idx, ts->msac.rng); 1772| 0| } 1773| | 1774| | // motion variation 1775| 3.54M| if (f->frame_hdr->switchable_motion_mode && ------------------ | Branch (1775:17): [True: 3.32M, False: 217k] ------------------ 1776| 3.32M| b->interintra_type == INTER_INTRA_NONE && imin(bw4, bh4) >= 2 && ------------------ | Branch (1776:17): [True: 3.21M, False: 108k] | Branch (1776:59): [True: 2.64M, False: 574k] ------------------ 1777| | // is not warped global motion 1778| 2.64M| !(!f->frame_hdr->force_integer_mv && b->inter_mode == GLOBALMV && ------------------ | Branch (1778:19): [True: 2.59M, False: 46.1k] | Branch (1778:54): [True: 1.77M, False: 822k] ------------------ 1779| 1.77M| f->frame_hdr->gmv[b->ref[0]].type > DAV1D_WM_TYPE_TRANSLATION) && ------------------ | Branch (1779:19): [True: 798k, False: 977k] ------------------ 1780| | // has overlappable neighbours 1781| 1.84M| ((have_left && findoddzero(&t->l.intra[by4 + 1], h4 >> 1)) || ------------------ | Branch (1781:19): [True: 1.10M, False: 745k] | Branch (1781:32): [True: 1.03M, False: 66.5k] ------------------ 1782| 811k| (have_top && findoddzero(&t->a->intra[bx4 + 1], w4 >> 1)))) ------------------ | Branch (1782:19): [True: 792k, False: 19.4k] | Branch (1782:31): [True: 774k, False: 17.7k] ------------------ 1783| 1.80M| { 1784| | // reaching here means the block allows obmc - check warp by 1785| | // finding matching-ref blocks in top/left edges 1786| 1.80M| uint64_t mask[2] = { 0, 0 }; 1787| 1.80M| find_matching_ref(t, intra_edge_flags, bw4, bh4, w4, h4, 1788| 1.80M| have_left, have_top, b->ref[0], mask); 1789| 1.80M| const int allow_warp = !f->svc[b->ref[0]][0].scale && ------------------ | Branch (1789:40): [True: 1.57M, False: 235k] ------------------ 1790| 1.57M| !f->frame_hdr->force_integer_mv && ------------------ | Branch (1790:21): [True: 1.55M, False: 16.0k] ------------------ 1791| 1.55M| f->frame_hdr->warp_motion && (mask[0] | mask[1]); ------------------ | Branch (1791:21): [True: 600k, False: 956k] | Branch (1791:50): [True: 541k, False: 59.2k] ------------------ 1792| | 1793| 1.80M| b->motion_mode = allow_warp ? ------------------ | Branch (1793:34): [True: 541k, False: 1.26M] ------------------ 1794| 541k| dav1d_msac_decode_symbol_adapt4(&ts->msac, ------------------ | | 47| 541k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 1795| 541k| ts->cdf.m.motion_mode[bs], 2) : 1796| 1.80M| dav1d_msac_decode_bool_adapt(&ts->msac, ts->cdf.m.obmc[bs]); ------------------ | | 52| 1.26M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 1797| 1.80M| if (b->motion_mode == MM_WARP) { ------------------ | Branch (1797:21): [True: 169k, False: 1.63M] ------------------ 1798| 169k| has_subpel_filter = 0; 1799| 169k| derive_warpmv(t, bw4, bh4, mask, b->mv[0], &t->warpmv); 1800| 169k|#define signabs(v) v < 0 ? '-' : ' ', abs(v) 1801| 169k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 169k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 169k] | | ------------------ | | 35| 169k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 169k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1802| 0| printf("[ %c%x %c%x %c%x\n %c%x %c%x %c%x ]\n" 1803| 0| "alpha=%c%x, beta=%c%x, gamma=%c%x, delta=%c%x, " 1804| 0| "mv=y:%d,x:%d\n", 1805| 0| signabs(t->warpmv.matrix[0]), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1806| 0| signabs(t->warpmv.matrix[1]), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1807| 0| signabs(t->warpmv.matrix[2]), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1808| 0| signabs(t->warpmv.matrix[3]), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1809| 0| signabs(t->warpmv.matrix[4]), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1810| 0| signabs(t->warpmv.matrix[5]), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1811| 0| signabs(t->warpmv.u.p.alpha), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1812| 0| signabs(t->warpmv.u.p.beta), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1813| 0| signabs(t->warpmv.u.p.gamma), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1814| 0| signabs(t->warpmv.u.p.delta), ------------------ | | 1800| 0|#define signabs(v) v < 0 ? '-' : ' ', abs(v) | | ------------------ | | | Branch (1800:20): [True: 0, False: 0] | | ------------------ ------------------ 1815| 0| b->mv[0].y, b->mv[0].x); 1816| 169k|#undef signabs 1817| 169k| if (t->frame_thread.pass) { ------------------ | Branch (1817:25): [True: 169k, False: 455] ------------------ 1818| 169k| if (t->warpmv.type == DAV1D_WM_TYPE_AFFINE) { ------------------ | Branch (1818:29): [True: 153k, False: 15.6k] ------------------ 1819| 153k| b->matrix[0] = t->warpmv.matrix[2] - 0x10000; 1820| 153k| b->matrix[1] = t->warpmv.matrix[3]; 1821| 153k| b->matrix[2] = t->warpmv.matrix[4]; 1822| 153k| b->matrix[3] = t->warpmv.matrix[5] - 0x10000; 1823| 153k| } else { 1824| 15.6k| b->matrix[0] = INT16_MIN; 1825| 15.6k| } 1826| 169k| } 1827| 169k| } 1828| | 1829| 1.80M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 1.80M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.80M] | | ------------------ | | 35| 1.80M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.80M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1830| 0| printf("Post-motionmode[%d]: r=%d [mask: 0x%" PRIx64 "/0x%" 1831| 0| PRIx64 "]\n", b->motion_mode, ts->msac.rng, mask[0], 1832| 0| mask[1]); 1833| 1.80M| } else { 1834| 1.73M| b->motion_mode = MM_TRANSLATION; 1835| 1.73M| } 1836| 3.54M| } 1837| | 1838| | // subpel filter 1839| 3.94M| enum Dav1dFilterMode filter[2]; 1840| 3.94M| if (f->frame_hdr->subpel_filter_mode == DAV1D_FILTER_SWITCHABLE) { ------------------ | Branch (1840:13): [True: 2.18M, False: 1.76M] ------------------ 1841| 2.18M| if (has_subpel_filter) { ------------------ | Branch (1841:17): [True: 910k, False: 1.27M] ------------------ 1842| 910k| const int comp = b->comp_type != COMP_INTER_NONE; 1843| 910k| const int ctx1 = get_filter_ctx(t->a, &t->l, comp, 0, b->ref[0], 1844| 910k| by4, bx4); 1845| 910k| filter[0] = dav1d_msac_decode_symbol_adapt4(&ts->msac, ------------------ | | 47| 910k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 1846| 910k| ts->cdf.m.filter[0][ctx1], 1847| 910k| DAV1D_N_SWITCHABLE_FILTERS - 1); 1848| 910k| if (f->seq_hdr->dual_filter) { ------------------ | Branch (1848:21): [True: 452k, False: 458k] ------------------ 1849| 452k| const int ctx2 = get_filter_ctx(t->a, &t->l, comp, 1, 1850| 452k| b->ref[0], by4, bx4); 1851| 452k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 452k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 452k] | | ------------------ | | 35| 452k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 452k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1852| 0| printf("Post-subpel_filter1[%d,ctx=%d]: r=%d\n", 1853| 0| filter[0], ctx1, ts->msac.rng); 1854| 452k| filter[1] = dav1d_msac_decode_symbol_adapt4(&ts->msac, ------------------ | | 47| 452k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 1855| 452k| ts->cdf.m.filter[1][ctx2], 1856| 452k| DAV1D_N_SWITCHABLE_FILTERS - 1); 1857| 452k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 452k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 452k] | | ------------------ | | 35| 452k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 452k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1858| 0| printf("Post-subpel_filter2[%d,ctx=%d]: r=%d\n", 1859| 0| filter[1], ctx2, ts->msac.rng); 1860| 458k| } else { 1861| 458k| filter[1] = filter[0]; 1862| 458k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 458k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 458k] | | ------------------ | | 35| 458k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 458k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1863| 0| printf("Post-subpel_filter[%d,ctx=%d]: r=%d\n", 1864| 0| filter[0], ctx1, ts->msac.rng); 1865| 458k| } 1866| 1.27M| } else { 1867| 1.27M| filter[0] = filter[1] = DAV1D_FILTER_8TAP_REGULAR; 1868| 1.27M| } 1869| 2.18M| } else { 1870| 1.76M| filter[0] = filter[1] = f->frame_hdr->subpel_filter_mode; 1871| 1.76M| } 1872| 3.94M| b->filter2d = dav1d_filter_2d[filter[1]][filter[0]]; 1873| | 1874| 3.94M| read_vartx_tree(t, b, bs, bx4, by4); 1875| | 1876| | // reconstruction 1877| 3.94M| if (t->frame_thread.pass == 1) { ------------------ | Branch (1877:13): [True: 3.94M, False: 1.83k] ------------------ 1878| 3.94M| f->bd_fn.read_coef_blocks(t, bs, b); 1879| 3.94M| } else { 1880| 1.83k| if (f->bd_fn.recon_b_inter(t, bs, b)) return -1; ------------------ | Branch (1880:17): [True: 0, False: 1.83k] ------------------ 1881| 1.83k| } 1882| | 1883| 3.94M| if (f->frame_hdr->loopfilter.level_y[0] || ------------------ | Branch (1883:13): [True: 3.41M, False: 525k] ------------------ 1884| 525k| f->frame_hdr->loopfilter.level_y[1]) ------------------ | Branch (1884:13): [True: 222k, False: 302k] ------------------ 1885| 3.63M| { 1886| 3.63M| const int is_globalmv = 1887| 3.63M| b->inter_mode == (is_comp ? GLOBALMV_GLOBALMV : GLOBALMV); ------------------ | Branch (1887:35): [True: 330k, False: 3.30M] ------------------ 1888| 3.63M| const uint8_t (*const lf_lvls)[8][2] = (const uint8_t (*)[8][2]) 1889| 3.63M| &ts->lflvl[b->seg_id][0][b->ref[0] + 1][!is_globalmv]; 1890| 3.63M| const uint16_t tx_split[2] = { b->tx_split0, b->tx_split1 }; 1891| 3.63M| enum RectTxfmSize ytx = b->max_ytx, uvtx = b->uvtx; 1892| 3.63M| if (f->frame_hdr->segmentation.lossless[b->seg_id]) { ------------------ | Branch (1892:17): [True: 29.6k, False: 3.60M] ------------------ 1893| 29.6k| ytx = (enum RectTxfmSize) TX_4X4; 1894| 29.6k| uvtx = (enum RectTxfmSize) TX_4X4; 1895| 29.6k| } 1896| 3.63M| dav1d_create_lf_mask_inter(t->lf_mask, f->lf.level, f->b4_stride, lf_lvls, 1897| 3.63M| t->bx, t->by, f->w4, f->h4, b->skip, bs, 1898| 3.63M| ytx, tx_split, uvtx, f->cur.p.layout, 1899| 3.63M| &t->a->tx_lpf_y[bx4], &t->l.tx_lpf_y[by4], 1900| 3.63M| has_chroma ? &t->a->tx_lpf_uv[cbx4] : NULL, ------------------ | Branch (1900:40): [True: 1.11M, False: 2.51M] ------------------ 1901| 3.63M| has_chroma ? &t->l.tx_lpf_uv[cby4] : NULL); ------------------ | Branch (1901:40): [True: 1.11M, False: 2.51M] ------------------ 1902| 3.63M| } 1903| | 1904| | // context updates 1905| 3.94M| if (is_comp) ------------------ | Branch (1905:13): [True: 398k, False: 3.54M] ------------------ 1906| 398k| splat_tworef_mv(f->c, t, bs, b, bw4, bh4); 1907| 3.54M| else 1908| 3.54M| splat_oneref_mv(f->c, t, bs, b, bw4, bh4); 1909| 3.94M| BlockContext *edge = t->a; 1910| 11.7M| for (int i = 0, off = bx4; i < 2; i++, off = by4, edge = &t->l) { ------------------ | Branch (1910:36): [True: 7.84M, False: 3.94M] ------------------ 1911| 7.84M|#define set_ctx(rep_macro) \ 1912| 7.84M| rep_macro(edge->seg_pred, off, seg_pred); \ 1913| 7.84M| rep_macro(edge->skip_mode, off, b->skip_mode); \ 1914| 7.84M| rep_macro(edge->intra, off, 0); \ 1915| 7.84M| rep_macro(edge->skip, off, b->skip); \ 1916| 7.84M| rep_macro(edge->pal_sz, off, 0); \ 1917| | /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ 1918| 7.84M| rep_macro(t->pal_sz_uv[i], off, 0); \ 1919| 7.84M| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ 1920| 7.84M| rep_macro(edge->comp_type, off, b->comp_type); \ 1921| 7.84M| rep_macro(edge->filter[0], off, filter[0]); \ 1922| 7.84M| rep_macro(edge->filter[1], off, filter[1]); \ 1923| 7.84M| rep_macro(edge->mode, off, b->inter_mode); \ 1924| 7.84M| rep_macro(edge->ref[0], off, b->ref[0]); \ 1925| 7.84M| rep_macro(edge->ref[1], off, ((uint8_t) b->ref[1])) 1926| 7.84M| case_set(b_dim[2 + i]); ------------------ | | 70| 7.84M| switch (var) { \ | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 1912| 780k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1913| 780k| rep_macro(edge->skip_mode, off, b->skip_mode); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1914| 780k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1915| 780k| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1916| 780k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1917| 780k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1918| 780k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1919| 780k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1920| 780k| rep_macro(edge->comp_type, off, b->comp_type); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1921| 780k| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1922| 780k| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1923| 780k| rep_macro(edge->mode, off, b->inter_mode); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1924| 780k| rep_macro(edge->ref[0], off, b->ref[0]); \ | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1925| 780k| rep_macro(edge->ref[1], off, ((uint8_t) b->ref[1])) | | | | ------------------ | | | | | | 71| 780k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 780k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (71:5): [True: 780k, False: 7.06M] | | ------------------ | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 1912| 1.46M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1913| 1.46M| rep_macro(edge->skip_mode, off, b->skip_mode); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1914| 1.46M| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1915| 1.46M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1916| 1.46M| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1917| 1.46M| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1918| 1.46M| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1919| 1.46M| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1920| 1.46M| rep_macro(edge->comp_type, off, b->comp_type); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1921| 1.46M| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1922| 1.46M| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1923| 1.46M| rep_macro(edge->mode, off, b->inter_mode); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1924| 1.46M| rep_macro(edge->ref[0], off, b->ref[0]); \ | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1925| 1.46M| rep_macro(edge->ref[1], off, ((uint8_t) b->ref[1])) | | | | ------------------ | | | | | | 72| 1.46M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.46M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (72:5): [True: 1.46M, False: 6.37M] | | ------------------ | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 1912| 1.26M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1913| 1.26M| rep_macro(edge->skip_mode, off, b->skip_mode); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1914| 1.26M| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1915| 1.26M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1916| 1.26M| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1917| 1.26M| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1918| 1.26M| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1919| 1.26M| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1920| 1.26M| rep_macro(edge->comp_type, off, b->comp_type); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1921| 1.26M| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1922| 1.26M| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1923| 1.26M| rep_macro(edge->mode, off, b->inter_mode); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1924| 1.26M| rep_macro(edge->ref[0], off, b->ref[0]); \ | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1925| 1.26M| rep_macro(edge->ref[1], off, ((uint8_t) b->ref[1])) | | | | ------------------ | | | | | | 73| 1.26M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.26M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (73:5): [True: 1.26M, False: 6.58M] | | ------------------ | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 1912| 512k| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1913| 512k| rep_macro(edge->skip_mode, off, b->skip_mode); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1914| 512k| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1915| 512k| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1916| 512k| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1917| 512k| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1918| 512k| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1919| 512k| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1920| 512k| rep_macro(edge->comp_type, off, b->comp_type); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1921| 512k| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1922| 512k| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1923| 512k| rep_macro(edge->mode, off, b->inter_mode); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1924| 512k| rep_macro(edge->ref[0], off, b->ref[0]); \ | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1925| 512k| rep_macro(edge->ref[1], off, ((uint8_t) b->ref[1])) | | | | ------------------ | | | | | | 74| 512k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 512k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (74:5): [True: 512k, False: 7.33M] | | ------------------ | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 1912| 2.38M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1913| 2.38M| rep_macro(edge->skip_mode, off, b->skip_mode); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1914| 2.38M| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1915| 2.38M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1916| 2.38M| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1917| 2.38M| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1918| 2.38M| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1919| 2.38M| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1920| 2.38M| rep_macro(edge->comp_type, off, b->comp_type); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1921| 2.38M| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1922| 2.38M| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1923| 2.38M| rep_macro(edge->mode, off, b->inter_mode); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1924| 2.38M| rep_macro(edge->ref[0], off, b->ref[0]); \ | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1925| 2.38M| rep_macro(edge->ref[1], off, ((uint8_t) b->ref[1])) | | | | ------------------ | | | | | | 75| 2.38M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.38M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.38M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.38M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.38M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (75:5): [True: 2.38M, False: 5.45M] | | ------------------ | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | ------------------ | | | | 1912| 1.44M| rep_macro(edge->seg_pred, off, seg_pred); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1913| 1.44M| rep_macro(edge->skip_mode, off, b->skip_mode); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1914| 1.44M| rep_macro(edge->intra, off, 0); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1915| 1.44M| rep_macro(edge->skip, off, b->skip); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1916| 1.44M| rep_macro(edge->pal_sz, off, 0); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1917| 1.44M| /* see aomedia bug 2183 for why this is outside if (has_chroma) */ \ | | | | 1918| 1.44M| rep_macro(t->pal_sz_uv[i], off, 0); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1919| 1.44M| rep_macro(edge->tx_intra, off, b_dim[2 + i]); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1920| 1.44M| rep_macro(edge->comp_type, off, b->comp_type); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1921| 1.44M| rep_macro(edge->filter[0], off, filter[0]); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1922| 1.44M| rep_macro(edge->filter[1], off, filter[1]); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1923| 1.44M| rep_macro(edge->mode, off, b->inter_mode); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1924| 1.44M| rep_macro(edge->ref[0], off, b->ref[0]); \ | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1925| 1.44M| rep_macro(edge->ref[1], off, ((uint8_t) b->ref[1])) | | | | ------------------ | | | | | | 76| 1.44M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 1.44M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 1.44M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 1.44M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 1.44M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | ------------------ | | | Branch (76:5): [True: 1.44M, False: 6.39M] | | ------------------ | | 77| 0| default: assert(0); \ | | ------------------ | | | Branch (77:5): [True: 0, False: 7.84M] | | ------------------ | | 78| 7.84M| } ------------------ | Branch (1926:13): [Folded, False: 0] ------------------ 1927| 7.84M|#undef set_ctx 1928| 7.84M| } 1929| 3.94M| if (has_chroma) { ------------------ | Branch (1929:13): [True: 1.35M, False: 2.58M] ------------------ 1930| 1.35M| dav1d_memset_pow2[ulog2(cbw4)](&t->a->uvmode[cbx4], DC_PRED); 1931| 1.35M| dav1d_memset_pow2[ulog2(cbh4)](&t->l.uvmode[cby4], DC_PRED); 1932| 1.35M| } 1933| 3.94M| } 1934| | 1935| | // update contexts 1936| 11.8M| if (f->frame_hdr->segmentation.enabled && ------------------ | Branch (1936:9): [True: 3.46M, False: 8.34M] ------------------ 1937| 3.46M| f->frame_hdr->segmentation.update_map) ------------------ | Branch (1937:9): [True: 2.56M, False: 905k] ------------------ 1938| 2.56M| { 1939| 2.56M| uint8_t *seg_ptr = &f->cur_segmap[t->by * f->b4_stride + t->bx]; 1940| 2.56M|#define set_ctx(rep_macro) \ 1941| 2.56M| for (int y = 0; y < bh4; y++) { \ 1942| 2.56M| rep_macro(seg_ptr, 0, b->seg_id); \ 1943| 2.56M| seg_ptr += f->b4_stride; \ 1944| 2.56M| } 1945| 2.56M| case_set(b_dim[2]); ------------------ | | 70| 2.56M| switch (var) { \ | | 71| 163k| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 1941| 488k| for (int y = 0; y < bh4; y++) { \ | | | | ------------------ | | | | | Branch (1941:25): [True: 324k, False: 163k] | | | | ------------------ | | | | 1942| 324k| rep_macro(seg_ptr, 0, b->seg_id); \ | | | | ------------------ | | | | | | 71| 324k| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 324k| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 1943| 324k| seg_ptr += f->b4_stride; \ | | | | 1944| 324k| } | | ------------------ | | | Branch (71:5): [True: 163k, False: 2.40M] | | ------------------ | | 72| 297k| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 1941| 1.16M| for (int y = 0; y < bh4; y++) { \ | | | | ------------------ | | | | | Branch (1941:25): [True: 864k, False: 297k] | | | | ------------------ | | | | 1942| 864k| rep_macro(seg_ptr, 0, b->seg_id); \ | | | | ------------------ | | | | | | 72| 864k| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 864k| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 1943| 864k| seg_ptr += f->b4_stride; \ | | | | 1944| 864k| } | | ------------------ | | | Branch (72:5): [True: 297k, False: 2.26M] | | ------------------ | | 73| 382k| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 1941| 2.13M| for (int y = 0; y < bh4; y++) { \ | | | | ------------------ | | | | | Branch (1941:25): [True: 1.75M, False: 382k] | | | | ------------------ | | | | 1942| 1.75M| rep_macro(seg_ptr, 0, b->seg_id); \ | | | | ------------------ | | | | | | 73| 1.75M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 1.75M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 1943| 1.75M| seg_ptr += f->b4_stride; \ | | | | 1944| 1.75M| } | | ------------------ | | | Branch (73:5): [True: 382k, False: 2.18M] | | ------------------ | | 74| 145k| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 1941| 992k| for (int y = 0; y < bh4; y++) { \ | | | | ------------------ | | | | | Branch (1941:25): [True: 846k, False: 145k] | | | | ------------------ | | | | 1942| 846k| rep_macro(seg_ptr, 0, b->seg_id); \ | | | | ------------------ | | | | | | 74| 846k| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 846k| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 1943| 846k| seg_ptr += f->b4_stride; \ | | | | 1944| 846k| } | | ------------------ | | | Branch (74:5): [True: 145k, False: 2.41M] | | ------------------ | | 75| 722k| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 1941| 12.2M| for (int y = 0; y < bh4; y++) { \ | | | | ------------------ | | | | | Branch (1941:25): [True: 11.4M, False: 722k] | | | | ------------------ | | | | 1942| 11.4M| rep_macro(seg_ptr, 0, b->seg_id); \ | | | | ------------------ | | | | | | 75| 11.4M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 11.4M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 11.4M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 11.4M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 11.4M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1943| 11.4M| seg_ptr += f->b4_stride; \ | | | | 1944| 11.4M| } | | ------------------ | | | Branch (75:5): [True: 722k, False: 1.84M] | | ------------------ | | 76| 854k| case 5: set_ctx(set_ctx32); break; \ | | ------------------ | | | | 1941| 28.0M| for (int y = 0; y < bh4; y++) { \ | | | | ------------------ | | | | | Branch (1941:25): [True: 27.2M, False: 854k] | | | | ------------------ | | | | 1942| 27.2M| rep_macro(seg_ptr, 0, b->seg_id); \ | | | | ------------------ | | | | | | 76| 27.2M| case 5: set_ctx(set_ctx32); break; \ | | | | | | ------------------ | | | | | | | | 66| 27.2M|#define set_ctx32(var, off, val) do { \ | | | | | | | | 67| 27.2M| memset(&(var)[off], val, 32); \ | | | | | | | | 68| 27.2M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (68:14): [Folded, False: 27.2M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 1943| 27.2M| seg_ptr += f->b4_stride; \ | | | | 1944| 27.2M| } | | ------------------ | | | Branch (76:5): [True: 854k, False: 1.70M] | | ------------------ | | 77| 0| default: assert(0); \ | | ------------------ | | | Branch (77:5): [True: 0, False: 2.56M] | | ------------------ | | 78| 2.56M| } ------------------ | Branch (1945:9): [Folded, False: 0] ------------------ 1946| 2.56M|#undef set_ctx 1947| 2.56M| } 1948| 11.8M| if (!b->skip) { ------------------ | Branch (1948:9): [True: 6.60M, False: 5.21M] ------------------ 1949| 6.60M| uint16_t (*noskip_mask)[2] = &t->lf_mask->noskip_mask[by4 >> 1]; 1950| 6.60M| const unsigned mask = (~0U >> (32 - bw4)) << (bx4 & 15); 1951| 6.60M| const int bx_idx = (bx4 & 16) >> 4; 1952| 22.9M| for (int y = 0; y < bh4; y += 2, noskip_mask++) { ------------------ | Branch (1952:25): [True: 16.3M, False: 6.60M] ------------------ 1953| 16.3M| (*noskip_mask)[bx_idx] |= mask; 1954| 16.3M| if (bw4 == 32) // this should be mask >> 16, but it's 0xffffffff anyway ------------------ | Branch (1954:17): [True: 2.96M, False: 13.3M] ------------------ 1955| 2.96M| (*noskip_mask)[1] |= mask; 1956| 16.3M| } 1957| 6.60M| } 1958| | 1959| 11.8M| if (t->frame_thread.pass == 1 && !b->intra && IS_INTER_OR_SWITCH(f->frame_hdr)) { ------------------ | | 36| 5.26M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 3.93M, False: 1.32M] | | ------------------ ------------------ | Branch (1959:9): [True: 11.8M, False: 18.4E] | Branch (1959:38): [True: 5.26M, False: 6.57M] ------------------ 1960| 3.93M| const int sby = (t->by - ts->tiling.row_start) >> f->sb_shift; 1961| 3.93M| int (*const lowest_px)[2] = ts->lowest_pixel[sby]; 1962| | 1963| | // keep track of motion vectors for each reference 1964| 3.93M| if (b->comp_type == COMP_INTER_NONE) { ------------------ | Branch (1964:13): [True: 3.54M, False: 396k] ------------------ 1965| | // y 1966| 3.54M| if (imin(bw4, bh4) > 1 && ------------------ | Branch (1966:17): [True: 2.90M, False: 638k] ------------------ 1967| 2.90M| ((b->inter_mode == GLOBALMV && f->gmv_warp_allowed[b->ref[0]]) || ------------------ | Branch (1967:19): [True: 1.92M, False: 974k] | Branch (1967:48): [True: 753k, False: 1.17M] ------------------ 1968| 2.15M| (b->motion_mode == MM_WARP && t->warpmv.type > DAV1D_WM_TYPE_TRANSLATION))) ------------------ | Branch (1968:19): [True: 169k, False: 1.98M] | Branch (1968:48): [True: 154k, False: 15.6k] ------------------ 1969| 907k| { 1970| 907k| affine_lowest_px_luma(t, &lowest_px[b->ref[0]][0], b_dim, 1971| 907k| b->motion_mode == MM_WARP ? &t->warpmv : ------------------ | Branch (1971:39): [True: 154k, False: 753k] ------------------ 1972| 907k| &f->frame_hdr->gmv[b->ref[0]]); 1973| 2.63M| } else { 1974| 2.63M| mc_lowest_px(&lowest_px[b->ref[0]][0], t->by, bh4, b->mv[0].y, 1975| 2.63M| 0, &f->svc[b->ref[0]][1]); 1976| 2.63M| if (b->motion_mode == MM_OBMC) { ------------------ | Branch (1976:21): [True: 421k, False: 2.21M] ------------------ 1977| 421k| obmc_lowest_px(t, lowest_px, 0, b_dim, bx4, by4, w4, h4); 1978| 421k| } 1979| 2.63M| } 1980| | 1981| | // uv 1982| 3.54M| if (has_chroma) { ------------------ | Branch (1982:17): [True: 1.19M, False: 2.34M] ------------------ 1983| | // sub8x8 derivation 1984| 1.19M| int is_sub8x8 = bw4 == ss_hor || bh4 == ss_ver; ------------------ | Branch (1984:33): [True: 61.2k, False: 1.13M] | Branch (1984:50): [True: 48.2k, False: 1.08M] ------------------ 1985| 1.19M| refmvs_block *const *r; 1986| 1.19M| if (is_sub8x8) { ------------------ | Branch (1986:21): [True: 109k, False: 1.08M] ------------------ 1987| 109k| assert(ss_hor == 1); ------------------ | Branch (1987:21): [True: 109k, False: 18.4E] ------------------ 1988| 109k| r = &t->rt.r[(t->by & 31) + 5]; 1989| 109k| if (bw4 == 1) is_sub8x8 &= r[0][t->bx - 1].ref.ref[0] > 0; ------------------ | Branch (1989:25): [True: 61.4k, False: 48.4k] ------------------ 1990| 109k| if (bh4 == ss_ver) is_sub8x8 &= r[-1][t->bx].ref.ref[0] > 0; ------------------ | Branch (1990:25): [True: 62.0k, False: 47.8k] ------------------ 1991| 109k| if (bw4 == 1 && bh4 == ss_ver) ------------------ | Branch (1991:25): [True: 61.4k, False: 48.4k] | Branch (1991:37): [True: 13.6k, False: 47.8k] ------------------ 1992| 13.6k| is_sub8x8 &= r[-1][t->bx - 1].ref.ref[0] > 0; 1993| 109k| } 1994| | 1995| | // chroma prediction 1996| 1.19M| if (is_sub8x8) { ------------------ | Branch (1996:21): [True: 96.5k, False: 1.10M] ------------------ 1997| 96.5k| assert(ss_hor == 1); ------------------ | Branch (1997:21): [True: 96.5k, False: 0] ------------------ 1998| 96.5k| if (bw4 == 1 && bh4 == ss_ver) { ------------------ | Branch (1998:25): [True: 54.1k, False: 42.3k] | Branch (1998:37): [True: 11.2k, False: 42.9k] ------------------ 1999| 11.2k| const refmvs_block *const rr = &r[-1][t->bx - 1]; 2000| 11.2k| mc_lowest_px(&lowest_px[rr->ref.ref[0] - 1][1], 2001| 11.2k| t->by - 1, bh4, rr->mv.mv[0].y, ss_ver, 2002| 11.2k| &f->svc[rr->ref.ref[0] - 1][1]); 2003| 11.2k| } 2004| 96.5k| if (bw4 == 1) { ------------------ | Branch (2004:25): [True: 54.1k, False: 42.3k] ------------------ 2005| 54.1k| const refmvs_block *const rr = &r[0][t->bx - 1]; 2006| 54.1k| mc_lowest_px(&lowest_px[rr->ref.ref[0] - 1][1], 2007| 54.1k| t->by, bh4, rr->mv.mv[0].y, ss_ver, 2008| 54.1k| &f->svc[rr->ref.ref[0] - 1][1]); 2009| 54.1k| } 2010| 96.5k| if (bh4 == ss_ver) { ------------------ | Branch (2010:25): [True: 53.5k, False: 42.9k] ------------------ 2011| 53.5k| const refmvs_block *const rr = &r[-1][t->bx]; 2012| 53.5k| mc_lowest_px(&lowest_px[rr->ref.ref[0] - 1][1], 2013| 53.5k| t->by - 1, bh4, rr->mv.mv[0].y, ss_ver, 2014| 53.5k| &f->svc[rr->ref.ref[0] - 1][1]); 2015| 53.5k| } 2016| 96.5k| mc_lowest_px(&lowest_px[b->ref[0]][1], t->by, bh4, 2017| 96.5k| b->mv[0].y, ss_ver, &f->svc[b->ref[0]][1]); 2018| 1.10M| } else { 2019| 1.10M| if (imin(cbw4, cbh4) > 1 && ------------------ | Branch (2019:25): [True: 636k, False: 465k] ------------------ 2020| 636k| ((b->inter_mode == GLOBALMV && f->gmv_warp_allowed[b->ref[0]]) || ------------------ | Branch (2020:27): [True: 149k, False: 487k] | Branch (2020:56): [True: 24.5k, False: 125k] ------------------ 2021| 612k| (b->motion_mode == MM_WARP && t->warpmv.type > DAV1D_WM_TYPE_TRANSLATION))) ------------------ | Branch (2021:27): [True: 64.2k, False: 548k] | Branch (2021:56): [True: 61.9k, False: 2.22k] ------------------ 2022| 86.5k| { 2023| 86.5k| affine_lowest_px_chroma(t, &lowest_px[b->ref[0]][1], b_dim, 2024| 86.5k| b->motion_mode == MM_WARP ? &t->warpmv : ------------------ | Branch (2024:49): [True: 61.9k, False: 24.5k] ------------------ 2025| 86.5k| &f->frame_hdr->gmv[b->ref[0]]); 2026| 1.01M| } else { 2027| 1.01M| mc_lowest_px(&lowest_px[b->ref[0]][1], 2028| 1.01M| t->by & ~ss_ver, bh4 << (bh4 == ss_ver), 2029| 1.01M| b->mv[0].y, ss_ver, &f->svc[b->ref[0]][1]); 2030| 1.01M| if (b->motion_mode == MM_OBMC) { ------------------ | Branch (2030:29): [True: 310k, False: 705k] ------------------ 2031| 310k| obmc_lowest_px(t, lowest_px, 1, b_dim, bx4, by4, w4, h4); 2032| 310k| } 2033| 1.01M| } 2034| 1.10M| } 2035| 1.19M| } 2036| 3.54M| } else { 2037| | // y 2038| 1.19M| for (int i = 0; i < 2; i++) { ------------------ | Branch (2038:29): [True: 794k, False: 396k] ------------------ 2039| 794k| if (b->inter_mode == GLOBALMV_GLOBALMV && f->gmv_warp_allowed[b->ref[i]]) { ------------------ | Branch (2039:21): [True: 80.5k, False: 714k] | Branch (2039:59): [True: 10.9k, False: 69.5k] ------------------ 2040| 10.9k| affine_lowest_px_luma(t, &lowest_px[b->ref[i]][0], b_dim, 2041| 10.9k| &f->frame_hdr->gmv[b->ref[i]]); 2042| 783k| } else { 2043| 783k| mc_lowest_px(&lowest_px[b->ref[i]][0], t->by, bh4, 2044| 783k| b->mv[i].y, 0, &f->svc[b->ref[i]][1]); 2045| 783k| } 2046| 794k| } 2047| | 2048| | // uv 2049| 476k| if (has_chroma) for (int i = 0; i < 2; i++) { ------------------ | Branch (2049:17): [True: 158k, False: 237k] | Branch (2049:45): [True: 317k, False: 158k] ------------------ 2050| 317k| if (b->inter_mode == GLOBALMV_GLOBALMV && ------------------ | Branch (2050:21): [True: 21.1k, False: 296k] ------------------ 2051| 21.1k| imin(cbw4, cbh4) > 1 && f->gmv_warp_allowed[b->ref[i]]) ------------------ | Branch (2051:21): [True: 14.7k, False: 6.36k] | Branch (2051:45): [True: 2.95k, False: 11.8k] ------------------ 2052| 2.95k| { 2053| 2.95k| affine_lowest_px_chroma(t, &lowest_px[b->ref[i]][1], b_dim, 2054| 2.95k| &f->frame_hdr->gmv[b->ref[i]]); 2055| 314k| } else { 2056| 314k| mc_lowest_px(&lowest_px[b->ref[i]][1], t->by, bh4, 2057| 314k| b->mv[i].y, ss_ver, &f->svc[b->ref[i]][1]); 2058| 314k| } 2059| 317k| } 2060| 396k| } 2061| 3.93M| } 2062| | 2063| 11.8M| return 0; 2064| 11.8M|} decode.c:get_prev_frame_segid: 499| 763k|{ 500| 763k| assert(f->frame_hdr->primary_ref_frame != DAV1D_PRIMARY_REF_NONE); ------------------ | Branch (500:5): [True: 763k, False: 18.4E] ------------------ 501| | 502| 763k| unsigned seg_id = 8; 503| 763k| ref_seg_map += by * stride + bx; 504| 829k| do { 505| 11.5M| for (int x = 0; x < w4; x++) ------------------ | Branch (505:25): [True: 10.7M, False: 829k] ------------------ 506| 10.7M| seg_id = imin(seg_id, ref_seg_map[x]); 507| 829k| ref_seg_map += stride; 508| 829k| } while (--h4 > 0 && seg_id); ------------------ | Branch (508:14): [True: 799k, False: 29.6k] | Branch (508:26): [True: 66.0k, False: 733k] ------------------ 509| 763k| assert(seg_id < 8); ------------------ | Branch (509:5): [True: 760k, False: 2.87k] ------------------ 510| | 511| 760k| return seg_id; 512| 763k|} decode.c:neg_deinterleave: 169| 2.31M|static int neg_deinterleave(int diff, int ref, int max) { 170| 2.31M| if (!ref) return diff; ------------------ | Branch (170:9): [True: 1.96M, False: 354k] ------------------ 171| 354k| if (ref >= (max - 1)) return max - diff - 1; ------------------ | Branch (171:9): [True: 83.7k, False: 270k] ------------------ 172| 270k| if (2 * ref < max) { ------------------ | Branch (172:9): [True: 183k, False: 86.8k] ------------------ 173| 183k| if (diff <= 2 * ref) { ------------------ | Branch (173:13): [True: 144k, False: 39.2k] ------------------ 174| 144k| if (diff & 1) ------------------ | Branch (174:17): [True: 18.4k, False: 125k] ------------------ 175| 18.4k| return ref + ((diff + 1) >> 1); 176| 125k| else 177| 125k| return ref - (diff >> 1); 178| 144k| } 179| 39.2k| return diff; 180| 183k| } else { 181| 86.8k| if (diff <= 2 * (max - ref - 1)) { ------------------ | Branch (181:13): [True: 72.8k, False: 13.9k] ------------------ 182| 72.8k| if (diff & 1) ------------------ | Branch (182:17): [True: 13.8k, False: 59.0k] ------------------ 183| 13.8k| return ref + ((diff + 1) >> 1); 184| 59.0k| else 185| 59.0k| return ref - (diff >> 1); 186| 72.8k| } 187| 13.9k| return max - (diff + 1); 188| 86.8k| } 189| 270k|} decode.c:read_pal_indices: 419| 199k|{ 420| 199k| Dav1dTileState *const ts = t->ts; 421| 199k| const ptrdiff_t stride = bw4 * 4; 422| 199k| assert(pal_idx); ------------------ | Branch (422:5): [True: 199k, False: 3] ------------------ 423| 199k| uint8_t *const pal_tmp = t->scratch.pal_idx_uv; 424| 199k| pal_tmp[0] = dav1d_msac_decode_uniform(&ts->msac, pal_sz); 425| 199k| uint16_t (*const color_map_cdf)[8] = 426| 199k| ts->cdf.m.color_map[pl][pal_sz - 2]; 427| 199k| uint8_t (*const order)[8] = t->scratch.pal_order; 428| 199k| uint8_t *const ctx = t->scratch.pal_ctx; 429| 5.68M| for (int i = 1; i < 4 * (w4 + h4) - 1; i++) { ------------------ | Branch (429:21): [True: 5.48M, False: 199k] ------------------ 430| | // top/left-to-bottom/right diagonals ("wave-front") 431| 5.48M| const int first = imin(i, w4 * 4 - 1); 432| 5.48M| const int last = imax(0, i - h4 * 4 + 1); 433| 5.48M| order_palette(pal_tmp, stride, i, first, last, order, ctx); 434| 60.8M| for (int j = first, m = 0; j >= last; j--, m++) { ------------------ | Branch (434:36): [True: 55.3M, False: 5.48M] ------------------ 435| 55.3M| const int color_idx = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 55.3M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 436| 55.3M| color_map_cdf[ctx[m]], pal_sz - 1); 437| 55.3M| pal_tmp[(i - j) * stride + j] = order[m][color_idx]; 438| 55.3M| } 439| 5.48M| } 440| | 441| 199k| t->c->pal_dsp.pal_idx_finish(pal_idx, pal_tmp, bw4 * 4, bh4 * 4, 442| 199k| w4 * 4, h4 * 4); 443| 199k|} decode.c:order_palette: 356| 5.49M|{ 357| 5.49M| int have_top = i > first; 358| | 359| 5.49M| assert(pal_idx); ------------------ | Branch (359:5): [True: 5.49M, False: 618] ------------------ 360| 5.49M| pal_idx += first + (i - first) * stride; 361| 57.8M| for (int j = first, n = 0; j >= last; have_top = 1, j--, n++, pal_idx += stride - 1) { ------------------ | Branch (361:32): [True: 52.5M, False: 5.25M] ------------------ 362| 52.5M| const int have_left = j > 0; 363| | 364| 52.5M| assert(have_left || have_top); ------------------ | Branch (364:9): [True: 50.1M, False: 2.38M] | Branch (364:9): [True: 2.38M, False: 0] ------------------ 365| | 366| 52.5M|#define add(v_in) do { \ 367| 52.5M| const int v = v_in; \ 368| 52.5M| assert((unsigned)v < 8U); \ 369| 52.5M| order[n][o_idx++] = v; \ 370| 52.5M| mask |= 1 << v; \ 371| 52.5M| } while (0) 372| | 373| 52.5M| unsigned mask = 0; 374| 52.5M| int o_idx = 0; 375| 52.5M| if (!have_left) { ------------------ | Branch (375:13): [True: 2.38M, False: 50.1M] ------------------ 376| 2.38M| ctx[n] = 0; 377| 2.38M| add(pal_idx[-stride]); ------------------ | | 366| 2.38M|#define add(v_in) do { \ | | 367| 2.38M| const int v = v_in; \ | | 368| 2.38M| assert((unsigned)v < 8U); \ | | 369| 2.38M| order[n][o_idx++] = v; \ | | 370| 2.38M| mask |= 1 << v; \ | | 371| 2.38M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 2.38M] | | ------------------ ------------------ | Branch (377:13): [True: 2.38M, False: 18.4E] ------------------ 378| 50.1M| } else if (!have_top) { ------------------ | Branch (378:20): [True: 3.11M, False: 47.0M] ------------------ 379| 3.11M| ctx[n] = 0; 380| 3.11M| add(pal_idx[-1]); ------------------ | | 366| 3.11M|#define add(v_in) do { \ | | 367| 3.11M| const int v = v_in; \ | | 368| 3.11M| assert((unsigned)v < 8U); \ | | 369| 3.11M| order[n][o_idx++] = v; \ | | 370| 3.11M| mask |= 1 << v; \ | | 371| 3.11M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 3.11M] | | ------------------ ------------------ | Branch (380:13): [True: 3.11M, False: 18.4E] ------------------ 381| 47.0M| } else { 382| 47.0M| const int l = pal_idx[-1], t = pal_idx[-stride], tl = pal_idx[-(stride + 1)]; 383| 47.0M| const int same_t_l = t == l; 384| 47.0M| const int same_t_tl = t == tl; 385| 47.0M| const int same_l_tl = l == tl; 386| 47.0M| const int same_all = same_t_l & same_t_tl & same_l_tl; 387| | 388| 47.0M| if (same_all) { ------------------ | Branch (388:17): [True: 21.2M, False: 25.8M] ------------------ 389| 21.2M| ctx[n] = 4; 390| 21.2M| add(t); ------------------ | | 366| 21.2M|#define add(v_in) do { \ | | 367| 21.2M| const int v = v_in; \ | | 368| 21.2M| assert((unsigned)v < 8U); \ | | 369| 21.2M| order[n][o_idx++] = v; \ | | 370| 21.2M| mask |= 1 << v; \ | | 371| 21.2M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 21.2M] | | ------------------ ------------------ | Branch (390:17): [True: 21.2M, False: 11.1k] ------------------ 391| 25.8M| } else if (same_t_l) { ------------------ | Branch (391:24): [True: 4.90M, False: 20.9M] ------------------ 392| 4.90M| ctx[n] = 3; 393| 4.90M| add(t); ------------------ | | 366| 4.90M|#define add(v_in) do { \ | | 367| 4.90M| const int v = v_in; \ | | 368| 4.90M| assert((unsigned)v < 8U); \ | | 369| 4.90M| order[n][o_idx++] = v; \ | | 370| 4.90M| mask |= 1 << v; \ | | 371| 4.90M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 4.90M] | | ------------------ ------------------ | Branch (393:17): [True: 4.90M, False: 448] ------------------ 394| 4.90M| add(tl); ------------------ | | 366| 4.90M|#define add(v_in) do { \ | | 367| 4.90M| const int v = v_in; \ | | 368| 4.90M| assert((unsigned)v < 8U); \ | | 369| 4.90M| order[n][o_idx++] = v; \ | | 370| 4.90M| mask |= 1 << v; \ | | 371| 4.90M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 4.90M] | | ------------------ ------------------ | Branch (394:17): [True: 4.90M, False: 18.4E] ------------------ 395| 20.9M| } else if (same_t_tl | same_l_tl) { ------------------ | Branch (395:24): [True: 16.9M, False: 3.92M] ------------------ 396| 16.9M| ctx[n] = 2; 397| 16.9M| add(tl); ------------------ | | 366| 16.9M|#define add(v_in) do { \ | | 367| 16.9M| const int v = v_in; \ | | 368| 16.9M| assert((unsigned)v < 8U); \ | | 369| 16.9M| order[n][o_idx++] = v; \ | | 370| 16.9M| mask |= 1 << v; \ | | 371| 16.9M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 16.9M] | | ------------------ ------------------ | Branch (397:17): [True: 16.9M, False: 4.63k] ------------------ 398| 16.9M| add(same_t_tl ? l : t); ------------------ | | 366| 16.9M|#define add(v_in) do { \ | | 367| 33.9M| const int v = v_in; \ | | ------------------ | | | Branch (367:23): [True: 8.48M, False: 8.51M] | | ------------------ | | 368| 16.9M| assert((unsigned)v < 8U); \ | | 369| 16.9M| order[n][o_idx++] = v; \ | | 370| 16.9M| mask |= 1 << v; \ | | 371| 16.9M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 16.9M] | | ------------------ ------------------ | Branch (398:17): [True: 16.9M, False: 18.4E] ------------------ 399| 16.9M| } else { 400| 3.92M| ctx[n] = 1; 401| 3.92M| add(imin(t, l)); ------------------ | | 366| 3.92M|#define add(v_in) do { \ | | 367| 3.92M| const int v = v_in; \ | | 368| 3.92M| assert((unsigned)v < 8U); \ | | 369| 6.94M| order[n][o_idx++] = v; \ | | 370| 6.94M| mask |= 1 << v; \ | | 371| 6.94M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 6.94M] | | ------------------ ------------------ | Branch (401:17): [True: 6.94M, False: 18.4E] ------------------ 402| 6.94M| add(imax(t, l)); ------------------ | | 366| 6.94M|#define add(v_in) do { \ | | 367| 6.94M| const int v = v_in; \ | | 368| 6.94M| assert((unsigned)v < 8U); \ | | 369| 6.94M| order[n][o_idx++] = v; \ | | 370| 6.94M| mask |= 1 << v; \ | | 371| 6.94M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 6.94M] | | ------------------ ------------------ | Branch (402:17): [True: 6.94M, False: 2.68k] ------------------ 403| 6.94M| add(tl); ------------------ | | 366| 6.94M|#define add(v_in) do { \ | | 367| 6.94M| const int v = v_in; \ | | 368| 6.94M| assert((unsigned)v < 8U); \ | | 369| 6.94M| order[n][o_idx++] = v; \ | | 370| 6.94M| mask |= 1 << v; \ | | 371| 6.94M| } while (0) | | ------------------ | | | Branch (371:14): [Folded, False: 6.94M] | | ------------------ ------------------ | Branch (403:17): [True: 6.94M, False: 4.20k] ------------------ 404| 6.94M| } 405| 47.0M| } 406| 461M| for (unsigned m = 1, bit = 0; m < 0x100; m <<= 1, bit++) ------------------ | Branch (406:39): [True: 405M, False: 55.5M] ------------------ 407| 405M| if (!(mask & m)) ------------------ | Branch (407:17): [True: 323M, False: 82.8M] ------------------ 408| 323M| order[n][o_idx++] = bit; 409| | assert(o_idx == 8); ------------------ | Branch (409:9): [True: 52.3M, False: 3.23M] ------------------ 410| 55.5M|#undef add 411| 55.5M| } 412| 5.49M|} decode.c:splat_intraref: 566| 5.05M|{ 567| 5.05M| const refmvs_block ALIGN(tmpl, 16) = (refmvs_block) { 568| 5.05M| .ref.ref = { 0, -1 }, 569| 5.05M| .mv.mv[0].n = INVALID_MV, ------------------ | | 40| 5.05M|#define INVALID_MV 0x80008000 ------------------ 570| 5.05M| .bs = bs, 571| 5.05M| .mf = 0, 572| 5.05M| }; 573| 5.05M| c->refmvs_dsp.splat_mv(&t->rt.r[(t->by & 31) + 5], &tmpl, t->bx, bw4, bh4); 574| 5.05M|} decode.c:read_mv_residual: 109| 2.11M|{ 110| 2.11M| MsacContext *const msac = &ts->msac; 111| 2.11M| const enum MVJoint mv_joint = 112| 2.11M| dav1d_msac_decode_symbol_adapt4(msac, ts->cdf.mv.joint, N_MV_JOINTS - 1); ------------------ | | 47| 2.11M|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 113| 2.11M| if (mv_joint & MV_JOINT_V) ------------------ | Branch (113:9): [True: 1.61M, False: 503k] ------------------ 114| 1.61M| ref_mv->y += read_mv_component_diff(msac, &ts->cdf.mv.comp[0], mv_prec); 115| 2.11M| if (mv_joint & MV_JOINT_H) ------------------ | Branch (115:9): [True: 1.49M, False: 625k] ------------------ 116| 1.49M| ref_mv->x += read_mv_component_diff(msac, &ts->cdf.mv.comp[1], mv_prec); 117| 2.11M|} decode.c:read_mv_component_diff: 79| 3.10M|{ 80| 3.10M| const int sign = dav1d_msac_decode_bool_adapt(msac, mv_comp->sign); ------------------ | | 52| 3.10M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 81| 3.10M| const int cl = dav1d_msac_decode_symbol_adapt16(msac, mv_comp->classes, 10); ------------------ | | 57| 3.10M|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 82| 3.10M| int up, fp = 3, hp = 1; 83| | 84| 3.10M| if (!cl) { ------------------ | Branch (84:9): [True: 1.45M, False: 1.64M] ------------------ 85| 1.45M| up = dav1d_msac_decode_bool_adapt(msac, mv_comp->class0); ------------------ | | 52| 1.45M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 86| 1.45M| if (mv_prec >= 0) { // !force_integer_mv ------------------ | Branch (86:13): [True: 740k, False: 710k] ------------------ 87| 740k| fp = dav1d_msac_decode_symbol_adapt4(msac, mv_comp->class0_fp[up], 3); ------------------ | | 47| 740k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 88| 740k| if (mv_prec > 0) // allow_high_precision_mv ------------------ | Branch (88:17): [True: 354k, False: 386k] ------------------ 89| 354k| hp = dav1d_msac_decode_bool_adapt(msac, mv_comp->class0_hp); ------------------ | | 52| 354k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 90| 740k| } 91| 1.64M| } else { 92| 1.64M| up = 1 << cl; 93| 16.3M| for (int n = 0; n < cl; n++) ------------------ | Branch (93:25): [True: 14.6M, False: 1.64M] ------------------ 94| 14.6M| up |= dav1d_msac_decode_bool_adapt(msac, mv_comp->classN[n]) << n; ------------------ | | 52| 14.6M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 95| 1.64M| if (mv_prec >= 0) { // !force_integer_mv ------------------ | Branch (95:13): [True: 263k, False: 1.38M] ------------------ 96| 263k| fp = dav1d_msac_decode_symbol_adapt4(msac, mv_comp->classN_fp, 3); ------------------ | | 47| 263k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 97| 263k| if (mv_prec > 0) // allow_high_precision_mv ------------------ | Branch (97:17): [True: 112k, False: 151k] ------------------ 98| 112k| hp = dav1d_msac_decode_bool_adapt(msac, mv_comp->classN_hp); ------------------ | | 52| 112k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 99| 263k| } 100| 1.64M| } 101| | 102| 3.10M| const int diff = ((up << 3) | (fp << 1) | hp) + 1; 103| | 104| 3.10M| return sign ? -diff : diff; ------------------ | Branch (104:12): [True: 2.29M, False: 806k] ------------------ 105| 3.10M|} decode.c:read_vartx_tree: 448| 5.26M|{ 449| 5.26M| const Dav1dFrameContext *const f = t->f; 450| 5.26M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 451| 5.26M| const int bw4 = b_dim[0], bh4 = b_dim[1]; 452| | 453| | // var-tx tree coding 454| 5.26M| uint16_t tx_split[2] = { 0 }; 455| 5.26M| b->max_ytx = dav1d_max_txfm_size_for_bs[bs][0]; 456| 5.26M| if (!b->skip && (f->frame_hdr->segmentation.lossless[b->seg_id] || ------------------ | Branch (456:9): [True: 1.85M, False: 3.41M] | Branch (456:22): [True: 10.8k, False: 1.84M] ------------------ 457| 1.84M| b->max_ytx == TX_4X4)) ------------------ | Branch (457:22): [True: 102k, False: 1.73M] ------------------ 458| 114k| { 459| 114k| b->max_ytx = b->uvtx = TX_4X4; 460| 114k| if (f->frame_hdr->txfm_mode == DAV1D_TX_SWITCHABLE) { ------------------ | Branch (460:13): [True: 32.0k, False: 82.2k] ------------------ 461| 32.0k| dav1d_memset_pow2[b_dim[2]](&t->a->tx[bx4], TX_4X4); 462| 32.0k| dav1d_memset_pow2[b_dim[3]](&t->l.tx[by4], TX_4X4); 463| 32.0k| } 464| 5.15M| } else if (f->frame_hdr->txfm_mode != DAV1D_TX_SWITCHABLE || b->skip) { ------------------ | Branch (464:16): [True: 4.13M, False: 1.01M] | Branch (464:66): [True: 509k, False: 505k] ------------------ 465| 4.64M| if (f->frame_hdr->txfm_mode == DAV1D_TX_SWITCHABLE) { ------------------ | Branch (465:13): [True: 509k, False: 4.13M] ------------------ 466| 509k| dav1d_memset_pow2[b_dim[2]](&t->a->tx[bx4], b_dim[2 + 0]); 467| 509k| dav1d_memset_pow2[b_dim[3]](&t->l.tx[by4], b_dim[2 + 1]); 468| 509k| } 469| 4.64M| b->uvtx = dav1d_max_txfm_size_for_bs[bs][f->cur.p.layout]; 470| 4.64M| } else { 471| 503k| assert(bw4 <= 16 || bh4 <= 16 || b->max_ytx == TX_64X64); ------------------ | Branch (471:9): [True: 497k, False: 6.12k] | Branch (471:9): [True: 2.60k, False: 3.52k] | Branch (471:9): [True: 3.52k, False: 0] ------------------ 472| 505k| int y, x, y_off, x_off; 473| 505k| const TxfmInfo *const ytx = &dav1d_txfm_dimensions[b->max_ytx]; 474| 1.01M| for (y = 0, y_off = 0; y < bh4; y += ytx->h, y_off++) { ------------------ | Branch (474:32): [True: 510k, False: 505k] ------------------ 475| 1.03M| for (x = 0, x_off = 0; x < bw4; x += ytx->w, x_off++) { ------------------ | Branch (475:36): [True: 520k, False: 510k] ------------------ 476| 520k| read_tx_tree(t, b->max_ytx, 0, tx_split, x_off, y_off); 477| | // contexts are updated inside read_tx_tree() 478| 520k| t->bx += ytx->w; 479| 520k| } 480| 510k| t->bx -= x; 481| 510k| t->by += ytx->h; 482| 510k| } 483| 505k| t->by -= y; 484| 505k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 505k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 505k] | | ------------------ | | 35| 505k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 505k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 485| 0| printf("Post-vartxtree[%x/%x]: r=%d\n", 486| 0| tx_split[0], tx_split[1], t->ts->msac.rng); 487| 505k| b->uvtx = dav1d_max_txfm_size_for_bs[bs][f->cur.p.layout]; 488| 505k| } 489| 5.26M| assert(!(tx_split[0] & ~0x33)); ------------------ | Branch (489:5): [True: 5.26M, False: 178] ------------------ 490| 5.26M| b->tx_split0 = (uint8_t)tx_split[0]; 491| 5.26M| b->tx_split1 = tx_split[1]; 492| 5.26M|} decode.c:read_tx_tree: 123| 968k|{ 124| 968k| const Dav1dFrameContext *const f = t->f; 125| 968k| const int bx4 = t->bx & 31, by4 = t->by & 31; 126| 968k| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[from]; 127| 968k| const int txw = t_dim->lw, txh = t_dim->lh; 128| 968k| int is_split; 129| | 130| 968k| if (depth < 2 && from > (int) TX_4X4) { ------------------ | Branch (130:9): [True: 809k, False: 158k] | Branch (130:22): [True: 809k, False: 1] ------------------ 131| 809k| const int cat = 2 * (TX_64X64 - t_dim->max) - depth; 132| 809k| const int a = t->a->tx[bx4] < txw; 133| 809k| const int l = t->l.tx[by4] < txh; 134| | 135| 809k| is_split = dav1d_msac_decode_bool_adapt(&t->ts->msac, ------------------ | | 52| 809k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 136| 809k| t->ts->cdf.m.txpart[cat][a + l]); 137| 809k| if (is_split) ------------------ | Branch (137:13): [True: 216k, False: 593k] ------------------ 138| 216k| masks[depth] |= 1 << (y_off * 4 + x_off); 139| 809k| } else { 140| 158k| is_split = 0; 141| 158k| } 142| | 143| 968k| if (is_split && t_dim->max > TX_8X8) { ------------------ | Branch (143:9): [True: 216k, False: 752k] | Branch (143:21): [True: 155k, False: 60.7k] ------------------ 144| 155k| const enum RectTxfmSize sub = t_dim->sub; 145| 155k| const TxfmInfo *const sub_t_dim = &dav1d_txfm_dimensions[sub]; 146| 155k| const int txsw = sub_t_dim->w, txsh = sub_t_dim->h; 147| | 148| 155k| read_tx_tree(t, sub, depth + 1, masks, x_off * 2 + 0, y_off * 2 + 0); 149| 155k| t->bx += txsw; 150| 155k| if (txw >= txh && t->bx < f->bw) ------------------ | Branch (150:13): [True: 118k, False: 36.5k] | Branch (150:27): [True: 117k, False: 769] ------------------ 151| 117k| read_tx_tree(t, sub, depth + 1, masks, x_off * 2 + 1, y_off * 2 + 0); 152| 155k| t->bx -= txsw; 153| 155k| t->by += txsh; 154| 155k| if (txh >= txw && t->by < f->bh) { ------------------ | Branch (154:13): [True: 107k, False: 47.6k] | Branch (154:27): [True: 106k, False: 1.30k] ------------------ 155| 106k| read_tx_tree(t, sub, depth + 1, masks, x_off * 2 + 0, y_off * 2 + 1); 156| 106k| t->bx += txsw; 157| 106k| if (txw >= txh && t->bx < f->bw) ------------------ | Branch (157:17): [True: 69.7k, False: 36.5k] | Branch (157:31): [True: 68.9k, False: 742] ------------------ 158| 68.9k| read_tx_tree(t, sub, depth + 1, masks, 159| 68.9k| x_off * 2 + 1, y_off * 2 + 1); 160| 106k| t->bx -= txsw; 161| 106k| } 162| 155k| t->by -= txsh; 163| 812k| } else { 164| 812k| dav1d_memset_pow2[t_dim->lw](&t->a->tx[bx4], is_split ? TX_4X4 : txw); ------------------ | Branch (164:54): [True: 60.8k, False: 752k] ------------------ 165| 812k| dav1d_memset_pow2[t_dim->lh](&t->l.tx[by4], is_split ? TX_4X4 : txh); ------------------ | Branch (165:53): [True: 60.7k, False: 752k] ------------------ 166| 812k| } 167| 968k|} decode.c:splat_intrabc_mv: 535| 1.32M|{ 536| 1.32M| const refmvs_block ALIGN(tmpl, 16) = (refmvs_block) { 537| 1.32M| .ref.ref = { 0, -1 }, 538| 1.32M| .mv.mv[0] = b->mv[0], 539| 1.32M| .bs = bs, 540| 1.32M| .mf = 0, 541| 1.32M| }; 542| 1.32M| c->refmvs_dsp.splat_mv(&t->rt.r[(t->by & 31) + 5], &tmpl, t->bx, bw4, bh4); 543| 1.32M|} decode.c:findoddzero: 339| 1.89M|static inline int findoddzero(const uint8_t *buf, int len) { 340| 2.13M| for (int n = 0; n < len; n++) ------------------ | Branch (340:21): [True: 2.05M, False: 83.7k] ------------------ 341| 2.05M| if (!buf[n * 2]) return 1; ------------------ | Branch (341:13): [True: 1.80M, False: 247k] ------------------ 342| 83.7k| return 0; 343| 1.89M|} decode.c:find_matching_ref: 197| 1.80M|{ 198| 1.80M| /*const*/ refmvs_block *const *r = &t->rt.r[(t->by & 31) + 5]; 199| 1.80M| int count = 0; 200| 1.80M| int have_topleft = have_top && have_left; ------------------ | Branch (200:24): [True: 1.69M, False: 114k] | Branch (200:36): [True: 961k, False: 732k] ------------------ 201| 1.80M| int have_topright = imax(bw4, bh4) < 32 && ------------------ | Branch (201:25): [True: 1.24M, False: 559k] ------------------ 202| 1.24M| have_top && t->bx + bw4 < t->ts->tiling.col_end && ------------------ | Branch (202:25): [True: 1.16M, False: 85.4k] | Branch (202:37): [True: 917k, False: 246k] ------------------ 203| 917k| (intra_edge_flags & EDGE_I444_TOP_HAS_RIGHT); ------------------ | Branch (203:25): [True: 620k, False: 296k] ------------------ 204| | 205| 1.80M|#define bs(rp) dav1d_block_dimensions[(rp)->bs] 206| 1.80M|#define matches(rp) ((rp)->ref.ref[0] == ref + 1 && (rp)->ref.ref[1] == -1) 207| | 208| 1.80M| if (have_top) { ------------------ | Branch (208:9): [True: 1.69M, False: 114k] ------------------ 209| 1.69M| const refmvs_block *r2 = &r[-1][t->bx]; 210| 1.69M| if (matches(r2)) { ------------------ | | 206| 1.69M|#define matches(rp) ((rp)->ref.ref[0] == ref + 1 && (rp)->ref.ref[1] == -1) | | ------------------ | | | Branch (206:22): [True: 1.52M, False: 167k] | | | Branch (206:53): [True: 1.45M, False: 71.6k] | | ------------------ ------------------ 211| 1.45M| masks[0] |= 1; 212| 1.45M| count = 1; 213| 1.45M| } 214| 1.69M| int aw4 = bs(r2)[0]; ------------------ | | 205| 1.69M|#define bs(rp) dav1d_block_dimensions[(rp)->bs] ------------------ 215| 1.69M| if (aw4 >= bw4) { ------------------ | Branch (215:13): [True: 1.57M, False: 118k] ------------------ 216| 1.57M| const int off = t->bx & (aw4 - 1); 217| 1.57M| if (off) have_topleft = 0; ------------------ | Branch (217:17): [True: 126k, False: 1.44M] ------------------ 218| 1.57M| if (aw4 - off > bw4) have_topright = 0; ------------------ | Branch (218:17): [True: 131k, False: 1.44M] ------------------ 219| 1.57M| } else { 220| 118k| unsigned mask = 1 << aw4; 221| 290k| for (int x = aw4; x < w4; x += aw4) { ------------------ | Branch (221:31): [True: 172k, False: 118k] ------------------ 222| 172k| r2 += aw4; 223| 172k| if (matches(r2)) { ------------------ | | 206| 172k|#define matches(rp) ((rp)->ref.ref[0] == ref + 1 && (rp)->ref.ref[1] == -1) | | ------------------ | | | Branch (206:22): [True: 117k, False: 54.5k] | | | Branch (206:53): [True: 107k, False: 9.62k] | | ------------------ ------------------ 224| 107k| masks[0] |= mask; 225| 107k| if (++count >= 8) return; ------------------ | Branch (225:25): [True: 453, False: 107k] ------------------ 226| 107k| } 227| 171k| aw4 = bs(r2)[0]; ------------------ | | 205| 171k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] ------------------ 228| 171k| mask <<= aw4; 229| 171k| } 230| 118k| } 231| 1.69M| } 232| 1.80M| if (have_left) { ------------------ | Branch (232:9): [True: 1.07M, False: 732k] ------------------ 233| 1.07M| /*const*/ refmvs_block *const *r2 = r; 234| 1.07M| if (matches(&r2[0][t->bx - 1])) { ------------------ | | 206| 1.07M|#define matches(rp) ((rp)->ref.ref[0] == ref + 1 && (rp)->ref.ref[1] == -1) | | ------------------ | | | Branch (206:22): [True: 902k, False: 173k] | | | Branch (206:53): [True: 828k, False: 73.9k] | | ------------------ ------------------ 235| 828k| masks[1] |= 1; 236| 828k| if (++count >= 8) return; ------------------ | Branch (236:17): [True: 241, False: 828k] ------------------ 237| 828k| } 238| 1.07M| int lh4 = bs(&r2[0][t->bx - 1])[1]; ------------------ | | 205| 1.07M|#define bs(rp) dav1d_block_dimensions[(rp)->bs] ------------------ 239| 1.07M| if (lh4 >= bh4) { ------------------ | Branch (239:13): [True: 936k, False: 138k] ------------------ 240| 936k| if (t->by & (lh4 - 1)) have_topleft = 0; ------------------ | Branch (240:17): [True: 156k, False: 780k] ------------------ 241| 936k| } else { 242| 138k| unsigned mask = 1 << lh4; 243| 334k| for (int y = lh4; y < h4; y += lh4) { ------------------ | Branch (243:31): [True: 197k, False: 137k] ------------------ 244| 197k| r2 += lh4; 245| 197k| if (matches(&r2[0][t->bx - 1])) { ------------------ | | 206| 197k|#define matches(rp) ((rp)->ref.ref[0] == ref + 1 && (rp)->ref.ref[1] == -1) | | ------------------ | | | Branch (206:22): [True: 136k, False: 60.9k] | | | Branch (206:53): [True: 125k, False: 10.6k] | | ------------------ ------------------ 246| 125k| masks[1] |= mask; 247| 125k| if (++count >= 8) return; ------------------ | Branch (247:25): [True: 1.41k, False: 124k] ------------------ 248| 125k| } 249| 195k| lh4 = bs(&r2[0][t->bx - 1])[1]; ------------------ | | 205| 195k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] ------------------ 250| 195k| mask <<= lh4; 251| 195k| } 252| 138k| } 253| 1.07M| } 254| 1.80M| if (have_topleft && matches(&r[-1][t->bx - 1])) { ------------------ | | 206| 676k|#define matches(rp) ((rp)->ref.ref[0] == ref + 1 && (rp)->ref.ref[1] == -1) | | ------------------ | | | Branch (206:22): [True: 544k, False: 131k] | | | Branch (206:53): [True: 500k, False: 44.2k] | | ------------------ ------------------ | Branch (254:9): [True: 676k, False: 1.12M] ------------------ 255| 500k| masks[1] |= 1ULL << 32; 256| 500k| if (++count >= 8) return; ------------------ | Branch (256:13): [True: 944, False: 499k] ------------------ 257| 500k| } 258| 1.80M| if (have_topright && matches(&r[-1][t->bx + bw4])) { ------------------ | | 206| 489k|#define matches(rp) ((rp)->ref.ref[0] == ref + 1 && (rp)->ref.ref[1] == -1) | | ------------------ | | | Branch (206:22): [True: 398k, False: 91.6k] | | | Branch (206:53): [True: 371k, False: 26.9k] | | ------------------ ------------------ | Branch (258:9): [True: 489k, False: 1.31M] ------------------ 259| 371k| masks[0] |= 1ULL << 32; 260| 371k| } 261| 1.80M|#undef matches 262| 1.80M|} decode.c:derive_warpmv: 268| 169k|{ 269| 169k| int pts[8][2 /* in, out */][2 /* x, y */], np = 0; 270| 169k| /*const*/ refmvs_block *const *r = &t->rt.r[(t->by & 31) + 5]; 271| | 272| 169k|#define add_sample(dx, dy, sx, sy, rp) do { \ 273| 169k| pts[np][0][0] = 16 * (2 * dx + sx * bs(rp)[0]) - 8; \ 274| 169k| pts[np][0][1] = 16 * (2 * dy + sy * bs(rp)[1]) - 8; \ 275| 169k| pts[np][1][0] = pts[np][0][0] + (rp)->mv.mv[0].x; \ 276| 169k| pts[np][1][1] = pts[np][0][1] + (rp)->mv.mv[0].y; \ 277| 169k| np++; \ 278| 169k|} while (0) 279| | 280| | // use masks[] to find the projectable motion vectors in the edges 281| 169k| if ((unsigned) masks[0] == 1 && !(masks[1] >> 32)) { ------------------ | Branch (281:9): [True: 118k, False: 51.8k] | Branch (281:37): [True: 60.6k, False: 57.4k] ------------------ 282| 60.6k| const int off = t->bx & (bs(&r[-1][t->bx])[0] - 1); ------------------ | | 205| 60.6k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] ------------------ 283| 60.6k| add_sample(-off, 0, 1, -1, &r[-1][t->bx]); ------------------ | | 272| 60.6k|#define add_sample(dx, dy, sx, sy, rp) do { \ | | 273| 60.6k| pts[np][0][0] = 16 * (2 * dx + sx * bs(rp)[0]) - 8; \ | | ------------------ | | | | 205| 60.6k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 274| 60.6k| pts[np][0][1] = 16 * (2 * dy + sy * bs(rp)[1]) - 8; \ | | ------------------ | | | | 205| 60.6k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 275| 60.6k| pts[np][1][0] = pts[np][0][0] + (rp)->mv.mv[0].x; \ | | 276| 60.6k| pts[np][1][1] = pts[np][0][1] + (rp)->mv.mv[0].y; \ | | 277| 60.6k| np++; \ | | 278| 60.6k|} while (0) | | ------------------ | | | Branch (278:10): [Folded, False: 60.6k] | | ------------------ ------------------ 284| 203k| } else for (unsigned off = 0, xmask = (uint32_t) masks[0]; np < 8 && xmask;) { // top ------------------ | Branch (284:64): [True: 203k, False: 379] | Branch (284:74): [True: 94.5k, False: 108k] ------------------ 285| 94.5k| const int tz = ctz(xmask); 286| 94.5k| off += tz; 287| 94.5k| xmask >>= tz; 288| 94.5k| add_sample(off, 0, 1, -1, &r[-1][t->bx + off]); ------------------ | | 272| 94.5k|#define add_sample(dx, dy, sx, sy, rp) do { \ | | 273| 94.5k| pts[np][0][0] = 16 * (2 * dx + sx * bs(rp)[0]) - 8; \ | | ------------------ | | | | 205| 94.5k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 274| 94.5k| pts[np][0][1] = 16 * (2 * dy + sy * bs(rp)[1]) - 8; \ | | ------------------ | | | | 205| 94.5k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 275| 94.5k| pts[np][1][0] = pts[np][0][0] + (rp)->mv.mv[0].x; \ | | 276| 94.5k| pts[np][1][1] = pts[np][0][1] + (rp)->mv.mv[0].y; \ | | 277| 94.5k| np++; \ | | 278| 94.5k|} while (0) | | ------------------ | | | Branch (278:10): [Folded, False: 94.5k] | | ------------------ ------------------ 289| 94.5k| xmask &= ~1; 290| 94.5k| } 291| 169k| if (np < 8 && masks[1] == 1) { ------------------ | Branch (291:9): [True: 169k, False: 362] | Branch (291:19): [True: 59.9k, False: 109k] ------------------ 292| 59.9k| const int off = t->by & (bs(&r[0][t->bx - 1])[1] - 1); ------------------ | | 205| 59.9k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] ------------------ 293| 59.9k| add_sample(0, -off, -1, 1, &r[-off][t->bx - 1]); ------------------ | | 272| 59.9k|#define add_sample(dx, dy, sx, sy, rp) do { \ | | 273| 59.9k| pts[np][0][0] = 16 * (2 * dx + sx * bs(rp)[0]) - 8; \ | | ------------------ | | | | 205| 59.9k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 274| 59.9k| pts[np][0][1] = 16 * (2 * dy + sy * bs(rp)[1]) - 8; \ | | ------------------ | | | | 205| 59.9k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 275| 59.9k| pts[np][1][0] = pts[np][0][0] + (rp)->mv.mv[0].x; \ | | 276| 59.9k| pts[np][1][1] = pts[np][0][1] + (rp)->mv.mv[0].y; \ | | 277| 59.9k| np++; \ | | 278| 59.9k|} while (0) | | ------------------ | | | Branch (278:10): [Folded, False: 59.9k] | | ------------------ ------------------ 294| 207k| } else for (unsigned off = 0, ymask = (uint32_t) masks[1]; np < 8 && ymask;) { // left ------------------ | Branch (294:64): [True: 207k, False: 831] | Branch (294:74): [True: 98.0k, False: 109k] ------------------ 295| 98.0k| const int tz = ctz(ymask); 296| 98.0k| off += tz; 297| 98.0k| ymask >>= tz; 298| 98.0k| add_sample(0, off, -1, 1, &r[off][t->bx - 1]); ------------------ | | 272| 98.0k|#define add_sample(dx, dy, sx, sy, rp) do { \ | | 273| 98.0k| pts[np][0][0] = 16 * (2 * dx + sx * bs(rp)[0]) - 8; \ | | ------------------ | | | | 205| 98.0k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 274| 98.0k| pts[np][0][1] = 16 * (2 * dy + sy * bs(rp)[1]) - 8; \ | | ------------------ | | | | 205| 98.0k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 275| 98.0k| pts[np][1][0] = pts[np][0][0] + (rp)->mv.mv[0].x; \ | | 276| 98.0k| pts[np][1][1] = pts[np][0][1] + (rp)->mv.mv[0].y; \ | | 277| 98.0k| np++; \ | | 278| 98.0k|} while (0) | | ------------------ | | | Branch (278:10): [Folded, False: 98.0k] | | ------------------ ------------------ 299| 98.0k| ymask &= ~1; 300| 98.0k| } 301| 169k| if (np < 8 && masks[1] >> 32) // top/left ------------------ | Branch (301:9): [True: 168k, False: 974] | Branch (301:19): [True: 75.6k, False: 93.2k] ------------------ 302| 75.6k| add_sample(0, 0, -1, -1, &r[-1][t->bx - 1]); ------------------ | | 272| 75.6k|#define add_sample(dx, dy, sx, sy, rp) do { \ | | 273| 75.6k| pts[np][0][0] = 16 * (2 * dx + sx * bs(rp)[0]) - 8; \ | | ------------------ | | | | 205| 75.6k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 274| 75.6k| pts[np][0][1] = 16 * (2 * dy + sy * bs(rp)[1]) - 8; \ | | ------------------ | | | | 205| 75.6k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 275| 75.6k| pts[np][1][0] = pts[np][0][0] + (rp)->mv.mv[0].x; \ | | 276| 75.6k| pts[np][1][1] = pts[np][0][1] + (rp)->mv.mv[0].y; \ | | 277| 75.6k| np++; \ | | 278| 75.6k|} while (0) | | ------------------ | | | Branch (278:10): [Folded, False: 75.6k] | | ------------------ ------------------ 303| 169k| if (np < 8 && masks[0] >> 32) // top/right ------------------ | Branch (303:9): [True: 168k, False: 1.20k] | Branch (303:19): [True: 46.6k, False: 122k] ------------------ 304| 46.6k| add_sample(bw4, 0, 1, -1, &r[-1][t->bx + bw4]); ------------------ | | 272| 46.6k|#define add_sample(dx, dy, sx, sy, rp) do { \ | | 273| 46.6k| pts[np][0][0] = 16 * (2 * dx + sx * bs(rp)[0]) - 8; \ | | ------------------ | | | | 205| 46.6k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 274| 46.6k| pts[np][0][1] = 16 * (2 * dy + sy * bs(rp)[1]) - 8; \ | | ------------------ | | | | 205| 46.6k|#define bs(rp) dav1d_block_dimensions[(rp)->bs] | | ------------------ | | 275| 46.6k| pts[np][1][0] = pts[np][0][0] + (rp)->mv.mv[0].x; \ | | 276| 46.6k| pts[np][1][1] = pts[np][0][1] + (rp)->mv.mv[0].y; \ | | 277| 46.6k| np++; \ | | 278| 46.6k|} while (0) | | ------------------ | | | Branch (278:10): [Folded, False: 46.6k] | | ------------------ ------------------ 305| 169k| assert(np > 0 && np <= 8); ------------------ | Branch (305:5): [True: 169k, False: 204] | Branch (305:5): [True: 169k, False: 0] ------------------ 306| 169k|#undef bs 307| | 308| | // select according to motion vector difference against a threshold 309| 169k| int mvd[8], ret = 0; 310| 169k| const int thresh = 4 * iclip(imax(bw4, bh4), 4, 28); 311| 604k| for (int i = 0; i < np; i++) { ------------------ | Branch (311:21): [True: 435k, False: 169k] ------------------ 312| 435k| mvd[i] = abs(pts[i][1][0] - pts[i][0][0] - mv.x) + 313| 435k| abs(pts[i][1][1] - pts[i][0][1] - mv.y); 314| 435k| if (mvd[i] > thresh) ------------------ | Branch (314:13): [True: 145k, False: 289k] ------------------ 315| 145k| mvd[i] = -1; 316| 289k| else 317| 289k| ret++; 318| 435k| } 319| 169k| if (!ret) { ------------------ | Branch (319:9): [True: 33.9k, False: 135k] ------------------ 320| 33.9k| ret = 1; 321| 158k| } else for (int i = 0, j = np - 1, k = 0; k < np - ret; k++, i++, j--) { ------------------ | Branch (321:47): [True: 58.5k, False: 100k] ------------------ 322| 100k| while (mvd[i] != -1) i++; ------------------ | Branch (322:16): [True: 41.7k, False: 58.5k] ------------------ 323| 109k| while (mvd[j] == -1) j--; ------------------ | Branch (323:16): [True: 50.8k, False: 58.5k] ------------------ 324| 58.5k| assert(i != j); ------------------ | Branch (324:9): [True: 58.5k, False: 0] ------------------ 325| 58.5k| if (i > j) break; ------------------ | Branch (325:13): [True: 35.6k, False: 22.8k] ------------------ 326| | // replace the discarded samples; 327| 22.8k| mvd[i] = mvd[j]; 328| 22.8k| memcpy(pts[i], pts[j], sizeof(*pts)); 329| 22.8k| } 330| | 331| 169k| if (!dav1d_find_affine_int(pts, ret, bw4, bh4, mv, wmp, t->bx, t->by) && ------------------ | Branch (331:9): [True: 161k, False: 8.30k] ------------------ 332| 161k| !dav1d_get_shear_params(wmp)) ------------------ | Branch (332:9): [True: 153k, False: 7.62k] ------------------ 333| 153k| { 334| 153k| wmp->type = DAV1D_WM_TYPE_AFFINE; 335| 153k| } else 336| 15.9k| wmp->type = DAV1D_WM_TYPE_IDENTITY; 337| 169k|} decode.c:splat_tworef_mv: 550| 398k|{ 551| 398k| assert(bw4 >= 2 && bh4 >= 2); ------------------ | Branch (551:5): [True: 398k, False: 52] | Branch (551:5): [True: 398k, False: 18.4E] ------------------ 552| 398k| const enum CompInterPredMode mode = b->inter_mode; 553| 398k| const refmvs_block ALIGN(tmpl, 16) = (refmvs_block) { 554| 398k| .ref.ref = { b->ref[0] + 1, b->ref[1] + 1 }, 555| 398k| .mv.mv = { b->mv[0], b->mv[1] }, 556| 398k| .bs = bs, 557| 398k| .mf = (mode == GLOBALMV_GLOBALMV) | !!((1 << mode) & (0xbc)) * 2, 558| 398k| }; 559| 398k| c->refmvs_dsp.splat_mv(&t->rt.r[(t->by & 31) + 5], &tmpl, t->bx, bw4, bh4); 560| 398k|} decode.c:splat_oneref_mv: 519| 3.54M|{ 520| 3.54M| const enum InterPredMode mode = b->inter_mode; 521| 3.54M| const refmvs_block ALIGN(tmpl, 16) = (refmvs_block) { 522| 3.54M| .ref.ref = { b->ref[0] + 1, b->interintra_type ? 0 : -1 }, ------------------ | Branch (522:37): [True: 113k, False: 3.42M] ------------------ 523| 3.54M| .mv.mv[0] = b->mv[0], 524| 3.54M| .bs = bs, 525| 3.54M| .mf = (mode == GLOBALMV && imin(bw4, bh4) >= 2) | ((mode == NEWMV) * 2), ------------------ | Branch (525:16): [True: 2.05M, False: 1.48M] | Branch (525:36): [True: 1.92M, False: 129k] ------------------ 526| 3.54M| }; 527| 3.54M| c->refmvs_dsp.splat_mv(&t->rt.r[(t->by & 31) + 5], &tmpl, t->bx, bw4, bh4); 528| 3.54M|} decode.c:affine_lowest_px_luma: 619| 936k|{ 620| 936k| affine_lowest_px(t, dst, b_dim, wmp, 0, 0); 621| 936k|} decode.c:affine_lowest_px: 597| 1.00M|{ 598| 1.00M| const int h_mul = 4 >> ss_hor, v_mul = 4 >> ss_ver; 599| 1.00M| assert(!((b_dim[0] * h_mul) & 7) && !((b_dim[1] * v_mul) & 7)); ------------------ | Branch (599:5): [True: 1.00M, False: 18.4E] | Branch (599:5): [True: 1.00M, False: 6] ------------------ 600| 1.00M| const int32_t *const mat = wmp->matrix; 601| 1.00M| const int y = b_dim[1] * v_mul - 8; // lowest y 602| | 603| 1.00M| const int src_y = t->by * 4 + ((y + 4) << ss_ver); 604| 1.00M| const int64_t mat5_y = (int64_t) mat[5] * src_y + mat[1]; 605| | // check left- and right-most blocks 606| 2.93M| for (int x = 0; x < b_dim[0] * h_mul; x += imax(8, b_dim[0] * h_mul - 8)) { ------------------ | Branch (606:21): [True: 1.92M, False: 1.00M] ------------------ 607| | // calculate transformation relative to center of 8x8 block in 608| | // luma pixel units 609| 1.92M| const int src_x = t->bx * 4 + ((x + 4) << ss_hor); 610| 1.92M| const int64_t mvy = ((int64_t) mat[4] * src_x + mat5_y) >> ss_ver; 611| 1.92M| const int dy = (int) (mvy >> 16) - 4; 612| 1.92M| *dst = imax(*dst, dy + 4 + 8); 613| 1.92M| } 614| 1.00M|} decode.c:mc_lowest_px: 579| 6.23M|{ 580| 6.23M| const int v_mul = 4 >> ss_ver; 581| 6.23M| if (!smp->scale) { ------------------ | Branch (581:9): [True: 4.43M, False: 1.80M] ------------------ 582| 4.43M| const int my = mvy >> (3 + ss_ver), dy = mvy & (15 >> !ss_ver); 583| 4.43M| *dst = imax(*dst, (by4 + bh4) * v_mul + my + 4 * !!dy); 584| 4.43M| } else { 585| 1.80M| int y = (by4 * v_mul << 4) + mvy * (1 << !ss_ver); 586| 1.80M| const int64_t tmp = (int64_t)(y) * smp->scale + (smp->scale - 0x4000) * 8; 587| 1.80M| y = apply_sign64((int)((llabs(tmp) + 128) >> 8), tmp) + 32; 588| 1.80M| const int bottom = ((y + (bh4 * v_mul - 1) * smp->step) >> 10) + 1 + 4; 589| 1.80M| *dst = imax(*dst, bottom); 590| 1.80M| } 591| 6.23M|} decode.c:obmc_lowest_px: 639| 731k|{ 640| 731k| assert(!(t->bx & 1) && !(t->by & 1)); ------------------ | Branch (640:5): [True: 731k, False: 18] | Branch (640:5): [True: 731k, False: 18.4E] ------------------ 641| 731k| const Dav1dFrameContext *const f = t->f; 642| 731k| /*const*/ refmvs_block **r = &t->rt.r[(t->by & 31) + 5]; 643| 731k| const int ss_ver = is_chroma && f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (643:24): [True: 310k, False: 421k] | Branch (643:37): [True: 150k, False: 159k] ------------------ 644| 731k| const int ss_hor = is_chroma && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (644:24): [True: 310k, False: 421k] | Branch (644:37): [True: 150k, False: 159k] ------------------ 645| 731k| const int h_mul = 4 >> ss_hor, v_mul = 4 >> ss_ver; 646| | 647| 731k| if (t->by > t->ts->tiling.row_start && ------------------ | Branch (647:9): [True: 624k, False: 106k] ------------------ 648| 624k| (!is_chroma || b_dim[0] * h_mul + b_dim[1] * v_mul >= 16)) ------------------ | Branch (648:10): [True: 361k, False: 263k] | Branch (648:24): [True: 158k, False: 104k] ------------------ 649| 520k| { 650| 1.09M| for (int i = 0, x = 0; x < w4 && i < imin(b_dim[2], 4); ) { ------------------ | Branch (650:32): [True: 578k, False: 518k] | Branch (650:42): [True: 575k, False: 2.21k] ------------------ 651| | // only odd blocks are considered for overlap handling, hence +1 652| 575k| const refmvs_block *const a_r = &r[-1][t->bx + x + 1]; 653| 575k| const uint8_t *const a_b_dim = dav1d_block_dimensions[a_r->bs]; 654| | 655| 575k| if (a_r->ref.ref[0] > 0) { ------------------ | Branch (655:17): [True: 548k, False: 27.4k] ------------------ 656| 548k| const int oh4 = imin(b_dim[1], 16) >> 1; 657| 548k| mc_lowest_px(&dst[a_r->ref.ref[0] - 1][is_chroma], t->by, 658| 548k| (oh4 * 3 + 3) >> 2, a_r->mv.mv[0].y, ss_ver, 659| 548k| &f->svc[a_r->ref.ref[0] - 1][1]); 660| 548k| i++; 661| 548k| } 662| 575k| x += imax(a_b_dim[0], 2); 663| 575k| } 664| 520k| } 665| | 666| 731k| if (t->bx > t->ts->tiling.col_start) ------------------ | Branch (666:9): [True: 710k, False: 20.6k] ------------------ 667| 1.49M| for (int i = 0, y = 0; y < h4 && i < imin(b_dim[3], 4); ) { ------------------ | Branch (667:32): [True: 790k, False: 709k] | Branch (667:42): [True: 788k, False: 1.58k] ------------------ 668| | // only odd blocks are considered for overlap handling, hence +1 669| 788k| const refmvs_block *const l_r = &r[y + 1][t->bx - 1]; 670| 788k| const uint8_t *const l_b_dim = dav1d_block_dimensions[l_r->bs]; 671| | 672| 788k| if (l_r->ref.ref[0] > 0) { ------------------ | Branch (672:17): [True: 749k, False: 39.2k] ------------------ 673| 749k| const int oh4 = iclip(l_b_dim[1], 2, b_dim[1]); 674| 749k| mc_lowest_px(&dst[l_r->ref.ref[0] - 1][is_chroma], 675| 749k| t->by + y, oh4, l_r->mv.mv[0].y, ss_ver, 676| 749k| &f->svc[l_r->ref.ref[0] - 1][1]); 677| 749k| i++; 678| 749k| } 679| 788k| y += imax(l_b_dim[1], 2); 680| 788k| } 681| 731k|} decode.c:affine_lowest_px_chroma: 626| 89.5k|{ 627| 89.5k| const Dav1dFrameContext *const f = t->f; 628| 89.5k| assert(f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400); ------------------ | Branch (628:5): [True: 89.5k, False: 1] ------------------ 629| 89.5k| if (f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I444) ------------------ | Branch (629:9): [True: 17.4k, False: 72.0k] ------------------ 630| 17.4k| affine_lowest_px_luma(t, dst, b_dim, wmp); 631| 72.0k| else 632| 72.0k| affine_lowest_px(t, dst, b_dim, wmp, f->cur.p.layout & DAV1D_PIXEL_LAYOUT_I420, 1); 633| 89.5k|} decode.c:read_restoration_info: 2516| 260k|{ 2517| 260k| const Dav1dFrameContext *const f = t->f; 2518| 260k| Dav1dTileState *const ts = t->ts; 2519| 260k| const Av1RestorationUnit *const lr_ref = ts->lr_ref[p]; 2520| | 2521| 260k| if (frame_type == DAV1D_RESTORATION_SWITCHABLE) { ------------------ | Branch (2521:9): [True: 135k, False: 125k] ------------------ 2522| 135k| const int filter = dav1d_msac_decode_symbol_adapt4(&ts->msac, ------------------ | | 47| 135k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 2523| 135k| ts->cdf.m.restore_switchable, 2); 2524| 135k| lr->type = filter + !!filter; /* NONE/WIENER/SGRPROJ */ 2525| 135k| } else { 2526| 125k| const unsigned type = 2527| 125k| dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 125k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 2528| 125k| frame_type == DAV1D_RESTORATION_WIENER ? ------------------ | Branch (2528:17): [True: 41.3k, False: 83.7k] ------------------ 2529| 83.7k| ts->cdf.m.restore_wiener : ts->cdf.m.restore_sgrproj); 2530| 125k| lr->type = type ? frame_type : DAV1D_RESTORATION_NONE; ------------------ | Branch (2530:20): [True: 53.3k, False: 71.7k] ------------------ 2531| 125k| } 2532| | 2533| 260k| if (lr->type == DAV1D_RESTORATION_WIENER) { ------------------ | Branch (2533:9): [True: 33.1k, False: 227k] ------------------ 2534| 33.1k| lr->filter_v[0] = p ? 0 : ------------------ | Branch (2534:27): [True: 19.3k, False: 13.7k] ------------------ 2535| 33.1k| dav1d_msac_decode_subexp(&ts->msac, 2536| 13.7k| lr_ref->filter_v[0] + 5, 16, 1) - 5; 2537| 33.1k| lr->filter_v[1] = 2538| 33.1k| dav1d_msac_decode_subexp(&ts->msac, 2539| 33.1k| lr_ref->filter_v[1] + 23, 32, 2) - 23; 2540| 33.1k| lr->filter_v[2] = 2541| 33.1k| dav1d_msac_decode_subexp(&ts->msac, 2542| 33.1k| lr_ref->filter_v[2] + 17, 64, 3) - 17; 2543| | 2544| 33.1k| lr->filter_h[0] = p ? 0 : ------------------ | Branch (2544:27): [True: 19.3k, False: 13.7k] ------------------ 2545| 33.1k| dav1d_msac_decode_subexp(&ts->msac, 2546| 13.7k| lr_ref->filter_h[0] + 5, 16, 1) - 5; 2547| 33.1k| lr->filter_h[1] = 2548| 33.1k| dav1d_msac_decode_subexp(&ts->msac, 2549| 33.1k| lr_ref->filter_h[1] + 23, 32, 2) - 23; 2550| 33.1k| lr->filter_h[2] = 2551| 33.1k| dav1d_msac_decode_subexp(&ts->msac, 2552| 33.1k| lr_ref->filter_h[2] + 17, 64, 3) - 17; 2553| 33.1k| memcpy(lr->sgr_weights, lr_ref->sgr_weights, sizeof(lr->sgr_weights)); 2554| 33.1k| ts->lr_ref[p] = lr; 2555| 33.1k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 33.1k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 33.1k] | | ------------------ | | 35| 33.1k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 33.1k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2556| 0| printf("Post-lr_wiener[pl=%d,v[%d,%d,%d],h[%d,%d,%d]]: r=%d\n", 2557| 0| p, lr->filter_v[0], lr->filter_v[1], 2558| 0| lr->filter_v[2], lr->filter_h[0], 2559| 0| lr->filter_h[1], lr->filter_h[2], ts->msac.rng); 2560| 227k| } else if (lr->type == DAV1D_RESTORATION_SGRPROJ) { ------------------ | Branch (2560:16): [True: 51.7k, False: 175k] ------------------ 2561| 51.7k| const unsigned idx = dav1d_msac_decode_bools(&ts->msac, 4); 2562| 51.7k| const uint16_t *const sgr_params = dav1d_sgr_params[idx]; 2563| 51.7k| lr->type += idx; 2564| 51.7k| lr->sgr_weights[0] = sgr_params[0] ? dav1d_msac_decode_subexp(&ts->msac, ------------------ | Branch (2564:30): [True: 42.5k, False: 9.16k] ------------------ 2565| 42.5k| lr_ref->sgr_weights[0] + 96, 128, 4) - 96 : 0; 2566| 51.7k| lr->sgr_weights[1] = sgr_params[1] ? dav1d_msac_decode_subexp(&ts->msac, ------------------ | Branch (2566:30): [True: 29.0k, False: 22.7k] ------------------ 2567| 29.0k| lr_ref->sgr_weights[1] + 32, 128, 4) - 32 : 95; 2568| 51.7k| memcpy(lr->filter_v, lr_ref->filter_v, sizeof(lr->filter_v)); 2569| 51.7k| memcpy(lr->filter_h, lr_ref->filter_h, sizeof(lr->filter_h)); 2570| 51.7k| ts->lr_ref[p] = lr; 2571| 51.7k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 51.7k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 51.7k] | | ------------------ | | 35| 51.7k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 51.7k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2572| 0| printf("Post-lr_sgrproj[pl=%d,idx=%d,w[%d,%d]]: r=%d\n", 2573| 0| p, idx, lr->sgr_weights[0], 2574| 0| lr->sgr_weights[1], ts->msac.rng); 2575| 51.7k| } 2576| 260k|} decode.c:init_quant_tables: 57| 375k|{ 58| 941k| for (int i = 0; i < (frame_hdr->segmentation.enabled ? 8 : 1); i++) { ------------------ | Branch (58:21): [True: 565k, False: 375k] | Branch (58:26): [True: 244k, False: 697k] ------------------ 59| 565k| const int yac = frame_hdr->segmentation.enabled ? ------------------ | Branch (59:25): [True: 217k, False: 348k] ------------------ 60| 348k| iclip_u8(qidx + frame_hdr->segmentation.seg_data.d[i].delta_q) : qidx; 61| 565k| const int ydc = iclip_u8(yac + frame_hdr->quant.ydc_delta); 62| 565k| const int uac = iclip_u8(yac + frame_hdr->quant.uac_delta); 63| 565k| const int udc = iclip_u8(yac + frame_hdr->quant.udc_delta); 64| 565k| const int vac = iclip_u8(yac + frame_hdr->quant.vac_delta); 65| 565k| const int vdc = iclip_u8(yac + frame_hdr->quant.vdc_delta); 66| | 67| 565k| dq[i][0][0] = dav1d_dq_tbl[seq_hdr->hbd][ydc][0]; 68| 565k| dq[i][0][1] = dav1d_dq_tbl[seq_hdr->hbd][yac][1]; 69| 565k| dq[i][1][0] = dav1d_dq_tbl[seq_hdr->hbd][udc][0]; 70| 565k| dq[i][1][1] = dav1d_dq_tbl[seq_hdr->hbd][uac][1]; 71| 565k| dq[i][2][0] = dav1d_dq_tbl[seq_hdr->hbd][vdc][0]; 72| 565k| dq[i][2][1] = dav1d_dq_tbl[seq_hdr->hbd][vac][1]; 73| 565k| } 74| 375k|} decode.c:setup_tile: 2432| 315k|{ 2433| 315k| const int col_sb_start = f->frame_hdr->tiling.col_start_sb[tile_col]; 2434| 315k| const int col_sb128_start = col_sb_start >> !f->seq_hdr->sb128; 2435| 315k| const int col_sb_end = f->frame_hdr->tiling.col_start_sb[tile_col + 1]; 2436| 315k| const int row_sb_start = f->frame_hdr->tiling.row_start_sb[tile_row]; 2437| 315k| const int row_sb_end = f->frame_hdr->tiling.row_start_sb[tile_row + 1]; 2438| 315k| const int sb_shift = f->sb_shift; 2439| | 2440| 315k| const uint8_t *const size_mul = ss_size_mul[f->cur.p.layout]; 2441| 946k| for (int p = 0; p < 2; p++) { ------------------ | Branch (2441:21): [True: 630k, False: 315k] ------------------ 2442| 630k| ts->frame_thread[p].pal_idx = f->frame_thread.pal_idx ? ------------------ | Branch (2442:39): [True: 543k, False: 86.7k] ------------------ 2443| 543k| &f->frame_thread.pal_idx[(size_t)tile_start_off * size_mul[1] / 8] : 2444| 630k| NULL; 2445| 630k| ts->frame_thread[p].cbi = f->frame_thread.cbi ? ------------------ | Branch (2445:35): [True: 630k, False: 62] ------------------ 2446| 630k| &f->frame_thread.cbi[(size_t)tile_start_off * size_mul[0] / 64] : 2447| 630k| NULL; 2448| 630k| ts->frame_thread[p].cf = f->frame_thread.cf ? ------------------ | Branch (2448:34): [True: 630k, False: 66] ------------------ 2449| 630k| (uint8_t*)f->frame_thread.cf + 2450| 630k| (((size_t)tile_start_off * size_mul[0]) >> !f->seq_hdr->hbd) : 2451| 630k| NULL; 2452| 630k| } 2453| | 2454| 315k| dav1d_cdf_thread_copy(&ts->cdf, &f->in_cdf); 2455| 315k| ts->last_qidx = f->frame_hdr->quant.yac; 2456| 315k| ts->last_delta_lf.u32 = 0; 2457| | 2458| 315k| dav1d_msac_init(&ts->msac, data, sz, f->frame_hdr->disable_cdf_update); 2459| | 2460| 315k| ts->tiling.row = tile_row; 2461| 315k| ts->tiling.col = tile_col; 2462| 315k| ts->tiling.col_start = col_sb_start << sb_shift; 2463| 315k| ts->tiling.col_end = imin(col_sb_end << sb_shift, f->bw); 2464| 315k| ts->tiling.row_start = row_sb_start << sb_shift; 2465| 315k| ts->tiling.row_end = imin(row_sb_end << sb_shift, f->bh); 2466| | 2467| | // Reference Restoration Unit (used for exp coding) 2468| 315k| int sb_idx, unit_idx; 2469| 315k| if (f->frame_hdr->width[0] != f->frame_hdr->width[1]) { ------------------ | Branch (2469:9): [True: 20.3k, False: 295k] ------------------ 2470| | // vertical components only 2471| 20.3k| sb_idx = (ts->tiling.row_start >> 5) * f->sr_sb128w; 2472| 20.3k| unit_idx = (ts->tiling.row_start & 16) >> 3; 2473| 295k| } else { 2474| 295k| sb_idx = (ts->tiling.row_start >> 5) * f->sb128w + col_sb128_start; 2475| 295k| unit_idx = ((ts->tiling.row_start & 16) >> 3) + 2476| 295k| ((ts->tiling.col_start & 16) >> 4); 2477| 295k| } 2478| 1.26M| for (int p = 0; p < 3; p++) { ------------------ | Branch (2478:21): [True: 945k, False: 315k] ------------------ 2479| 945k| if (!((f->lf.restore_planes >> p) & 1U)) ------------------ | Branch (2479:13): [True: 890k, False: 54.8k] ------------------ 2480| 890k| continue; 2481| | 2482| 54.8k| if (f->frame_hdr->width[0] != f->frame_hdr->width[1]) { ------------------ | Branch (2482:13): [True: 15.9k, False: 38.9k] ------------------ 2483| 15.9k| const int ss_hor = p && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (2483:32): [True: 10.4k, False: 5.47k] | Branch (2483:37): [True: 7.30k, False: 3.14k] ------------------ 2484| 15.9k| const int d = f->frame_hdr->super_res.width_scale_denominator; 2485| 15.9k| const int unit_size_log2 = f->frame_hdr->restoration.unit_size[!!p]; 2486| 15.9k| const int rnd = (8 << unit_size_log2) - 1, shift = unit_size_log2 + 3; 2487| 15.9k| const int x = ((4 * ts->tiling.col_start * d >> ss_hor) + rnd) >> shift; 2488| 15.9k| const int px_x = x << (unit_size_log2 + ss_hor); 2489| 15.9k| const int u_idx = unit_idx + ((px_x & 64) >> 6); 2490| 15.9k| const int sb128x = px_x >> 7; 2491| 15.9k| if (sb128x >= f->sr_sb128w) continue; ------------------ | Branch (2491:17): [True: 321, False: 15.6k] ------------------ 2492| 15.6k| ts->lr_ref[p] = &f->lf.lr_mask[sb_idx + sb128x].lr[p][u_idx]; 2493| 38.9k| } else { 2494| 38.9k| ts->lr_ref[p] = &f->lf.lr_mask[sb_idx].lr[p][unit_idx]; 2495| 38.9k| } 2496| | 2497| 54.5k| ts->lr_ref[p]->filter_v[0] = 3; 2498| 54.5k| ts->lr_ref[p]->filter_v[1] = -7; 2499| 54.5k| ts->lr_ref[p]->filter_v[2] = 15; 2500| 54.5k| ts->lr_ref[p]->filter_h[0] = 3; 2501| 54.5k| ts->lr_ref[p]->filter_h[1] = -7; 2502| 54.5k| ts->lr_ref[p]->filter_h[2] = 15; 2503| 54.5k| ts->lr_ref[p]->sgr_weights[0] = -32; 2504| 54.5k| ts->lr_ref[p]->sgr_weights[1] = 31; 2505| 54.5k| } 2506| | 2507| 315k| if (f->c->n_tc > 1) { ------------------ | Branch (2507:9): [True: 315k, False: 361] ------------------ 2508| 945k| for (int p = 0; p < 2; p++) ------------------ | Branch (2508:25): [True: 630k, False: 315k] ------------------ 2509| 630k| atomic_init(&ts->progress[p], row_sb_start); 2510| 315k| } 2511| 315k|} decode.c:get_upscale_x0: 3324| 78.2k|static int get_upscale_x0(const int in_w, const int out_w, const int step) { 3325| 78.2k| const int err = out_w * step - (in_w << 14); 3326| 78.2k| const int x0 = (-((out_w - in_w) << 13) + (out_w >> 1)) / out_w + 128 - (err / 2); 3327| 78.2k| return x0 & 0x3fff; 3328| 78.2k|} obu.c:get_poc_diff: 239| 1.23M|{ 240| 1.23M| if (!order_hint_n_bits) return 0; ------------------ | Branch (240:9): [True: 0, False: 1.23M] ------------------ 241| 1.23M| const int mask = 1 << (order_hint_n_bits - 1); 242| 1.23M| const int diff = poc0 - poc1; 243| 1.23M| return (diff & (mask - 1)) - (diff & mask); 244| 1.23M|} refmvs.c:get_gmv_2d: 482| 4.33M|{ 483| 4.33M| switch (gmv->type) { 484| 1.09M| case DAV1D_WM_TYPE_ROT_ZOOM: ------------------ | Branch (484:5): [True: 1.09M, False: 3.23M] ------------------ 485| 1.09M| assert(gmv->matrix[5] == gmv->matrix[2]); ------------------ | Branch (485:9): [True: 1.09M, False: 18.4E] ------------------ 486| 1.09M| assert(gmv->matrix[4] == -gmv->matrix[3]); ------------------ | Branch (486:9): [True: 1.09M, False: 18.4E] ------------------ 487| | // fall-through 488| 1.09M| default: ------------------ | Branch (488:5): [True: 0, False: 4.33M] ------------------ 489| 1.38M| case DAV1D_WM_TYPE_AFFINE: { ------------------ | Branch (489:5): [True: 289k, False: 4.04M] ------------------ 490| 1.38M| const int x = bx4 * 4 + bw4 * 2 - 1; 491| 1.38M| const int y = by4 * 4 + bh4 * 2 - 1; 492| 1.38M| const int xc = (gmv->matrix[2] - (1 << 16)) * x + 493| 1.38M| gmv->matrix[3] * y + gmv->matrix[0]; 494| 1.38M| const int yc = (gmv->matrix[5] - (1 << 16)) * y + 495| 1.38M| gmv->matrix[4] * x + gmv->matrix[1]; 496| 1.38M| const int shift = 16 - (3 - !hdr->hp); 497| 1.38M| const int round = (1 << shift) >> 1; 498| 1.38M| mv res = (mv) { 499| 1.38M| .y = apply_sign(((abs(yc) + round) >> shift) << !hdr->hp, yc), 500| 1.38M| .x = apply_sign(((abs(xc) + round) >> shift) << !hdr->hp, xc), 501| 1.38M| }; 502| 1.38M| if (hdr->force_integer_mv) ------------------ | Branch (502:13): [True: 29.9k, False: 1.35M] ------------------ 503| 29.9k| fix_int_mv_precision(&res); 504| 1.38M| return res; 505| 1.09M| } 506| 88.6k| case DAV1D_WM_TYPE_TRANSLATION: { ------------------ | Branch (506:5): [True: 88.6k, False: 4.24M] ------------------ 507| 88.6k| mv res = (mv) { 508| 88.6k| .y = gmv->matrix[0] >> 13, 509| 88.6k| .x = gmv->matrix[1] >> 13, 510| 88.6k| }; 511| 88.6k| if (hdr->force_integer_mv) ------------------ | Branch (511:13): [True: 1.29k, False: 87.3k] ------------------ 512| 1.29k| fix_int_mv_precision(&res); 513| 88.6k| return res; 514| 1.09M| } 515| 2.86M| case DAV1D_WM_TYPE_IDENTITY: ------------------ | Branch (515:5): [True: 2.86M, False: 1.47M] ------------------ 516| 2.86M| return (mv) { .x = 0, .y = 0 }; 517| 4.33M| } 518| 4.33M|} refmvs.c:fix_int_mv_precision: 462| 37.3k|static inline void fix_int_mv_precision(mv *const mv) { 463| 37.3k| mv->x = (mv->x - (mv->x >> 15) + 3) & ~7U; 464| 37.3k| mv->y = (mv->y - (mv->y >> 15) + 3) & ~7U; 465| 37.3k|} refmvs.c:fix_mv_precision: 469| 2.92M|{ 470| 2.92M| if (hdr->force_integer_mv) { ------------------ | Branch (470:9): [True: 6.10k, False: 2.92M] ------------------ 471| 6.10k| fix_int_mv_precision(mv); 472| 2.92M| } else if (!hdr->hp) { ------------------ | Branch (472:16): [True: 301k, False: 2.62M] ------------------ 473| 301k| mv->x = (mv->x - (mv->x >> 15)) & ~1U; 474| 301k| mv->y = (mv->y - (mv->y >> 15)) & ~1U; 475| 301k| } 476| 2.92M|} refmvs.c:get_poc_diff: 239| 5.89M|{ 240| 5.89M| if (!order_hint_n_bits) return 0; ------------------ | Branch (240:9): [True: 3.22M, False: 2.67M] ------------------ 241| 2.67M| const int mask = 1 << (order_hint_n_bits - 1); 242| 2.67M| const int diff = poc0 - poc1; 243| 2.67M| return (diff & (mask - 1)) - (diff & mask); 244| 5.89M|} decode.c:get_partition_ctx: 87| 8.89M|{ 88| 8.89M| return ((a->partition[xb8] >> (4 - bl)) & 1) + 89| 8.89M| (((l->partition[yb8] >> (4 - bl)) & 1) << 1); 90| 8.89M|} decode.c:get_cur_frame_segid: 445| 2.46M|{ 446| 2.46M| cur_seg_map += bx + by * stride; 447| 2.46M| if (have_left && have_top) { ------------------ | Branch (447:9): [True: 1.29M, False: 1.16M] | Branch (447:22): [True: 1.10M, False: 193k] ------------------ 448| 1.10M| const int l = cur_seg_map[-1]; 449| 1.10M| const int a = cur_seg_map[-stride]; 450| 1.10M| const int al = cur_seg_map[-(stride + 1)]; 451| | 452| 1.10M| if (l == a && al == l) *seg_ctx = 2; ------------------ | Branch (452:13): [True: 804k, False: 298k] | Branch (452:23): [True: 786k, False: 17.9k] ------------------ 453| 316k| else if (l == a || al == l || a == al) *seg_ctx = 1; ------------------ | Branch (453:18): [True: 16.9k, False: 299k] | Branch (453:28): [True: 130k, False: 169k] | Branch (453:39): [True: 108k, False: 61.3k] ------------------ 454| 60.0k| else *seg_ctx = 0; 455| 1.10M| return a == al ? a : l; ------------------ | Branch (455:16): [True: 894k, False: 208k] ------------------ 456| 1.35M| } else { 457| 1.35M| *seg_ctx = 0; 458| 1.35M| return have_left ? cur_seg_map[-1] : have_top ? cur_seg_map[-stride] : 0; ------------------ | Branch (458:16): [True: 195k, False: 1.16M] | Branch (458:46): [True: 1.15M, False: 8.82k] ------------------ 459| 1.35M| } 460| 2.46M|} decode.c:get_intra_ctx: 63| 2.37M|{ 64| 2.37M| if (have_left) { ------------------ | Branch (64:9): [True: 2.27M, False: 99.0k] ------------------ 65| 2.27M| if (have_top) { ------------------ | Branch (65:13): [True: 2.02M, False: 252k] ------------------ 66| 2.02M| const int ctx = l->intra[yb4] + a->intra[xb4]; 67| 2.02M| return ctx + (ctx == 2); 68| 2.02M| } else 69| 252k| return l->intra[yb4] * 2; 70| 2.27M| } else { 71| 99.0k| return have_top ? a->intra[xb4] * 2 : 0; ------------------ | Branch (71:16): [True: 79.6k, False: 19.4k] ------------------ 72| 99.0k| } 73| 2.37M|} decode.c:get_tx_ctx: 79| 933k|{ 80| 933k| return (l->tx_intra[yb4] >= max_tx->lh) + (a->tx_intra[xb4] >= max_tx->lw); 81| 933k|} decode.c:get_comp_ctx: 160| 786k|{ 161| 786k| if (have_top) { ------------------ | Branch (161:9): [True: 648k, False: 138k] ------------------ 162| 648k| if (have_left) { ------------------ | Branch (162:13): [True: 613k, False: 35.2k] ------------------ 163| 613k| if (a->comp_type[xb4]) { ------------------ | Branch (163:17): [True: 255k, False: 358k] ------------------ 164| 255k| if (l->comp_type[yb4]) { ------------------ | Branch (164:21): [True: 164k, False: 91.0k] ------------------ 165| 164k| return 4; 166| 164k| } else { 167| | // 4U means intra (-1) or bwd (>= 4) 168| 91.0k| return 2 + ((unsigned)l->ref[0][yb4] >= 4U); 169| 91.0k| } 170| 358k| } else if (l->comp_type[yb4]) { ------------------ | Branch (170:24): [True: 80.4k, False: 277k] ------------------ 171| | // 4U means intra (-1) or bwd (>= 4) 172| 80.4k| return 2 + ((unsigned)a->ref[0][xb4] >= 4U); 173| 277k| } else { 174| 277k| return (l->ref[0][yb4] >= 4) ^ (a->ref[0][xb4] >= 4); 175| 277k| } 176| 613k| } else { 177| 35.2k| return a->comp_type[xb4] ? 3 : a->ref[0][xb4] >= 4; ------------------ | Branch (177:20): [True: 11.8k, False: 23.3k] ------------------ 178| 35.2k| } 179| 648k| } else if (have_left) { ------------------ | Branch (179:16): [True: 132k, False: 5.81k] ------------------ 180| 132k| return l->comp_type[yb4] ? 3 : l->ref[0][yb4] >= 4; ------------------ | Branch (180:16): [True: 70.0k, False: 62.3k] ------------------ 181| 132k| } else { 182| 5.81k| return 1; 183| 5.81k| } 184| 786k|} decode.c:fix_mv_precision: 469| 1.45M|{ 470| 1.45M| if (hdr->force_integer_mv) { ------------------ | Branch (470:9): [True: 16.1k, False: 1.43M] ------------------ 471| 16.1k| fix_int_mv_precision(mv); 472| 1.43M| } else if (!hdr->hp) { ------------------ | Branch (472:16): [True: 661k, False: 772k] ------------------ 473| 661k| mv->x = (mv->x - (mv->x >> 15)) & ~1U; 474| 661k| mv->y = (mv->y - (mv->y >> 15)) & ~1U; 475| 661k| } 476| 1.45M|} decode.c:fix_int_mv_precision: 462| 32.8k|static inline void fix_int_mv_precision(mv *const mv) { 463| 32.8k| mv->x = (mv->x - (mv->x >> 15) + 3) & ~7U; 464| 32.8k| mv->y = (mv->y - (mv->y >> 15) + 3) & ~7U; 465| 32.8k|} decode.c:get_comp_dir_ctx: 190| 372k|{ 191| 372k|#define has_uni_comp(edge, off) \ 192| 372k| ((edge->ref[0][off] < 4) == (edge->ref[1][off] < 4)) 193| | 194| 372k| if (have_top && have_left) { ------------------ | Branch (194:9): [True: 295k, False: 77.3k] | Branch (194:21): [True: 282k, False: 12.4k] ------------------ 195| 282k| const int a_intra = a->intra[xb4], l_intra = l->intra[yb4]; 196| | 197| 282k| if (a_intra && l_intra) return 2; ------------------ | Branch (197:13): [True: 12.2k, False: 270k] | Branch (197:24): [True: 4.77k, False: 7.43k] ------------------ 198| 278k| if (a_intra || l_intra) { ------------------ | Branch (198:13): [True: 7.42k, False: 270k] | Branch (198:24): [True: 9.66k, False: 261k] ------------------ 199| 17.1k| const BlockContext *const edge = a_intra ? l : a; ------------------ | Branch (199:46): [True: 7.43k, False: 9.68k] ------------------ 200| 17.1k| const int off = a_intra ? yb4 : xb4; ------------------ | Branch (200:29): [True: 7.43k, False: 9.68k] ------------------ 201| | 202| 17.1k| if (edge->comp_type[off] == COMP_INTER_NONE) return 2; ------------------ | Branch (202:17): [True: 5.99k, False: 11.1k] ------------------ 203| 11.1k| return 1 + 2 * has_uni_comp(edge, off); ------------------ | | 192| 11.1k| ((edge->ref[0][off] < 4) == (edge->ref[1][off] < 4)) ------------------ 204| 17.1k| } 205| | 206| 260k| const int a_comp = a->comp_type[xb4] != COMP_INTER_NONE; 207| 260k| const int l_comp = l->comp_type[yb4] != COMP_INTER_NONE; 208| 260k| const int a_ref0 = a->ref[0][xb4], l_ref0 = l->ref[0][yb4]; 209| | 210| 260k| if (!a_comp && !l_comp) { ------------------ | Branch (210:13): [True: 68.8k, False: 192k] | Branch (210:24): [True: 27.2k, False: 41.6k] ------------------ 211| 27.2k| return 1 + 2 * ((a_ref0 >= 4) == (l_ref0 >= 4)); 212| 233k| } else if (!a_comp || !l_comp) { ------------------ | Branch (212:20): [True: 41.5k, False: 192k] | Branch (212:31): [True: 48.0k, False: 144k] ------------------ 213| 89.7k| const BlockContext *const edge = a_comp ? a : l; ------------------ | Branch (213:46): [True: 48.1k, False: 41.6k] ------------------ 214| 89.7k| const int off = a_comp ? xb4 : yb4; ------------------ | Branch (214:29): [True: 48.1k, False: 41.6k] ------------------ 215| | 216| 89.7k| if (!has_uni_comp(edge, off)) return 1; ------------------ | | 192| 89.7k| ((edge->ref[0][off] < 4) == (edge->ref[1][off] < 4)) ------------------ | Branch (216:17): [True: 74.1k, False: 15.6k] ------------------ 217| 15.6k| return 3 + ((a_ref0 >= 4) == (l_ref0 >= 4)); 218| 143k| } else { 219| 143k| const int a_uni = has_uni_comp(a, xb4), l_uni = has_uni_comp(l, yb4); ------------------ | | 192| 143k| ((edge->ref[0][off] < 4) == (edge->ref[1][off] < 4)) ------------------ const int a_uni = has_uni_comp(a, xb4), l_uni = has_uni_comp(l, yb4); ------------------ | | 192| 143k| ((edge->ref[0][off] < 4) == (edge->ref[1][off] < 4)) ------------------ 220| | 221| 143k| if (!a_uni && !l_uni) return 0; ------------------ | Branch (221:17): [True: 115k, False: 28.3k] | Branch (221:27): [True: 106k, False: 9.08k] ------------------ 222| 37.3k| if (!a_uni || !l_uni) return 2; ------------------ | Branch (222:17): [True: 8.83k, False: 28.5k] | Branch (222:27): [True: 15.3k, False: 13.2k] ------------------ 223| 12.9k| return 3 + ((a_ref0 == 4) == (l_ref0 == 4)); 224| 37.3k| } 225| 260k| } else if (have_top || have_left) { ------------------ | Branch (225:16): [True: 12.2k, False: 77.5k] | Branch (225:28): [True: 75.6k, False: 1.87k] ------------------ 226| 88.1k| const BlockContext *const edge = have_left ? l : a; ------------------ | Branch (226:42): [True: 75.6k, False: 12.5k] ------------------ 227| 88.1k| const int off = have_left ? yb4 : xb4; ------------------ | Branch (227:25): [True: 75.6k, False: 12.5k] ------------------ 228| | 229| 88.1k| if (edge->intra[off]) return 2; ------------------ | Branch (229:13): [True: 2.08k, False: 86.0k] ------------------ 230| 86.0k| if (edge->comp_type[off] == COMP_INTER_NONE) return 2; ------------------ | Branch (230:13): [True: 17.8k, False: 68.1k] ------------------ 231| 68.1k| return 4 * has_uni_comp(edge, off); ------------------ | | 192| 68.1k| ((edge->ref[0][off] < 4) == (edge->ref[1][off] < 4)) ------------------ 232| 86.0k| } else { 233| 1.63k| return 2; 234| 1.63k| } 235| 372k|} decode.c:av1_get_fwd_ref_ctx: 307| 1.32M|{ 308| 1.32M| int cnt[4] = { 0 }; 309| | 310| 1.32M| if (have_top && !a->intra[xb4]) { ------------------ | Branch (310:9): [True: 1.19M, False: 129k] | Branch (310:21): [True: 1.10M, False: 96.5k] ------------------ 311| 1.10M| if (a->ref[0][xb4] < 4) cnt[a->ref[0][xb4]]++; ------------------ | Branch (311:13): [True: 1.02M, False: 78.1k] ------------------ 312| 1.10M| if (a->comp_type[xb4] && a->ref[1][xb4] < 4) cnt[a->ref[1][xb4]]++; ------------------ | Branch (312:13): [True: 236k, False: 866k] | Branch (312:34): [True: 31.9k, False: 204k] ------------------ 313| 1.10M| } 314| | 315| 1.32M| if (have_left && !l->intra[yb4]) { ------------------ | Branch (315:9): [True: 1.28M, False: 43.4k] | Branch (315:22): [True: 1.18M, False: 98.3k] ------------------ 316| 1.18M| if (l->ref[0][yb4] < 4) cnt[l->ref[0][yb4]]++; ------------------ | Branch (316:13): [True: 1.10M, False: 85.5k] ------------------ 317| 1.18M| if (l->comp_type[yb4] && l->ref[1][yb4] < 4) cnt[l->ref[1][yb4]]++; ------------------ | Branch (317:13): [True: 282k, False: 904k] | Branch (317:34): [True: 29.8k, False: 252k] ------------------ 318| 1.18M| } 319| | 320| 1.32M| cnt[0] += cnt[1]; 321| 1.32M| cnt[2] += cnt[3]; 322| | 323| 1.32M| return cnt[0] == cnt[2] ? 1 : cnt[0] < cnt[2] ? 0 : 2; ------------------ | Branch (323:12): [True: 174k, False: 1.15M] | Branch (323:35): [True: 183k, False: 971k] ------------------ 324| 1.32M|} decode.c:av1_get_fwd_ref_2_ctx: 350| 278k|{ 351| 278k| int cnt[2] = { 0 }; 352| | 353| 278k| if (have_top && !a->intra[xb4]) { ------------------ | Branch (353:9): [True: 221k, False: 56.9k] | Branch (353:21): [True: 199k, False: 22.2k] ------------------ 354| 199k| if ((a->ref[0][xb4] ^ 2U) < 2) cnt[a->ref[0][xb4] - 2]++; ------------------ | Branch (354:13): [True: 119k, False: 80.1k] ------------------ 355| 199k| if (a->comp_type[xb4] && (a->ref[1][xb4] ^ 2U) < 2) cnt[a->ref[1][xb4] - 2]++; ------------------ | Branch (355:13): [True: 78.1k, False: 121k] | Branch (355:34): [True: 14.0k, False: 64.0k] ------------------ 356| 199k| } 357| | 358| 278k| if (have_left && !l->intra[yb4]) { ------------------ | Branch (358:9): [True: 267k, False: 11.5k] | Branch (358:22): [True: 246k, False: 20.4k] ------------------ 359| 246k| if ((l->ref[0][yb4] ^ 2U) < 2) cnt[l->ref[0][yb4] - 2]++; ------------------ | Branch (359:13): [True: 152k, False: 94.2k] ------------------ 360| 246k| if (l->comp_type[yb4] && (l->ref[1][yb4] ^ 2U) < 2) cnt[l->ref[1][yb4] - 2]++; ------------------ | Branch (360:13): [True: 112k, False: 134k] | Branch (360:34): [True: 16.9k, False: 95.3k] ------------------ 361| 246k| } 362| | 363| 278k| return cnt[0] == cnt[1] ? 1 : cnt[0] < cnt[1] ? 0 : 2; ------------------ | Branch (363:12): [True: 73.9k, False: 204k] | Branch (363:35): [True: 150k, False: 54.5k] ------------------ 364| 278k|} decode.c:av1_get_fwd_ref_1_ctx: 330| 1.08M|{ 331| 1.08M| int cnt[2] = { 0 }; 332| | 333| 1.08M| if (have_top && !a->intra[xb4]) { ------------------ | Branch (333:9): [True: 1.00M, False: 83.4k] | Branch (333:21): [True: 925k, False: 75.9k] ------------------ 334| 925k| if (a->ref[0][xb4] < 2) cnt[a->ref[0][xb4]]++; ------------------ | Branch (334:13): [True: 825k, False: 99.1k] ------------------ 335| 925k| if (a->comp_type[xb4] && a->ref[1][xb4] < 2) cnt[a->ref[1][xb4]]++; ------------------ | Branch (335:13): [True: 173k, False: 751k] | Branch (335:34): [True: 12.5k, False: 160k] ------------------ 336| 925k| } 337| | 338| 1.08M| if (have_left && !l->intra[yb4]) { ------------------ | Branch (338:9): [True: 1.05M, False: 33.2k] | Branch (338:22): [True: 972k, False: 79.0k] ------------------ 339| 972k| if (l->ref[0][yb4] < 2) cnt[l->ref[0][yb4]]++; ------------------ | Branch (339:13): [True: 864k, False: 107k] ------------------ 340| 972k| if (l->comp_type[yb4] && l->ref[1][yb4] < 2) cnt[l->ref[1][yb4]]++; ------------------ | Branch (340:13): [True: 195k, False: 777k] | Branch (340:34): [True: 11.8k, False: 183k] ------------------ 341| 972k| } 342| | 343| 1.08M| return cnt[0] == cnt[1] ? 1 : cnt[0] < cnt[1] ? 0 : 2; ------------------ | Branch (343:12): [True: 132k, False: 952k] | Branch (343:35): [True: 56.0k, False: 896k] ------------------ 344| 1.08M|} decode.c:av1_get_bwd_ref_ctx: 370| 826k|{ 371| 826k| int cnt[3] = { 0 }; 372| | 373| 826k| if (have_top && !a->intra[xb4]) { ------------------ | Branch (373:9): [True: 667k, False: 159k] | Branch (373:21): [True: 624k, False: 42.9k] ------------------ 374| 624k| if (a->ref[0][xb4] >= 4) cnt[a->ref[0][xb4] - 4]++; ------------------ | Branch (374:13): [True: 342k, False: 282k] ------------------ 375| 624k| if (a->comp_type[xb4] && a->ref[1][xb4] >= 4) cnt[a->ref[1][xb4] - 4]++; ------------------ | Branch (375:13): [True: 210k, False: 413k] | Branch (375:34): [True: 195k, False: 15.7k] ------------------ 376| 624k| } 377| | 378| 826k| if (have_left && !l->intra[yb4]) { ------------------ | Branch (378:9): [True: 787k, False: 39.7k] | Branch (378:22): [True: 733k, False: 53.4k] ------------------ 379| 733k| if (l->ref[0][yb4] >= 4) cnt[l->ref[0][yb4] - 4]++; ------------------ | Branch (379:13): [True: 397k, False: 336k] ------------------ 380| 733k| if (l->comp_type[yb4] && l->ref[1][yb4] >= 4) cnt[l->ref[1][yb4] - 4]++; ------------------ | Branch (380:13): [True: 253k, False: 479k] | Branch (380:34): [True: 238k, False: 14.9k] ------------------ 381| 733k| } 382| | 383| 826k| cnt[1] += cnt[0]; 384| | 385| 826k| return cnt[2] == cnt[1] ? 1 : cnt[1] < cnt[2] ? 0 : 2; ------------------ | Branch (385:12): [True: 145k, False: 681k] | Branch (385:35): [True: 462k, False: 219k] ------------------ 386| 826k|} decode.c:av1_get_bwd_ref_1_ctx: 392| 285k|{ 393| 285k| int cnt[3] = { 0 }; 394| | 395| 285k| if (have_top && !a->intra[xb4]) { ------------------ | Branch (395:9): [True: 252k, False: 32.7k] | Branch (395:21): [True: 233k, False: 19.1k] ------------------ 396| 233k| if (a->ref[0][xb4] >= 4) cnt[a->ref[0][xb4] - 4]++; ------------------ | Branch (396:13): [True: 97.9k, False: 135k] ------------------ 397| 233k| if (a->comp_type[xb4] && a->ref[1][xb4] >= 4) cnt[a->ref[1][xb4] - 4]++; ------------------ | Branch (397:13): [True: 102k, False: 131k] | Branch (397:34): [True: 92.0k, False: 10.2k] ------------------ 398| 233k| } 399| | 400| 285k| if (have_left && !l->intra[yb4]) { ------------------ | Branch (400:9): [True: 275k, False: 9.79k] | Branch (400:22): [True: 253k, False: 22.3k] ------------------ 401| 253k| if (l->ref[0][yb4] >= 4) cnt[l->ref[0][yb4] - 4]++; ------------------ | Branch (401:13): [True: 105k, False: 148k] ------------------ 402| 253k| if (l->comp_type[yb4] && l->ref[1][yb4] >= 4) cnt[l->ref[1][yb4] - 4]++; ------------------ | Branch (402:13): [True: 112k, False: 141k] | Branch (402:34): [True: 102k, False: 9.80k] ------------------ 403| 253k| } 404| | 405| 285k| return cnt[0] == cnt[1] ? 1 : cnt[0] < cnt[1] ? 0 : 2; ------------------ | Branch (405:12): [True: 64.6k, False: 220k] | Branch (405:35): [True: 76.0k, False: 144k] ------------------ 406| 285k|} decode.c:av1_get_ref_ctx: 287| 1.62M|{ 288| 1.62M| int cnt[2] = { 0 }; 289| | 290| 1.62M| if (have_top && !a->intra[xb4]) { ------------------ | Branch (290:9): [True: 1.43M, False: 183k] | Branch (290:21): [True: 1.31M, False: 124k] ------------------ 291| 1.31M| cnt[a->ref[0][xb4] >= 4]++; 292| 1.31M| if (a->comp_type[xb4]) cnt[a->ref[1][xb4] >= 4]++; ------------------ | Branch (292:13): [True: 137k, False: 1.17M] ------------------ 293| 1.31M| } 294| | 295| 1.62M| if (have_left && !l->intra[yb4]) { ------------------ | Branch (295:9): [True: 1.55M, False: 62.9k] | Branch (295:22): [True: 1.42M, False: 131k] ------------------ 296| 1.42M| cnt[l->ref[0][yb4] >= 4]++; 297| 1.42M| if (l->comp_type[yb4]) cnt[l->ref[1][yb4] >= 4]++; ------------------ | Branch (297:13): [True: 177k, False: 1.24M] ------------------ 298| 1.42M| } 299| | 300| 1.62M| return cnt[0] == cnt[1] ? 1 : cnt[0] < cnt[1] ? 0 : 2; ------------------ | Branch (300:12): [True: 219k, False: 1.40M] | Branch (300:35): [True: 440k, False: 960k] ------------------ 301| 1.62M|} decode.c:av1_get_uni_p1_ctx: 412| 52.3k|{ 413| 52.3k| int cnt[3] = { 0 }; 414| | 415| 52.3k| if (have_top && !a->intra[xb4]) { ------------------ | Branch (415:9): [True: 40.7k, False: 11.5k] | Branch (415:21): [True: 37.7k, False: 3.05k] ------------------ 416| 37.7k| if (a->ref[0][xb4] - 1U < 3) cnt[a->ref[0][xb4] - 1]++; ------------------ | Branch (416:13): [True: 5.58k, False: 32.1k] ------------------ 417| 37.7k| if (a->comp_type[xb4] && a->ref[1][xb4] - 1U < 3) cnt[a->ref[1][xb4] - 1]++; ------------------ | Branch (417:13): [True: 27.2k, False: 10.4k] | Branch (417:34): [True: 16.0k, False: 11.1k] ------------------ 418| 37.7k| } 419| | 420| 52.3k| if (have_left && !l->intra[yb4]) { ------------------ | Branch (420:9): [True: 50.4k, False: 1.89k] | Branch (420:22): [True: 48.5k, False: 1.93k] ------------------ 421| 48.5k| if (l->ref[0][yb4] - 1U < 3) cnt[l->ref[0][yb4] - 1]++; ------------------ | Branch (421:13): [True: 9.35k, False: 39.1k] ------------------ 422| 48.5k| if (l->comp_type[yb4] && l->ref[1][yb4] - 1U < 3) cnt[l->ref[1][yb4] - 1]++; ------------------ | Branch (422:13): [True: 37.2k, False: 11.2k] | Branch (422:34): [True: 20.3k, False: 16.9k] ------------------ 423| 48.5k| } 424| | 425| 52.3k| cnt[1] += cnt[2]; 426| | 427| 52.3k| return cnt[0] == cnt[1] ? 1 : cnt[0] < cnt[1] ? 0 : 2; ------------------ | Branch (427:12): [True: 16.7k, False: 35.5k] | Branch (427:35): [True: 23.4k, False: 12.1k] ------------------ 428| 52.3k|} decode.c:get_drl_context: 432| 852k|{ 433| 852k| if (ref_mv_stack[ref_idx].weight >= 640) ------------------ | Branch (433:9): [True: 702k, False: 149k] ------------------ 434| 702k| return ref_mv_stack[ref_idx + 1].weight < 640; 435| | 436| 18.4E| return ref_mv_stack[ref_idx + 1].weight < 640 ? 2 : 0; ------------------ | Branch (436:12): [True: 149k, False: 18.4E] ------------------ 437| 852k|} decode.c:get_gmv_2d: 482| 2.14M|{ 483| 2.14M| switch (gmv->type) { 484| 783k| case DAV1D_WM_TYPE_ROT_ZOOM: ------------------ | Branch (484:5): [True: 783k, False: 1.35M] ------------------ 485| 783k| assert(gmv->matrix[5] == gmv->matrix[2]); ------------------ | Branch (485:9): [True: 783k, False: 18.4E] ------------------ 486| 783k| assert(gmv->matrix[4] == -gmv->matrix[3]); ------------------ | Branch (486:9): [True: 783k, False: 6] ------------------ 487| | // fall-through 488| 783k| default: ------------------ | Branch (488:5): [True: 0, False: 2.14M] ------------------ 489| 918k| case DAV1D_WM_TYPE_AFFINE: { ------------------ | Branch (489:5): [True: 135k, False: 2.00M] ------------------ 490| 918k| const int x = bx4 * 4 + bw4 * 2 - 1; 491| 918k| const int y = by4 * 4 + bh4 * 2 - 1; 492| 918k| const int xc = (gmv->matrix[2] - (1 << 16)) * x + 493| 918k| gmv->matrix[3] * y + gmv->matrix[0]; 494| 918k| const int yc = (gmv->matrix[5] - (1 << 16)) * y + 495| 918k| gmv->matrix[4] * x + gmv->matrix[1]; 496| 918k| const int shift = 16 - (3 - !hdr->hp); 497| 918k| const int round = (1 << shift) >> 1; 498| 918k| mv res = (mv) { 499| 918k| .y = apply_sign(((abs(yc) + round) >> shift) << !hdr->hp, yc), 500| 918k| .x = apply_sign(((abs(xc) + round) >> shift) << !hdr->hp, xc), 501| 918k| }; 502| 918k| if (hdr->force_integer_mv) ------------------ | Branch (502:13): [True: 16.2k, False: 902k] ------------------ 503| 16.2k| fix_int_mv_precision(&res); 504| 918k| return res; 505| 783k| } 506| 65.1k| case DAV1D_WM_TYPE_TRANSLATION: { ------------------ | Branch (506:5): [True: 65.1k, False: 2.07M] ------------------ 507| 65.1k| mv res = (mv) { 508| 65.1k| .y = gmv->matrix[0] >> 13, 509| 65.1k| .x = gmv->matrix[1] >> 13, 510| 65.1k| }; 511| 65.1k| if (hdr->force_integer_mv) ------------------ | Branch (511:13): [True: 441, False: 64.7k] ------------------ 512| 441| fix_int_mv_precision(&res); 513| 65.1k| return res; 514| 783k| } 515| 1.15M| case DAV1D_WM_TYPE_IDENTITY: ------------------ | Branch (515:5): [True: 1.15M, False: 983k] ------------------ 516| 1.15M| return (mv) { .x = 0, .y = 0 }; 517| 2.14M| } 518| 2.14M|} decode.c:get_mask_comp_ctx: 266| 295k|{ 267| 295k| const int a_ctx = a->comp_type[xb4] >= COMP_INTER_SEG ? 1 : ------------------ | Branch (267:23): [True: 38.9k, False: 256k] ------------------ 268| 295k| a->ref[0][xb4] == 6 ? 3 : 0; ------------------ | Branch (268:23): [True: 14.8k, False: 241k] ------------------ 269| 295k| const int l_ctx = l->comp_type[yb4] >= COMP_INTER_SEG ? 1 : ------------------ | Branch (269:23): [True: 43.6k, False: 251k] ------------------ 270| 295k| l->ref[0][yb4] == 6 ? 3 : 0; ------------------ | Branch (270:23): [True: 18.0k, False: 233k] ------------------ 271| | 272| 295k| return imin(a_ctx + l_ctx, 5); 273| 295k|} decode.c:get_jnt_comp_ctx: 251| 228k|{ 252| 228k| const int d0 = abs(get_poc_diff(order_hint_n_bits, ref0poc, poc)); 253| 228k| const int d1 = abs(get_poc_diff(order_hint_n_bits, poc, ref1poc)); 254| 228k| const int offset = d0 == d1; 255| 228k| const int a_ctx = a->comp_type[xb4] >= COMP_INTER_AVG || ------------------ | Branch (255:23): [True: 100k, False: 128k] ------------------ 256| 128k| a->ref[0][xb4] == 6; ------------------ | Branch (256:23): [True: 10.3k, False: 117k] ------------------ 257| 228k| const int l_ctx = l->comp_type[yb4] >= COMP_INTER_AVG || ------------------ | Branch (257:23): [True: 115k, False: 113k] ------------------ 258| 113k| l->ref[0][yb4] == 6; ------------------ | Branch (258:23): [True: 12.1k, False: 101k] ------------------ 259| | 260| 228k| return 3 * offset + a_ctx + l_ctx; 261| 228k|} decode.c:get_filter_ctx: 139| 1.36M|{ 140| 1.36M| const int a_filter = (a->ref[0][xb4] == ref || a->ref[1][xb4] == ref) ? ------------------ | Branch (140:27): [True: 836k, False: 526k] | Branch (140:52): [True: 42.5k, False: 484k] ------------------ 141| 879k| a->filter[dir][xb4] : DAV1D_N_SWITCHABLE_FILTERS; 142| 1.36M| const int l_filter = (l->ref[0][yb4] == ref || l->ref[1][yb4] == ref) ? ------------------ | Branch (142:27): [True: 930k, False: 432k] | Branch (142:52): [True: 41.4k, False: 390k] ------------------ 143| 973k| l->filter[dir][yb4] : DAV1D_N_SWITCHABLE_FILTERS; 144| | 145| 1.36M| if (a_filter == l_filter) { ------------------ | Branch (145:9): [True: 812k, False: 550k] ------------------ 146| 812k| return comp * 4 + a_filter; 147| 812k| } else if (a_filter == DAV1D_N_SWITCHABLE_FILTERS) { ------------------ | Branch (147:16): [True: 297k, False: 252k] ------------------ 148| 297k| return comp * 4 + l_filter; 149| 297k| } else if (l_filter == DAV1D_N_SWITCHABLE_FILTERS) { ------------------ | Branch (149:16): [True: 203k, False: 48.4k] ------------------ 150| 203k| return comp * 4 + a_filter; 151| 203k| } else { 152| 48.4k| return comp * 4 + DAV1D_N_SWITCHABLE_FILTERS; 153| 48.4k| } 154| 1.36M|} decode.c:gather_top_partition_prob: 106| 362k|{ 107| | // Exploit the fact that cdfs for PARTITION_V, PARTITION_SPLIT and 108| | // PARTITION_T_TOP_SPLIT are neighbors. 109| 362k| unsigned out = in[PARTITION_V - 1] - in[PARTITION_T_TOP_SPLIT]; 110| | // Exploit the facts that cdfs for PARTITION_T_LEFT_SPLIT and 111| | // PARTITION_T_RIGHT_SPLIT are neighbors, the probability for 112| | // PARTITION_V4 is always zero, and the probability for 113| | // PARTITION_T_RIGHT_SPLIT is zero in 128x128 blocks. 114| 362k| out += in[PARTITION_T_LEFT_SPLIT - 1]; 115| 362k| if (bl != BL_128X128) ------------------ | Branch (115:9): [True: 316k, False: 45.7k] ------------------ 116| 316k| out += in[PARTITION_V4 - 1] - in[PARTITION_T_RIGHT_SPLIT]; 117| 362k| return out; 118| 362k|} decode.c:gather_left_partition_prob: 94| 371k|{ 95| 371k| unsigned out = in[PARTITION_H - 1] - in[PARTITION_H]; 96| | // Exploit the fact that cdfs for PARTITION_SPLIT, PARTITION_T_TOP_SPLIT, 97| | // PARTITION_T_BOTTOM_SPLIT and PARTITION_T_LEFT_SPLIT are neighbors. 98| 371k| out += in[PARTITION_SPLIT - 1] - in[PARTITION_T_LEFT_SPLIT]; 99| 371k| if (bl != BL_128X128) ------------------ | Branch (99:9): [True: 335k, False: 36.2k] ------------------ 100| 335k| out += in[PARTITION_H4 - 1] - in[PARTITION_H4]; 101| 371k| return out; 102| 371k|} decode.c:get_poc_diff: 239| 2.64M|{ 240| 2.64M| if (!order_hint_n_bits) return 0; ------------------ | Branch (240:9): [True: 60.1k, False: 2.58M] ------------------ 241| 2.58M| const int mask = 1 << (order_hint_n_bits - 1); 242| 2.58M| const int diff = poc0 - poc1; 243| 2.58M| return (diff & (mask - 1)) - (diff & mask); 244| 2.64M|} recon_tmpl.c:get_uv_inter_txtp: 122| 852k|{ 123| 852k| if (uvt_dim->max == TX_32X32) ------------------ | Branch (123:9): [True: 166k, False: 685k] ------------------ 124| 166k| return ytxtp == IDTX ? IDTX : DCT_DCT; ------------------ | Branch (124:16): [True: 4.45k, False: 162k] ------------------ 125| 685k| if (uvt_dim->min == TX_16X16 && ------------------ | Branch (125:9): [True: 70.5k, False: 614k] ------------------ 126| 70.5k| ((1 << ytxtp) & ((1 << H_FLIPADST) | (1 << V_FLIPADST) | ------------------ | Branch (126:9): [True: 582, False: 69.9k] ------------------ 127| 70.5k| (1 << H_ADST) | (1 << V_ADST)))) 128| 582| { 129| 582| return DCT_DCT; 130| 582| } 131| | 132| 684k| return ytxtp; 133| 685k|} dav1d_prep_grain_8bpc: 105| 7.70k|{ 106| 7.70k| const Dav1dFilmGrainData *const data = &out->frame_hdr->film_grain.data; 107| |#if BITDEPTH != 8 108| | const int bitdepth_max = (1 << out->p.bpc) - 1; 109| |#endif 110| | 111| | // Generate grain LUTs as needed 112| 7.70k| dsp->generate_grain_y(grain_lut[0], data HIGHBD_TAIL_SUFFIX); // always needed 113| 7.70k| if (data->num_uv_points[0] || data->chroma_scaling_from_luma) ------------------ | Branch (113:9): [True: 720, False: 6.98k] | Branch (113:35): [True: 712, False: 6.27k] ------------------ 114| 1.43k| dsp->generate_grain_uv[in->p.layout - 1](grain_lut[1], grain_lut[0], 115| 1.43k| data, 0 HIGHBD_TAIL_SUFFIX); 116| 7.70k| if (data->num_uv_points[1] || data->chroma_scaling_from_luma) ------------------ | Branch (116:9): [True: 3.21k, False: 4.49k] | Branch (116:35): [True: 712, False: 3.78k] ------------------ 117| 3.92k| dsp->generate_grain_uv[in->p.layout - 1](grain_lut[2], grain_lut[0], 118| 3.92k| data, 1 HIGHBD_TAIL_SUFFIX); 119| | 120| | // Generate scaling LUTs as needed 121| 7.70k| if (data->num_y_points || data->chroma_scaling_from_luma) ------------------ | Branch (121:9): [True: 3.78k, False: 3.92k] | Branch (121:31): [True: 700, False: 3.22k] ------------------ 122| 4.48k| generate_scaling(in->p.bpc, data->y_points, data->num_y_points, scaling[0]); 123| 7.70k| if (data->num_uv_points[0]) ------------------ | Branch (123:9): [True: 720, False: 6.98k] ------------------ 124| 720| generate_scaling(in->p.bpc, data->uv_points[0], data->num_uv_points[0], scaling[1]); 125| 7.70k| if (data->num_uv_points[1]) ------------------ | Branch (125:9): [True: 3.21k, False: 4.49k] ------------------ 126| 3.21k| generate_scaling(in->p.bpc, data->uv_points[1], data->num_uv_points[1], scaling[2]); 127| | 128| | // Copy over the non-modified planes 129| 7.70k| assert(out->stride[0] == in->stride[0]); ------------------ | Branch (129:5): [True: 7.70k, False: 0] ------------------ 130| 7.70k| if (!data->num_y_points) { ------------------ | Branch (130:9): [True: 3.92k, False: 3.78k] ------------------ 131| 3.92k| const ptrdiff_t stride = out->stride[0]; 132| 3.92k| const ptrdiff_t sz = out->p.h * stride; 133| 3.92k| if (sz < 0) ------------------ | Branch (133:13): [True: 0, False: 3.92k] ------------------ 134| 0| memcpy((uint8_t*) out->data[0] + sz - stride, 135| 0| (uint8_t*) in->data[0] + sz - stride, -sz); 136| 3.92k| else 137| 3.92k| memcpy(out->data[0], in->data[0], sz); 138| 3.92k| } 139| | 140| 7.70k| if (in->p.layout != DAV1D_PIXEL_LAYOUT_I400 && !data->chroma_scaling_from_luma) { ------------------ | Branch (140:9): [True: 7.20k, False: 501] | Branch (140:52): [True: 6.49k, False: 712] ------------------ 141| 6.49k| assert(out->stride[1] == in->stride[1]); ------------------ | Branch (141:9): [True: 6.49k, False: 0] ------------------ 142| 6.49k| const int ss_ver = in->p.layout == DAV1D_PIXEL_LAYOUT_I420; 143| 6.49k| const ptrdiff_t stride = out->stride[1]; 144| 6.49k| const ptrdiff_t sz = ((out->p.h + ss_ver) >> ss_ver) * stride; 145| 6.49k| if (sz < 0) { ------------------ | Branch (145:13): [True: 0, False: 6.49k] ------------------ 146| 0| if (!data->num_uv_points[0]) ------------------ | Branch (146:17): [True: 0, False: 0] ------------------ 147| 0| memcpy((uint8_t*) out->data[1] + sz - stride, 148| 0| (uint8_t*) in->data[1] + sz - stride, -sz); 149| 0| if (!data->num_uv_points[1]) ------------------ | Branch (149:17): [True: 0, False: 0] ------------------ 150| 0| memcpy((uint8_t*) out->data[2] + sz - stride, 151| 0| (uint8_t*) in->data[2] + sz - stride, -sz); 152| 6.49k| } else { 153| 6.49k| if (!data->num_uv_points[0]) ------------------ | Branch (153:17): [True: 5.77k, False: 720] ------------------ 154| 5.77k| memcpy(out->data[1], in->data[1], sz); 155| 6.49k| if (!data->num_uv_points[1]) ------------------ | Branch (155:17): [True: 3.28k, False: 3.21k] ------------------ 156| 3.28k| memcpy(out->data[2], in->data[2], sz); 157| 6.49k| } 158| 6.49k| } 159| 7.70k|} dav1d_apply_grain_row_8bpc: 167| 29.8k|{ 168| | // Synthesize grain for the affected planes 169| 29.8k| const Dav1dFilmGrainData *const data = &out->frame_hdr->film_grain.data; 170| 29.8k| const int ss_y = in->p.layout == DAV1D_PIXEL_LAYOUT_I420; 171| 29.8k| const int ss_x = in->p.layout != DAV1D_PIXEL_LAYOUT_I444; 172| 29.8k| const int cpw = (out->p.w + ss_x) >> ss_x; 173| 29.8k| const int is_id = out->seq_hdr->mtrx == DAV1D_MC_IDENTITY; 174| 29.8k| pixel *const luma_src = 175| 29.8k| ((pixel *) in->data[0]) + row * FG_BLOCK_SIZE * PXSTRIDE(in->stride[0]); ------------------ | | 37| 29.8k|#define FG_BLOCK_SIZE 32 ------------------ ((pixel *) in->data[0]) + row * FG_BLOCK_SIZE * PXSTRIDE(in->stride[0]); ------------------ | | 53| 29.8k|#define PXSTRIDE(x) (x) ------------------ 176| |#if BITDEPTH != 8 177| | const int bitdepth_max = (1 << out->p.bpc) - 1; 178| |#endif 179| | 180| 29.8k| if (data->num_y_points) { ------------------ | Branch (180:9): [True: 14.7k, False: 15.0k] ------------------ 181| 14.7k| const int bh = imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE); ------------------ | | 37| 14.7k|#define FG_BLOCK_SIZE 32 ------------------ const int bh = imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE); ------------------ | | 37| 14.7k|#define FG_BLOCK_SIZE 32 ------------------ 182| 14.7k| dsp->fgy_32x32xn(((pixel *) out->data[0]) + row * FG_BLOCK_SIZE * PXSTRIDE(out->stride[0]), ------------------ | | 37| 14.7k|#define FG_BLOCK_SIZE 32 ------------------ dsp->fgy_32x32xn(((pixel *) out->data[0]) + row * FG_BLOCK_SIZE * PXSTRIDE(out->stride[0]), ------------------ | | 53| 14.7k|#define PXSTRIDE(x) (x) ------------------ 183| 14.7k| luma_src, out->stride[0], data, 184| 14.7k| out->p.w, scaling[0], grain_lut[0], bh, row HIGHBD_TAIL_SUFFIX); 185| 14.7k| } 186| | 187| 29.8k| if (!data->num_uv_points[0] && !data->num_uv_points[1] && ------------------ | Branch (187:9): [True: 25.6k, False: 4.22k] | Branch (187:36): [True: 13.8k, False: 11.8k] ------------------ 188| 13.8k| !data->chroma_scaling_from_luma) ------------------ | Branch (188:9): [True: 10.3k, False: 3.42k] ------------------ 189| 10.3k| { 190| 10.3k| return; 191| 10.3k| } 192| | 193| 19.4k| const int bh = (imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE) + ss_y) >> ss_y; ------------------ | | 37| 19.4k|#define FG_BLOCK_SIZE 32 ------------------ const int bh = (imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE) + ss_y) >> ss_y; ------------------ | | 37| 19.4k|#define FG_BLOCK_SIZE 32 ------------------ 194| | 195| | // extend padding pixels 196| 19.4k| if (out->p.w & ss_x) { ------------------ | Branch (196:9): [True: 1.95k, False: 17.4k] ------------------ 197| 1.95k| pixel *ptr = luma_src; 198| 41.9k| for (int y = 0; y < bh; y++) { ------------------ | Branch (198:25): [True: 39.9k, False: 1.95k] ------------------ 199| 39.9k| ptr[out->p.w] = ptr[out->p.w - 1]; 200| 39.9k| ptr += PXSTRIDE(in->stride[0]) << ss_y; ------------------ | | 53| 39.9k|#define PXSTRIDE(x) (x) ------------------ 201| 39.9k| } 202| 1.95k| } 203| | 204| 19.4k| const ptrdiff_t uv_off = row * FG_BLOCK_SIZE * PXSTRIDE(out->stride[1]) >> ss_y; ------------------ | | 37| 19.4k|#define FG_BLOCK_SIZE 32 ------------------ const ptrdiff_t uv_off = row * FG_BLOCK_SIZE * PXSTRIDE(out->stride[1]) >> ss_y; ------------------ | | 53| 19.4k|#define PXSTRIDE(x) (x) ------------------ 205| 19.4k| if (data->chroma_scaling_from_luma) { ------------------ | Branch (205:9): [True: 3.35k, False: 16.0k] ------------------ 206| 10.0k| for (int pl = 0; pl < 2; pl++) ------------------ | Branch (206:26): [True: 6.70k, False: 3.35k] ------------------ 207| 6.70k| dsp->fguv_32x32xn[in->p.layout - 1](((pixel *) out->data[1 + pl]) + uv_off, 208| 6.70k| ((const pixel *) in->data[1 + pl]) + uv_off, 209| 6.70k| in->stride[1], data, cpw, 210| 6.70k| scaling[0], grain_lut[1 + pl], 211| 6.70k| bh, row, luma_src, in->stride[0], 212| 6.70k| pl, is_id HIGHBD_TAIL_SUFFIX); 213| 16.0k| } else { 214| 48.1k| for (int pl = 0; pl < 2; pl++) ------------------ | Branch (214:26): [True: 32.0k, False: 16.0k] ------------------ 215| 32.0k| if (data->num_uv_points[pl]) ------------------ | Branch (215:17): [True: 18.2k, False: 13.7k] ------------------ 216| 18.2k| dsp->fguv_32x32xn[in->p.layout - 1](((pixel *) out->data[1 + pl]) + uv_off, 217| 18.2k| ((const pixel *) in->data[1 + pl]) + uv_off, 218| 18.2k| in->stride[1], data, cpw, 219| 18.2k| scaling[1 + pl], grain_lut[1 + pl], 220| 18.2k| bh, row, luma_src, in->stride[0], 221| 18.2k| pl, is_id HIGHBD_TAIL_SUFFIX); 222| 16.0k| } 223| 19.4k|} fg_apply_tmpl.c:generate_scaling: 44| 8.41k|{ 45| 8.41k|#if BITDEPTH == 8 46| 8.41k| const int shift_x = 0; 47| 8.41k| const int scaling_size = SCALING_SIZE; ------------------ | | 39| 8.41k|#define SCALING_SIZE 256 ------------------ 48| |#else 49| | assert(bitdepth > 8); 50| | const int shift_x = bitdepth - 8; 51| | const int scaling_size = 1 << bitdepth; 52| |#endif 53| | 54| 8.41k| if (num == 0) { ------------------ | Branch (54:9): [True: 700, False: 7.71k] ------------------ 55| 700| memset(scaling, 0, scaling_size); 56| 700| return; 57| 700| } 58| | 59| | // Fill up the preceding entries with the initial value 60| 7.71k| memset(scaling, points[0][1], points[0][0] << shift_x); 61| | 62| | // Linearly interpolate the values in the middle 63| 19.3k| for (int i = 0; i < num - 1; i++) { ------------------ | Branch (63:21): [True: 11.6k, False: 7.71k] ------------------ 64| 11.6k| const int bx = points[i][0]; 65| 11.6k| const int by = points[i][1]; 66| 11.6k| const int ex = points[i+1][0]; 67| 11.6k| const int ey = points[i+1][1]; 68| 11.6k| const int dx = ex - bx; 69| 11.6k| const int dy = ey - by; 70| 11.6k| assert(dx > 0); ------------------ | Branch (70:9): [True: 11.6k, False: 0] ------------------ 71| 11.6k| const int delta = dy * ((0x10000 + (dx >> 1)) / dx); 72| 575k| for (int x = 0, d = 0x8000; x < dx; x++) { ------------------ | Branch (72:37): [True: 563k, False: 11.6k] ------------------ 73| 563k| scaling[(bx + x) << shift_x] = by + (d >> 16); 74| 563k| d += delta; 75| 563k| } 76| 11.6k| } 77| | 78| | // Fill up the remaining entries with the final value 79| 7.71k| const int n = points[num - 1][0] << shift_x; 80| 7.71k| memset(&scaling[n], points[num - 1][1], scaling_size - n); 81| | 82| |#if BITDEPTH != 8 83| | const int pad = 1 << shift_x, rnd = pad >> 1; 84| | for (int i = 0; i < num - 1; i++) { 85| | const int bx = points[i][0] << shift_x; 86| | const int ex = points[i+1][0] << shift_x; 87| | const int dx = ex - bx; 88| | for (int x = 0; x < dx; x += pad) { 89| | const int range = scaling[bx + x + pad] - scaling[bx + x]; 90| | for (int n = 1, r = rnd; n < pad; n++) { 91| | r += range; 92| | scaling[bx + x + n] = scaling[bx + x] + (r >> shift_x); 93| | } 94| | } 95| | } 96| |#endif 97| 7.71k|} dav1d_prep_grain_16bpc: 105| 2.40k|{ 106| 2.40k| const Dav1dFilmGrainData *const data = &out->frame_hdr->film_grain.data; 107| 2.40k|#if BITDEPTH != 8 108| 2.40k| const int bitdepth_max = (1 << out->p.bpc) - 1; 109| 2.40k|#endif 110| | 111| | // Generate grain LUTs as needed 112| 2.40k| dsp->generate_grain_y(grain_lut[0], data HIGHBD_TAIL_SUFFIX); // always needed ------------------ | | 74| 2.40k|#define HIGHBD_TAIL_SUFFIX , bitdepth_max ------------------ 113| 2.40k| if (data->num_uv_points[0] || data->chroma_scaling_from_luma) ------------------ | Branch (113:9): [True: 426, False: 1.98k] | Branch (113:35): [True: 1.44k, False: 533] ------------------ 114| 1.87k| dsp->generate_grain_uv[in->p.layout - 1](grain_lut[1], grain_lut[0], 115| 1.87k| data, 0 HIGHBD_TAIL_SUFFIX); ------------------ | | 74| 1.87k|#define HIGHBD_TAIL_SUFFIX , bitdepth_max ------------------ 116| 2.40k| if (data->num_uv_points[1] || data->chroma_scaling_from_luma) ------------------ | Branch (116:9): [True: 263, False: 2.14k] | Branch (116:35): [True: 1.44k, False: 696] ------------------ 117| 1.71k| dsp->generate_grain_uv[in->p.layout - 1](grain_lut[2], grain_lut[0], 118| 1.71k| data, 1 HIGHBD_TAIL_SUFFIX); ------------------ | | 74| 1.71k|#define HIGHBD_TAIL_SUFFIX , bitdepth_max ------------------ 119| | 120| | // Generate scaling LUTs as needed 121| 2.40k| if (data->num_y_points || data->chroma_scaling_from_luma) ------------------ | Branch (121:9): [True: 784, False: 1.62k] | Branch (121:31): [True: 1.02k, False: 595] ------------------ 122| 1.81k| generate_scaling(in->p.bpc, data->y_points, data->num_y_points, scaling[0]); 123| 2.40k| if (data->num_uv_points[0]) ------------------ | Branch (123:9): [True: 426, False: 1.98k] ------------------ 124| 426| generate_scaling(in->p.bpc, data->uv_points[0], data->num_uv_points[0], scaling[1]); 125| 2.40k| if (data->num_uv_points[1]) ------------------ | Branch (125:9): [True: 263, False: 2.14k] ------------------ 126| 263| generate_scaling(in->p.bpc, data->uv_points[1], data->num_uv_points[1], scaling[2]); 127| | 128| | // Copy over the non-modified planes 129| 2.40k| assert(out->stride[0] == in->stride[0]); ------------------ | Branch (129:5): [True: 2.40k, False: 0] ------------------ 130| 2.40k| if (!data->num_y_points) { ------------------ | Branch (130:9): [True: 1.62k, False: 784] ------------------ 131| 1.62k| const ptrdiff_t stride = out->stride[0]; 132| 1.62k| const ptrdiff_t sz = out->p.h * stride; 133| 1.62k| if (sz < 0) ------------------ | Branch (133:13): [True: 0, False: 1.62k] ------------------ 134| 0| memcpy((uint8_t*) out->data[0] + sz - stride, 135| 0| (uint8_t*) in->data[0] + sz - stride, -sz); 136| 1.62k| else 137| 1.62k| memcpy(out->data[0], in->data[0], sz); 138| 1.62k| } 139| | 140| 2.40k| if (in->p.layout != DAV1D_PIXEL_LAYOUT_I400 && !data->chroma_scaling_from_luma) { ------------------ | Branch (140:9): [True: 2.22k, False: 184] | Branch (140:52): [True: 775, False: 1.44k] ------------------ 141| 775| assert(out->stride[1] == in->stride[1]); ------------------ | Branch (141:9): [True: 775, False: 0] ------------------ 142| 775| const int ss_ver = in->p.layout == DAV1D_PIXEL_LAYOUT_I420; 143| 775| const ptrdiff_t stride = out->stride[1]; 144| 775| const ptrdiff_t sz = ((out->p.h + ss_ver) >> ss_ver) * stride; 145| 775| if (sz < 0) { ------------------ | Branch (145:13): [True: 0, False: 775] ------------------ 146| 0| if (!data->num_uv_points[0]) ------------------ | Branch (146:17): [True: 0, False: 0] ------------------ 147| 0| memcpy((uint8_t*) out->data[1] + sz - stride, 148| 0| (uint8_t*) in->data[1] + sz - stride, -sz); 149| 0| if (!data->num_uv_points[1]) ------------------ | Branch (149:17): [True: 0, False: 0] ------------------ 150| 0| memcpy((uint8_t*) out->data[2] + sz - stride, 151| 0| (uint8_t*) in->data[2] + sz - stride, -sz); 152| 775| } else { 153| 775| if (!data->num_uv_points[0]) ------------------ | Branch (153:17): [True: 349, False: 426] ------------------ 154| 349| memcpy(out->data[1], in->data[1], sz); 155| 775| if (!data->num_uv_points[1]) ------------------ | Branch (155:17): [True: 512, False: 263] ------------------ 156| 512| memcpy(out->data[2], in->data[2], sz); 157| 775| } 158| 775| } 159| 2.40k|} dav1d_apply_grain_row_16bpc: 167| 9.15k|{ 168| | // Synthesize grain for the affected planes 169| 9.15k| const Dav1dFilmGrainData *const data = &out->frame_hdr->film_grain.data; 170| 9.15k| const int ss_y = in->p.layout == DAV1D_PIXEL_LAYOUT_I420; 171| 9.15k| const int ss_x = in->p.layout != DAV1D_PIXEL_LAYOUT_I444; 172| 9.15k| const int cpw = (out->p.w + ss_x) >> ss_x; 173| 9.15k| const int is_id = out->seq_hdr->mtrx == DAV1D_MC_IDENTITY; 174| 9.15k| pixel *const luma_src = 175| 9.15k| ((pixel *) in->data[0]) + row * FG_BLOCK_SIZE * PXSTRIDE(in->stride[0]); ------------------ | | 37| 9.15k|#define FG_BLOCK_SIZE 32 ------------------ 176| 9.15k|#if BITDEPTH != 8 177| 9.15k| const int bitdepth_max = (1 << out->p.bpc) - 1; 178| 9.15k|#endif 179| | 180| 9.15k| if (data->num_y_points) { ------------------ | Branch (180:9): [True: 3.33k, False: 5.82k] ------------------ 181| 3.33k| const int bh = imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE); ------------------ | | 37| 3.33k|#define FG_BLOCK_SIZE 32 ------------------ const int bh = imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE); ------------------ | | 37| 3.33k|#define FG_BLOCK_SIZE 32 ------------------ 182| 3.33k| dsp->fgy_32x32xn(((pixel *) out->data[0]) + row * FG_BLOCK_SIZE * PXSTRIDE(out->stride[0]), ------------------ | | 37| 3.33k|#define FG_BLOCK_SIZE 32 ------------------ 183| 3.33k| luma_src, out->stride[0], data, 184| 3.33k| out->p.w, scaling[0], grain_lut[0], bh, row HIGHBD_TAIL_SUFFIX); ------------------ | | 74| 3.33k|#define HIGHBD_TAIL_SUFFIX , bitdepth_max ------------------ 185| 3.33k| } 186| | 187| 9.15k| if (!data->num_uv_points[0] && !data->num_uv_points[1] && ------------------ | Branch (187:9): [True: 8.03k, False: 1.12k] | Branch (187:36): [True: 7.36k, False: 668] ------------------ 188| 7.37k| !data->chroma_scaling_from_luma) ------------------ | Branch (188:9): [True: 1.73k, False: 5.64k] ------------------ 189| 1.73k| { 190| 1.73k| return; 191| 1.73k| } 192| | 193| 7.42k| const int bh = (imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE) + ss_y) >> ss_y; ------------------ | | 37| 7.42k|#define FG_BLOCK_SIZE 32 ------------------ const int bh = (imin(out->p.h - row * FG_BLOCK_SIZE, FG_BLOCK_SIZE) + ss_y) >> ss_y; ------------------ | | 37| 7.42k|#define FG_BLOCK_SIZE 32 ------------------ 194| | 195| | // extend padding pixels 196| 7.42k| if (out->p.w & ss_x) { ------------------ | Branch (196:9): [True: 2.57k, False: 4.85k] ------------------ 197| 2.57k| pixel *ptr = luma_src; 198| 36.6k| for (int y = 0; y < bh; y++) { ------------------ | Branch (198:25): [True: 34.1k, False: 2.57k] ------------------ 199| 34.1k| ptr[out->p.w] = ptr[out->p.w - 1]; 200| 34.1k| ptr += PXSTRIDE(in->stride[0]) << ss_y; 201| 34.1k| } 202| 2.57k| } 203| | 204| 7.42k| const ptrdiff_t uv_off = row * FG_BLOCK_SIZE * PXSTRIDE(out->stride[1]) >> ss_y; ------------------ | | 37| 7.42k|#define FG_BLOCK_SIZE 32 ------------------ 205| 7.42k| if (data->chroma_scaling_from_luma) { ------------------ | Branch (205:9): [True: 5.61k, False: 1.81k] ------------------ 206| 16.7k| for (int pl = 0; pl < 2; pl++) ------------------ | Branch (206:26): [True: 11.1k, False: 5.61k] ------------------ 207| 11.1k| dsp->fguv_32x32xn[in->p.layout - 1](((pixel *) out->data[1 + pl]) + uv_off, 208| 11.1k| ((const pixel *) in->data[1 + pl]) + uv_off, 209| 11.1k| in->stride[1], data, cpw, 210| 11.1k| scaling[0], grain_lut[1 + pl], 211| 11.1k| bh, row, luma_src, in->stride[0], 212| 11.1k| pl, is_id HIGHBD_TAIL_SUFFIX); ------------------ | | 74| 11.1k|#define HIGHBD_TAIL_SUFFIX , bitdepth_max ------------------ 213| 5.61k| } else { 214| 5.30k| for (int pl = 0; pl < 2; pl++) ------------------ | Branch (214:26): [True: 3.48k, False: 1.81k] ------------------ 215| 3.48k| if (data->num_uv_points[pl]) ------------------ | Branch (215:17): [True: 2.20k, False: 1.28k] ------------------ 216| 2.20k| dsp->fguv_32x32xn[in->p.layout - 1](((pixel *) out->data[1 + pl]) + uv_off, 217| 2.20k| ((const pixel *) in->data[1 + pl]) + uv_off, 218| 2.20k| in->stride[1], data, cpw, 219| 2.20k| scaling[1 + pl], grain_lut[1 + pl], 220| 2.20k| bh, row, luma_src, in->stride[0], 221| 2.20k| pl, is_id HIGHBD_TAIL_SUFFIX); ------------------ | | 74| 2.20k|#define HIGHBD_TAIL_SUFFIX , bitdepth_max ------------------ 222| 1.81k| } 223| 7.42k|} dav1d_film_grain_dsp_init_8bpc: 425| 3.46k|COLD void bitfn(dav1d_film_grain_dsp_init)(Dav1dFilmGrainDSPContext *const c) { 426| 3.46k| c->generate_grain_y = generate_grain_y_c; 427| 3.46k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I420 - 1] = generate_grain_uv_420_c; 428| 3.46k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I422 - 1] = generate_grain_uv_422_c; 429| 3.46k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I444 - 1] = generate_grain_uv_444_c; 430| | 431| 3.46k| c->fgy_32x32xn = fgy_32x32xn_c; 432| 3.46k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] = fguv_32x32xn_420_c; 433| 3.46k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I422 - 1] = fguv_32x32xn_422_c; 434| 3.46k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I444 - 1] = fguv_32x32xn_444_c; 435| | 436| 3.46k|#if HAVE_ASM 437| |#if ARCH_AARCH64 || ARCH_ARM 438| | film_grain_dsp_init_arm(c); 439| |#elif ARCH_X86 440| | film_grain_dsp_init_x86(c); 441| |#elif ARCH_RISCV 442| | film_grain_dsp_init_riscv(c); 443| |#endif 444| 3.46k|#endif 445| 3.46k|} dav1d_film_grain_dsp_init_16bpc: 425| 5.11k|COLD void bitfn(dav1d_film_grain_dsp_init)(Dav1dFilmGrainDSPContext *const c) { 426| 5.11k| c->generate_grain_y = generate_grain_y_c; 427| 5.11k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I420 - 1] = generate_grain_uv_420_c; 428| 5.11k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I422 - 1] = generate_grain_uv_422_c; 429| 5.11k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I444 - 1] = generate_grain_uv_444_c; 430| | 431| 5.11k| c->fgy_32x32xn = fgy_32x32xn_c; 432| 5.11k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] = fguv_32x32xn_420_c; 433| 5.11k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I422 - 1] = fguv_32x32xn_422_c; 434| 5.11k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I444 - 1] = fguv_32x32xn_444_c; 435| | 436| 5.11k|#if HAVE_ASM 437| |#if ARCH_AARCH64 || ARCH_ARM 438| | film_grain_dsp_init_arm(c); 439| |#elif ARCH_X86 440| | film_grain_dsp_init_x86(c); 441| |#elif ARCH_RISCV 442| | film_grain_dsp_init_riscv(c); 443| |#endif 444| 5.11k|#endif 445| 5.11k|} dav1d_init_get_bits: 38| 549k|{ 39| 549k| assert(sz); ------------------ | Branch (39:5): [True: 549k, False: 0] ------------------ 40| 549k| c->ptr = c->ptr_start = data; 41| 549k| c->ptr_end = &c->ptr_start[sz]; 42| 549k| c->state = 0; 43| 549k| c->bits_left = 0; 44| 549k| c->error = 0; 45| 549k|} dav1d_get_bit: 47| 12.3M|unsigned dav1d_get_bit(GetBits *const c) { 48| 12.3M| if (!c->bits_left) { ------------------ | Branch (48:9): [True: 2.03M, False: 10.2M] ------------------ 49| 2.03M| if (c->ptr >= c->ptr_end) { ------------------ | Branch (49:13): [True: 11.7k, False: 2.01M] ------------------ 50| 11.7k| c->error = 1; 51| 2.01M| } else { 52| 2.01M| const unsigned state = *c->ptr++; 53| 2.01M| c->bits_left = 7; 54| 2.01M| c->state = (uint64_t) state << 57; 55| 2.01M| return state >> 7; 56| 2.01M| } 57| 2.03M| } 58| | 59| 10.3M| const uint64_t state = c->state; 60| 10.3M| c->bits_left--; 61| 10.3M| c->state = state << 1; 62| 10.3M| return (unsigned) (state >> 63); 63| 12.3M|} dav1d_get_uleb128: 95| 187k|unsigned dav1d_get_uleb128(GetBits *const c) { 96| 187k| uint64_t val = 0; 97| 187k| unsigned i = 0, more; 98| | 99| 197k| do { 100| 197k| const int v = dav1d_get_bits(c, 8); 101| 197k| more = v & 0x80; 102| 197k| val |= ((uint64_t) (v & 0x7F)) << i; 103| 197k| i += 7; 104| 197k| } while (more && i < 56); ------------------ | Branch (104:14): [True: 9.86k, False: 187k] | Branch (104:22): [True: 9.77k, False: 88] ------------------ 105| | 106| 187k| if (val > UINT32_MAX || more) { ------------------ | Branch (106:9): [True: 669, False: 187k] | Branch (106:29): [True: 55, False: 187k] ------------------ 107| 724| c->error = 1; 108| 724| return 0; 109| 724| } 110| | 111| 187k| return (unsigned) val; 112| 187k|} dav1d_get_uniform: 114| 178k|unsigned dav1d_get_uniform(GetBits *const c, const unsigned max) { 115| | // Output in range [0..max-1] 116| | // max must be > 1, or else nothing is read from the bitstream 117| 178k| assert(max > 1); ------------------ | Branch (117:5): [True: 178k, False: 0] ------------------ 118| 178k| const int l = ulog2(max) + 1; 119| 178k| assert(l > 1); ------------------ | Branch (119:5): [True: 178k, False: 0] ------------------ 120| 178k| const unsigned m = (1U << l) - max; 121| 178k| const unsigned v = dav1d_get_bits(c, l - 1); 122| 178k| return v < m ? v : (v << 1) - m + dav1d_get_bit(c); ------------------ | Branch (122:12): [True: 165k, False: 12.8k] ------------------ 123| 178k|} dav1d_get_vlc: 125| 650|unsigned dav1d_get_vlc(GetBits *const c) { 126| 650| if (dav1d_get_bit(c)) ------------------ | Branch (126:9): [True: 260, False: 390] ------------------ 127| 260| return 0; 128| | 129| 390| int n_bits = 0; 130| 7.09k| do { 131| 7.09k| if (++n_bits == 32) ------------------ | Branch (131:13): [True: 73, False: 7.02k] ------------------ 132| 73| return UINT32_MAX; 133| 7.09k| } while (!dav1d_get_bit(c)); ------------------ | Branch (133:14): [True: 6.70k, False: 317] ------------------ 134| | 135| 317| return ((1U << n_bits) - 1) + dav1d_get_bits(c, n_bits); 136| 390|} dav1d_get_bits_subexp: 162| 178k|int dav1d_get_bits_subexp(GetBits *const c, const int ref, const unsigned n) { 163| 178k| return (int) get_bits_subexp_u(c, ref + (1 << n), 2 << n) - (1 << n); 164| 178k|} getbits.c:refill: 65| 2.10M|static inline void refill(GetBits *const c, const int n) { 66| 2.10M| assert(c->bits_left >= 0 && c->bits_left < 32); ------------------ | Branch (66:5): [True: 2.10M, False: 0] | Branch (66:5): [True: 2.10M, False: 0] ------------------ 67| 2.10M| unsigned state = 0; 68| 2.20M| do { 69| 2.20M| if (c->ptr >= c->ptr_end) { ------------------ | Branch (69:13): [True: 16.3k, False: 2.19M] ------------------ 70| 16.3k| c->error = 1; 71| 16.3k| if (state) break; ------------------ | Branch (71:17): [True: 1.04k, False: 15.3k] ------------------ 72| 15.3k| return; 73| 16.3k| } 74| 2.19M| state = (state << 8) | *c->ptr++; 75| 2.19M| c->bits_left += 8; 76| 2.19M| } while (n > c->bits_left); ------------------ | Branch (76:14): [True: 99.5k, False: 2.09M] ------------------ 77| 2.09M| c->state |= (uint64_t) state << (64 - c->bits_left); 78| 2.09M|} getbits.c:get_bits_subexp_u: 140| 178k|{ 141| 178k| unsigned v = 0; 142| | 143| 337k| for (int i = 0;; i++) { 144| 337k| const int b = i ? 3 + i - 1 : 3; ------------------ | Branch (144:23): [True: 158k, False: 178k] ------------------ 145| | 146| 337k| if (n < v + 3 * (1 << b)) { ------------------ | Branch (146:13): [True: 8.09k, False: 329k] ------------------ 147| 8.09k| v += dav1d_get_uniform(c, n - v + 1); 148| 8.09k| break; 149| 8.09k| } 150| | 151| 329k| if (!dav1d_get_bit(c)) { ------------------ | Branch (151:13): [True: 170k, False: 158k] ------------------ 152| 170k| v += dav1d_get_bits(c, b); 153| 170k| break; 154| 170k| } 155| | 156| 158k| v += 1 << b; 157| 158k| } 158| | 159| 178k| return ref * 2 <= n ? inv_recenter(ref, v) : n - inv_recenter(n - ref, v); ------------------ | Branch (159:12): [True: 168k, False: 10.1k] ------------------ 160| 178k|} obu.c:dav1d_bytealign_get_bits: 52| 756k|static inline void dav1d_bytealign_get_bits(GetBits *c) { 53| | // bits_left is never more than 7, because it is only incremented 54| | // by refill(), called by dav1d_get_bits and that never reads more 55| | // than 7 bits more than it needs. 56| | // 57| | // If this wasn't true, we would need to work out how many bits to 58| | // discard (bits_left % 8), subtract that from bits_left and then 59| | // shift state right by that amount. 60| 756k| assert(c->bits_left <= 7); ------------------ | Branch (60:5): [True: 756k, False: 0] ------------------ 61| | 62| 756k| c->bits_left = 0; 63| 756k| c->state = 0; 64| 756k|} dav1d_init_intra_edge_tree: 126| 1|COLD void dav1d_init_intra_edge_tree(void) { 127| | // This function is guaranteed to be called only once 128| 1| struct ModeSelMem mem; 129| | 130| 1| mem.nwc[BL_128X128] = &nodes.branch_sb128[1]; 131| 1| mem.nwc[BL_64X64] = &nodes.branch_sb128[1 + 4]; 132| 1| mem.nwc[BL_32X32] = &nodes.branch_sb128[1 + 4 + 16]; 133| 1| mem.nt = nodes.tip_sb128; 134| 1| init_mode_node(nodes.branch_sb128, BL_128X128, &mem, 1, 0); 135| 1| assert(mem.nwc[BL_128X128] == &nodes.branch_sb128[1 + 4]); ------------------ | Branch (135:5): [True: 1, False: 0] ------------------ 136| 1| assert(mem.nwc[BL_64X64] == &nodes.branch_sb128[1 + 4 + 16]); ------------------ | Branch (136:5): [True: 1, False: 0] ------------------ 137| 1| assert(mem.nwc[BL_32X32] == &nodes.branch_sb128[1 + 4 + 16 + 64]); ------------------ | Branch (137:5): [True: 1, False: 0] ------------------ 138| 1| assert(mem.nt == &nodes.tip_sb128[256]); ------------------ | Branch (138:5): [True: 1, False: 0] ------------------ 139| | 140| 1| mem.nwc[BL_128X128] = NULL; 141| 1| mem.nwc[BL_64X64] = &nodes.branch_sb64[1]; 142| 1| mem.nwc[BL_32X32] = &nodes.branch_sb64[1 + 4]; 143| 1| mem.nt = nodes.tip_sb64; 144| 1| init_mode_node(nodes.branch_sb64, BL_64X64, &mem, 1, 0); 145| 1| assert(mem.nwc[BL_64X64] == &nodes.branch_sb64[1 + 4]); ------------------ | Branch (145:5): [True: 1, False: 0] ------------------ 146| 1| assert(mem.nwc[BL_32X32] == &nodes.branch_sb64[1 + 4 + 16]); ------------------ | Branch (146:5): [True: 1, False: 0] ------------------ 147| 1| assert(mem.nt == &nodes.tip_sb64[64]); ------------------ | Branch (147:5): [True: 1, False: 0] ------------------ 148| 1|} intra_edge.c:init_mode_node: 101| 106|{ 102| 106| init_edges(&nwc->node, bl, 103| 106| (top_has_right ? EDGE_ALL_TOP_HAS_RIGHT : 0) | ------------------ | Branch (103:17): [True: 73, False: 33] ------------------ 104| 106| (left_has_bottom ? EDGE_ALL_LEFT_HAS_BOTTOM : 0)); ------------------ | Branch (104:17): [True: 33, False: 73] ------------------ 105| 106| if (bl == BL_16X16) { ------------------ | Branch (105:9): [True: 80, False: 26] ------------------ 106| 400| for (int n = 0; n < 4; n++) { ------------------ | Branch (106:25): [True: 320, False: 80] ------------------ 107| 320| EdgeTip *const nt = mem->nt++; 108| 320| nwc->split_offset[n] = PTR_OFFSET(nwc, nt); ------------------ | | 94| 320|#define PTR_OFFSET(a, b) ((uint16_t)((uintptr_t)(b) - (uintptr_t)(a))) ------------------ 109| 320| init_edges(&nt->node, bl + 1, 110| 320| ((n == 3 || (n == 1 && !top_has_right)) ? 0 : ------------------ | Branch (110:26): [True: 80, False: 240] | Branch (110:37): [True: 80, False: 160] | Branch (110:47): [True: 26, False: 54] ------------------ 111| 320| EDGE_ALL_TOP_HAS_RIGHT) | 112| 320| (!(n == 0 || (n == 2 && left_has_bottom)) ? 0 : ------------------ | Branch (112:27): [True: 80, False: 240] | Branch (112:38): [True: 80, False: 160] | Branch (112:48): [True: 26, False: 54] ------------------ 113| 320| EDGE_ALL_LEFT_HAS_BOTTOM)); 114| 320| } 115| 80| } else { 116| 130| for (int n = 0; n < 4; n++) { ------------------ | Branch (116:25): [True: 104, False: 26] ------------------ 117| 104| EdgeBranch *const nwc_child = mem->nwc[bl]++; 118| 104| nwc->split_offset[n] = PTR_OFFSET(nwc, nwc_child); ------------------ | | 94| 104|#define PTR_OFFSET(a, b) ((uint16_t)((uintptr_t)(b) - (uintptr_t)(a))) ------------------ 119| 104| init_mode_node(nwc_child, bl + 1, mem, 120| 104| !(n == 3 || (n == 1 && !top_has_right)), ------------------ | Branch (120:30): [True: 26, False: 78] | Branch (120:41): [True: 26, False: 52] | Branch (120:51): [True: 7, False: 19] ------------------ 121| 104| n == 0 || (n == 2 && left_has_bottom)); ------------------ | Branch (121:28): [True: 26, False: 78] | Branch (121:39): [True: 26, False: 52] | Branch (121:49): [True: 7, False: 19] ------------------ 122| 104| } 123| 26| } 124| 106|} intra_edge.c:init_edges: 58| 426|{ 59| 426| node->o = edge_flags; 60| 426| node->h[0] = edge_flags | EDGE_ALL_LEFT_HAS_BOTTOM; 61| 426| node->v[0] = edge_flags | EDGE_ALL_TOP_HAS_RIGHT; 62| | 63| 426| if (bl == BL_8X8) { ------------------ | Branch (63:9): [True: 320, False: 106] ------------------ 64| 320| EdgeTip *const nt = (EdgeTip *) node; 65| | 66| 320| node->h[1] = edge_flags & (EDGE_ALL_LEFT_HAS_BOTTOM | 67| 320| EDGE_I420_TOP_HAS_RIGHT); 68| 320| node->v[1] = edge_flags & (EDGE_ALL_TOP_HAS_RIGHT | 69| 320| EDGE_I420_LEFT_HAS_BOTTOM | 70| 320| EDGE_I422_LEFT_HAS_BOTTOM); 71| | 72| 320| nt->split[0] = (edge_flags & EDGE_ALL_TOP_HAS_RIGHT) | 73| 320| EDGE_I422_LEFT_HAS_BOTTOM; 74| 320| nt->split[1] = edge_flags | EDGE_I444_TOP_HAS_RIGHT; 75| 320| nt->split[2] = edge_flags & (EDGE_I420_TOP_HAS_RIGHT | 76| 320| EDGE_I420_LEFT_HAS_BOTTOM | 77| 320| EDGE_I422_LEFT_HAS_BOTTOM); 78| 320| } else { 79| 106| EdgeBranch *const nwc = (EdgeBranch *) node; 80| | 81| 106| node->h[1] = edge_flags & EDGE_ALL_LEFT_HAS_BOTTOM; 82| 106| node->v[1] = edge_flags & EDGE_ALL_TOP_HAS_RIGHT; 83| | 84| 106| nwc->h4 = EDGE_ALL_LEFT_HAS_BOTTOM; 85| 106| nwc->v4 = EDGE_ALL_TOP_HAS_RIGHT; 86| 106| if (bl == BL_16X16) { ------------------ | Branch (86:13): [True: 80, False: 26] ------------------ 87| 80| nwc->h4 |= edge_flags & EDGE_I420_TOP_HAS_RIGHT; 88| 80| nwc->v4 |= edge_flags & (EDGE_I420_LEFT_HAS_BOTTOM | 89| 80| EDGE_I422_LEFT_HAS_BOTTOM); 90| 80| } 91| 106| } 92| 426|} recon_tmpl.c:sm_flag: 95| 4.75M|static inline int sm_flag(const BlockContext *const b, const int idx) { 96| 4.75M| if (!b->intra[idx]) return 0; ------------------ | Branch (96:9): [True: 249k, False: 4.50M] ------------------ 97| 4.50M| const enum IntraPredMode m = b->mode[idx]; 98| 4.50M| return (m == SMOOTH_PRED || m == SMOOTH_H_PRED || ------------------ | Branch (98:13): [True: 165k, False: 4.33M] | Branch (98:33): [True: 78.9k, False: 4.26M] ------------------ 99| 4.26M| m == SMOOTH_V_PRED) ? ANGLE_SMOOTH_EDGE_FLAG : 0; ------------------ | | 93| 305k|#define ANGLE_SMOOTH_EDGE_FLAG 512 ------------------ | Branch (99:13): [True: 57.7k, False: 4.20M] ------------------ 100| 4.75M|} recon_tmpl.c:sm_uv_flag: 102| 2.65M|static inline int sm_uv_flag(const BlockContext *const b, const int idx) { 103| 2.65M| const enum IntraPredMode m = b->uvmode[idx]; 104| 2.65M| return (m == SMOOTH_PRED || m == SMOOTH_H_PRED || ------------------ | Branch (104:13): [True: 90.3k, False: 2.56M] | Branch (104:33): [True: 57.1k, False: 2.51M] ------------------ 105| 2.51M| m == SMOOTH_V_PRED) ? ANGLE_SMOOTH_EDGE_FLAG : 0; ------------------ | | 93| 188k|#define ANGLE_SMOOTH_EDGE_FLAG 512 ------------------ | Branch (105:13): [True: 40.0k, False: 2.47M] ------------------ 106| 2.65M|} dav1d_prepare_intra_edges_8bpc: 86| 13.6M|{ 87| 13.6M| const int bitdepth = bitdepth_from_max(bitdepth_max); ------------------ | | 58| 13.6M|#define bitdepth_from_max(x) 8 ------------------ 88| 13.6M| assert(y < h && x < w); ------------------ | Branch (88:5): [True: 13.6M, False: 4.12k] | Branch (88:5): [True: 13.6M, False: 18.4E] ------------------ 89| | 90| 13.6M| switch (mode) { 91| 203k| case VERT_PRED: ------------------ | Branch (91:5): [True: 203k, False: 13.4M] ------------------ 92| 554k| case HOR_PRED: ------------------ | Branch (92:5): [True: 351k, False: 13.3M] ------------------ 93| 603k| case DIAG_DOWN_LEFT_PRED: ------------------ | Branch (93:5): [True: 48.3k, False: 13.6M] ------------------ 94| 652k| case DIAG_DOWN_RIGHT_PRED: ------------------ | Branch (94:5): [True: 49.4k, False: 13.6M] ------------------ 95| 699k| case VERT_RIGHT_PRED: ------------------ | Branch (95:5): [True: 46.5k, False: 13.6M] ------------------ 96| 1.57M| case HOR_DOWN_PRED: ------------------ | Branch (96:5): [True: 874k, False: 12.7M] ------------------ 97| 1.72M| case HOR_UP_PRED: ------------------ | Branch (97:5): [True: 147k, False: 13.5M] ------------------ 98| 1.77M| case VERT_LEFT_PRED: { ------------------ | Branch (98:5): [True: 57.4k, False: 13.6M] ------------------ 99| 1.77M| *angle = av1_mode_to_angle_map[mode - VERT_PRED] + 3 * *angle; 100| | 101| 1.77M| if (*angle <= 90) ------------------ | Branch (101:13): [True: 260k, False: 1.51M] ------------------ 102| 260k| mode = *angle < 90 && have_top ? Z1_PRED : VERT_PRED; ------------------ | Branch (102:20): [True: 144k, False: 116k] | Branch (102:35): [True: 142k, False: 1.79k] ------------------ 103| 1.51M| else if (*angle < 180) ------------------ | Branch (103:18): [True: 1.06M, False: 452k] ------------------ 104| 1.06M| mode = Z2_PRED; 105| 452k| else 106| 452k| mode = *angle > 180 && have_left ? Z3_PRED : HOR_PRED; ------------------ | Branch (106:20): [True: 216k, False: 236k] | Branch (106:36): [True: 211k, False: 4.48k] ------------------ 107| 1.77M| break; 108| 1.72M| } 109| 10.9M| case DC_PRED: ------------------ | Branch (109:5): [True: 10.9M, False: 2.73M] ------------------ 110| 11.2M| case PAETH_PRED: ------------------ | Branch (110:5): [True: 297k, False: 13.3M] ------------------ 111| 11.2M| mode = av1_mode_conv[mode][have_left][have_top]; 112| 11.2M| break; 113| 672k| default: ------------------ | Branch (113:5): [True: 672k, False: 12.9M] ------------------ 114| 672k| break; 115| 13.6M| } 116| | 117| 13.6M| const pixel *dst_top; 118| 13.6M| if (have_top && ------------------ | Branch (118:9): [True: 13.5M, False: 167k] ------------------ 119| 13.5M| (av1_intra_prediction_edges[mode].needs_top || ------------------ | Branch (119:10): [True: 13.0M, False: 440k] ------------------ 120| 440k| av1_intra_prediction_edges[mode].needs_topleft || ------------------ | Branch (120:10): [True: 205k, False: 235k] ------------------ 121| 235k| (av1_intra_prediction_edges[mode].needs_left && !have_left))) ------------------ | Branch (121:11): [True: 235k, False: 0] | Branch (121:58): [True: 9.31k, False: 225k] ------------------ 122| 13.2M| { 123| 13.2M| if (prefilter_toplevel_sb_edge) { ------------------ | Branch (123:13): [True: 1.04M, False: 12.2M] ------------------ 124| 1.04M| dst_top = &prefilter_toplevel_sb_edge[x * 4]; 125| 12.2M| } else { 126| 12.2M| dst_top = &dst[-PXSTRIDE(stride)]; ------------------ | | 53| 12.2M|#define PXSTRIDE(x) (x) ------------------ 127| 12.2M| } 128| 13.2M| } 129| | 130| 13.6M| if (av1_intra_prediction_edges[mode].needs_left) { ------------------ | Branch (130:9): [True: 12.3M, False: 1.35M] ------------------ 131| 12.3M| const int sz = th << 2; 132| 12.3M| pixel *const left = &topleft_out[-sz]; 133| | 134| 12.3M| if (have_left) { ------------------ | Branch (134:13): [True: 12.2M, False: 116k] ------------------ 135| 12.2M| const int px_have = imin(sz, (h - y) << 2); 136| | 137| 95.9M| for (int i = 0; i < px_have; i++) ------------------ | Branch (137:29): [True: 83.7M, False: 12.2M] ------------------ 138| 83.7M| left[sz - 1 - i] = dst[PXSTRIDE(stride) * i - 1]; ------------------ | | 53| 83.7M|#define PXSTRIDE(x) (x) ------------------ 139| 12.2M| if (px_have < sz) ------------------ | Branch (139:17): [True: 21.8k, False: 12.1M] ------------------ 140| 21.8k| pixel_set(left, left[sz - px_have], sz - px_have); ------------------ | | 48| 21.8k|#define pixel_set memset ------------------ 141| 12.2M| } else { 142| 116k| pixel_set(left, have_top ? *dst_top : ((1 << bitdepth) >> 1) + 1, sz); ------------------ | | 48| 116k|#define pixel_set memset ------------------ | Branch (142:29): [True: 111k, False: 4.57k] ------------------ 143| 116k| } 144| | 145| 12.3M| if (av1_intra_prediction_edges[mode].needs_bottomleft) { ------------------ | Branch (145:13): [True: 211k, False: 12.1M] ------------------ 146| 211k| const int have_bottomleft = (!have_left || y + th >= h) ? 0 : ------------------ | Branch (146:42): [True: 18.4E, False: 211k] | Branch (146:56): [True: 4.36k, False: 207k] ------------------ 147| 211k| (edge_flags & EDGE_I444_LEFT_HAS_BOTTOM); 148| | 149| 211k| if (have_bottomleft) { ------------------ | Branch (149:17): [True: 45.6k, False: 165k] ------------------ 150| 45.6k| const int px_have = imin(sz, (h - y - th) << 2); 151| | 152| 417k| for (int i = 0; i < px_have; i++) ------------------ | Branch (152:33): [True: 371k, False: 45.6k] ------------------ 153| 371k| left[-(i + 1)] = dst[(sz + i) * PXSTRIDE(stride) - 1]; ------------------ | | 53| 371k|#define PXSTRIDE(x) (x) ------------------ 154| 45.6k| if (px_have < sz) ------------------ | Branch (154:21): [True: 418, False: 45.2k] ------------------ 155| 418| pixel_set(left - sz, left[-px_have], sz - px_have); ------------------ | | 48| 418|#define pixel_set memset ------------------ 156| 165k| } else { 157| 165k| pixel_set(left - sz, left[0], sz); ------------------ | | 48| 165k|#define pixel_set memset ------------------ 158| 165k| } 159| 211k| } 160| 12.3M| } 161| | 162| 13.6M| if (av1_intra_prediction_edges[mode].needs_top) { ------------------ | Branch (162:9): [True: 13.1M, False: 560k] ------------------ 163| 13.1M| const int sz = tw << 2; 164| 13.1M| pixel *const top = &topleft_out[1]; 165| | 166| 13.1M| if (have_top) { ------------------ | Branch (166:13): [True: 13.0M, False: 44.9k] ------------------ 167| 13.0M| const int px_have = imin(sz, (w - x) << 2); 168| 13.0M| pixel_copy(top, dst_top, px_have); ------------------ | | 47| 13.0M|#define pixel_copy memcpy ------------------ 169| 13.0M| if (px_have < sz) ------------------ | Branch (169:17): [True: 416k, False: 12.6M] ------------------ 170| 416k| pixel_set(top + px_have, top[px_have - 1], sz - px_have); ------------------ | | 48| 416k|#define pixel_set memset ------------------ 171| 13.0M| } else { 172| 44.9k| pixel_set(top, have_left ? dst[-1] : ((1 << bitdepth) >> 1) - 1, sz); ------------------ | | 48| 44.9k|#define pixel_set memset ------------------ | Branch (172:28): [True: 43.8k, False: 1.15k] ------------------ 173| 44.9k| } 174| | 175| 13.1M| if (av1_intra_prediction_edges[mode].needs_topright) { ------------------ | Branch (175:13): [True: 142k, False: 12.9M] ------------------ 176| 142k| const int have_topright = (!have_top || x + tw >= w) ? 0 : ------------------ | Branch (176:40): [True: 18.4E, False: 142k] | Branch (176:53): [True: 5.42k, False: 137k] ------------------ 177| 142k| (edge_flags & EDGE_I444_TOP_HAS_RIGHT); 178| | 179| 142k| if (have_topright) { ------------------ | Branch (179:17): [True: 100k, False: 41.6k] ------------------ 180| 100k| const int px_have = imin(sz, (w - x - tw) << 2); 181| | 182| 100k| pixel_copy(top + sz, &dst_top[sz], px_have); ------------------ | | 47| 100k|#define pixel_copy memcpy ------------------ 183| 100k| if (px_have < sz) ------------------ | Branch (183:21): [True: 1.52k, False: 99.2k] ------------------ 184| 1.52k| pixel_set(top + sz + px_have, top[sz + px_have - 1], ------------------ | | 48| 1.52k|#define pixel_set memset ------------------ 185| 1.52k| sz - px_have); 186| 100k| } else { 187| 41.6k| pixel_set(top + sz, top[sz - 1], sz); ------------------ | | 48| 41.6k|#define pixel_set memset ------------------ 188| 41.6k| } 189| 142k| } 190| 13.1M| } 191| | 192| 13.6M| if (av1_intra_prediction_edges[mode].needs_topleft) { ------------------ | Branch (192:9): [True: 1.91M, False: 11.7M] ------------------ 193| 1.91M| if (have_left) ------------------ | Branch (193:13): [True: 1.81M, False: 96.3k] ------------------ 194| 1.81M| *topleft_out = have_top ? dst_top[-1] : dst[-1]; ------------------ | Branch (194:28): [True: 1.77M, False: 35.9k] ------------------ 195| 96.3k| else 196| 96.3k| *topleft_out = have_top ? *dst_top : (1 << bitdepth) >> 1; ------------------ | Branch (196:28): [True: 93.2k, False: 3.09k] ------------------ 197| | 198| 1.91M| if (mode == Z2_PRED && tw + th >= 6 && filter_edge) ------------------ | Branch (198:13): [True: 1.06M, False: 846k] | Branch (198:32): [True: 49.5k, False: 1.01M] | Branch (198:48): [True: 45.3k, False: 4.21k] ------------------ 199| 45.3k| *topleft_out = ((topleft_out[-1] + topleft_out[1]) * 5 + 200| 45.3k| topleft_out[0] * 6 + 8) >> 4; 201| 1.91M| } 202| | 203| 13.6M| return mode; 204| 13.6M|} dav1d_prepare_intra_edges_16bpc: 86| 16.4M|{ 87| 16.4M| const int bitdepth = bitdepth_from_max(bitdepth_max); ------------------ | | 75| 16.4M|#define bitdepth_from_max(bitdepth_max) (32 - clz(bitdepth_max)) ------------------ 88| 16.4M| assert(y < h && x < w); ------------------ | Branch (88:5): [True: 16.4M, False: 443] | Branch (88:5): [True: 16.4M, False: 18.4E] ------------------ 89| | 90| 16.4M| switch (mode) { 91| 434k| case VERT_PRED: ------------------ | Branch (91:5): [True: 434k, False: 15.9M] ------------------ 92| 1.37M| case HOR_PRED: ------------------ | Branch (92:5): [True: 938k, False: 15.4M] ------------------ 93| 1.55M| case DIAG_DOWN_LEFT_PRED: ------------------ | Branch (93:5): [True: 184k, False: 16.2M] ------------------ 94| 1.78M| case DIAG_DOWN_RIGHT_PRED: ------------------ | Branch (94:5): [True: 225k, False: 16.1M] ------------------ 95| 1.86M| case VERT_RIGHT_PRED: ------------------ | Branch (95:5): [True: 86.5k, False: 16.3M] ------------------ 96| 2.18M| case HOR_DOWN_PRED: ------------------ | Branch (96:5): [True: 318k, False: 16.0M] ------------------ 97| 3.17M| case HOR_UP_PRED: ------------------ | Branch (97:5): [True: 992k, False: 15.4M] ------------------ 98| 3.36M| case VERT_LEFT_PRED: { ------------------ | Branch (98:5): [True: 190k, False: 16.2M] ------------------ 99| 3.36M| *angle = av1_mode_to_angle_map[mode - VERT_PRED] + 3 * *angle; 100| | 101| 3.36M| if (*angle <= 90) ------------------ | Branch (101:13): [True: 744k, False: 2.62M] ------------------ 102| 744k| mode = *angle < 90 && have_top ? Z1_PRED : VERT_PRED; ------------------ | Branch (102:20): [True: 493k, False: 250k] | Branch (102:35): [True: 477k, False: 16.1k] ------------------ 103| 2.62M| else if (*angle < 180) ------------------ | Branch (103:18): [True: 913k, False: 1.71M] ------------------ 104| 913k| mode = Z2_PRED; 105| 1.71M| else 106| 1.71M| mode = *angle > 180 && have_left ? Z3_PRED : HOR_PRED; ------------------ | Branch (106:20): [True: 1.18M, False: 528k] | Branch (106:36): [True: 1.17M, False: 8.89k] ------------------ 107| 3.36M| break; 108| 3.17M| } 109| 10.1M| case DC_PRED: ------------------ | Branch (109:5): [True: 10.1M, False: 6.30M] ------------------ 110| 10.8M| case PAETH_PRED: ------------------ | Branch (110:5): [True: 760k, False: 15.6M] ------------------ 111| 10.8M| mode = av1_mode_conv[mode][have_left][have_top]; 112| 10.8M| break; 113| 2.29M| default: ------------------ | Branch (113:5): [True: 2.29M, False: 14.1M] ------------------ 114| 2.29M| break; 115| 16.4M| } 116| | 117| 16.4M| const pixel *dst_top; 118| 16.4M| if (have_top && ------------------ | Branch (118:9): [True: 15.5M, False: 889k] ------------------ 119| 15.5M| (av1_intra_prediction_edges[mode].needs_top || ------------------ | Branch (119:10): [True: 13.8M, False: 1.65M] ------------------ 120| 1.65M| av1_intra_prediction_edges[mode].needs_topleft || ------------------ | Branch (120:10): [True: 1.13M, False: 519k] ------------------ 121| 519k| (av1_intra_prediction_edges[mode].needs_left && !have_left))) ------------------ | Branch (121:11): [True: 519k, False: 5] | Branch (121:58): [True: 9.36k, False: 509k] ------------------ 122| 15.0M| { 123| 15.0M| if (prefilter_toplevel_sb_edge) { ------------------ | Branch (123:13): [True: 780k, False: 14.2M] ------------------ 124| 780k| dst_top = &prefilter_toplevel_sb_edge[x * 4]; 125| 14.2M| } else { 126| 14.2M| dst_top = &dst[-PXSTRIDE(stride)]; 127| 14.2M| } 128| 15.0M| } 129| | 130| 16.4M| if (av1_intra_prediction_edges[mode].needs_left) { ------------------ | Branch (130:9): [True: 14.3M, False: 2.04M] ------------------ 131| 14.3M| const int sz = th << 2; 132| 14.3M| pixel *const left = &topleft_out[-sz]; 133| | 134| 14.3M| if (have_left) { ------------------ | Branch (134:13): [True: 14.3M, False: 35.9k] ------------------ 135| 14.3M| const int px_have = imin(sz, (h - y) << 2); 136| | 137| 106M| for (int i = 0; i < px_have; i++) ------------------ | Branch (137:29): [True: 92.3M, False: 14.3M] ------------------ 138| 92.3M| left[sz - 1 - i] = dst[PXSTRIDE(stride) * i - 1]; 139| 14.3M| if (px_have < sz) ------------------ | Branch (139:17): [True: 36.0k, False: 14.2M] ------------------ 140| 36.0k| pixel_set(left, left[sz - px_have], sz - px_have); 141| 14.3M| } else { 142| 35.9k| pixel_set(left, have_top ? *dst_top : ((1 << bitdepth) >> 1) + 1, sz); ------------------ | Branch (142:29): [True: 34.5k, False: 1.42k] ------------------ 143| 35.9k| } 144| | 145| 14.3M| if (av1_intra_prediction_edges[mode].needs_bottomleft) { ------------------ | Branch (145:13): [True: 1.17M, False: 13.1M] ------------------ 146| 1.17M| const int have_bottomleft = (!have_left || y + th >= h) ? 0 : ------------------ | Branch (146:42): [True: 45, False: 1.17M] | Branch (146:56): [True: 35.5k, False: 1.13M] ------------------ 147| 1.17M| (edge_flags & EDGE_I444_LEFT_HAS_BOTTOM); 148| | 149| 1.17M| if (have_bottomleft) { ------------------ | Branch (149:17): [True: 104k, False: 1.06M] ------------------ 150| 104k| const int px_have = imin(sz, (h - y - th) << 2); 151| | 152| 737k| for (int i = 0; i < px_have; i++) ------------------ | Branch (152:33): [True: 632k, False: 104k] ------------------ 153| 632k| left[-(i + 1)] = dst[(sz + i) * PXSTRIDE(stride) - 1]; 154| 104k| if (px_have < sz) ------------------ | Branch (154:21): [True: 536, False: 104k] ------------------ 155| 536| pixel_set(left - sz, left[-px_have], sz - px_have); 156| 1.06M| } else { 157| 1.06M| pixel_set(left - sz, left[0], sz); 158| 1.06M| } 159| 1.17M| } 160| 14.3M| } 161| | 162| 16.4M| if (av1_intra_prediction_edges[mode].needs_top) { ------------------ | Branch (162:9): [True: 14.0M, False: 2.39M] ------------------ 163| 14.0M| const int sz = tw << 2; 164| 14.0M| pixel *const top = &topleft_out[1]; 165| | 166| 14.0M| if (have_top) { ------------------ | Branch (166:13): [True: 13.8M, False: 127k] ------------------ 167| 13.8M| const int px_have = imin(sz, (w - x) << 2); 168| 13.8M| pixel_copy(top, dst_top, px_have); ------------------ | | 65| 13.8M|#define pixel_copy(a, b, c) memcpy(a, b, (c) << 1) ------------------ 169| 13.8M| if (px_have < sz) ------------------ | Branch (169:17): [True: 817k, False: 13.0M] ------------------ 170| 817k| pixel_set(top + px_have, top[px_have - 1], sz - px_have); 171| 13.8M| } else { 172| 127k| pixel_set(top, have_left ? dst[-1] : ((1 << bitdepth) >> 1) - 1, sz); ------------------ | Branch (172:28): [True: 121k, False: 6.34k] ------------------ 173| 127k| } 174| | 175| 14.0M| if (av1_intra_prediction_edges[mode].needs_topright) { ------------------ | Branch (175:13): [True: 476k, False: 13.5M] ------------------ 176| 476k| const int have_topright = (!have_top || x + tw >= w) ? 0 : ------------------ | Branch (176:40): [True: 8, False: 476k] | Branch (176:53): [True: 4.64k, False: 471k] ------------------ 177| 476k| (edge_flags & EDGE_I444_TOP_HAS_RIGHT); 178| | 179| 476k| if (have_topright) { ------------------ | Branch (179:17): [True: 427k, False: 48.8k] ------------------ 180| 427k| const int px_have = imin(sz, (w - x - tw) << 2); 181| | 182| 427k| pixel_copy(top + sz, &dst_top[sz], px_have); ------------------ | | 65| 427k|#define pixel_copy(a, b, c) memcpy(a, b, (c) << 1) ------------------ 183| 427k| if (px_have < sz) ------------------ | Branch (183:21): [True: 1.55k, False: 425k] ------------------ 184| 1.55k| pixel_set(top + sz + px_have, top[sz + px_have - 1], 185| 1.55k| sz - px_have); 186| 427k| } else { 187| 48.8k| pixel_set(top + sz, top[sz - 1], sz); 188| 48.8k| } 189| 476k| } 190| 14.0M| } 191| | 192| 16.4M| if (av1_intra_prediction_edges[mode].needs_topleft) { ------------------ | Branch (192:9): [True: 3.44M, False: 12.9M] ------------------ 193| 3.44M| if (have_left) ------------------ | Branch (193:13): [True: 3.42M, False: 16.0k] ------------------ 194| 3.42M| *topleft_out = have_top ? dst_top[-1] : dst[-1]; ------------------ | Branch (194:28): [True: 3.36M, False: 65.2k] ------------------ 195| 16.0k| else 196| 16.0k| *topleft_out = have_top ? *dst_top : (1 << bitdepth) >> 1; ------------------ | Branch (196:28): [True: 13.4k, False: 2.69k] ------------------ 197| | 198| 3.44M| if (mode == Z2_PRED && tw + th >= 6 && filter_edge) ------------------ | Branch (198:13): [True: 908k, False: 2.53M] | Branch (198:32): [True: 52.0k, False: 856k] | Branch (198:48): [True: 48.2k, False: 3.77k] ------------------ 199| 48.2k| *topleft_out = ((topleft_out[-1] + topleft_out[1]) * 5 + 200| 48.2k| topleft_out[0] * 6 + 8) >> 4; 201| 3.44M| } 202| | 203| 16.4M| return mode; 204| 16.4M|} dav1d_intra_pred_dsp_init_8bpc: 744| 3.46k|COLD void bitfn(dav1d_intra_pred_dsp_init)(Dav1dIntraPredDSPContext *const c) { 745| 3.46k| c->intra_pred[DC_PRED ] = ipred_dc_c; 746| 3.46k| c->intra_pred[DC_128_PRED ] = ipred_dc_128_c; 747| 3.46k| c->intra_pred[TOP_DC_PRED ] = ipred_dc_top_c; 748| 3.46k| c->intra_pred[LEFT_DC_PRED ] = ipred_dc_left_c; 749| 3.46k| c->intra_pred[HOR_PRED ] = ipred_h_c; 750| 3.46k| c->intra_pred[VERT_PRED ] = ipred_v_c; 751| 3.46k| c->intra_pred[PAETH_PRED ] = ipred_paeth_c; 752| 3.46k| c->intra_pred[SMOOTH_PRED ] = ipred_smooth_c; 753| 3.46k| c->intra_pred[SMOOTH_V_PRED] = ipred_smooth_v_c; 754| 3.46k| c->intra_pred[SMOOTH_H_PRED] = ipred_smooth_h_c; 755| 3.46k| c->intra_pred[Z1_PRED ] = ipred_z1_c; 756| 3.46k| c->intra_pred[Z2_PRED ] = ipred_z2_c; 757| 3.46k| c->intra_pred[Z3_PRED ] = ipred_z3_c; 758| 3.46k| c->intra_pred[FILTER_PRED ] = ipred_filter_c; 759| | 760| 3.46k| c->cfl_ac[DAV1D_PIXEL_LAYOUT_I420 - 1] = cfl_ac_420_c; 761| 3.46k| c->cfl_ac[DAV1D_PIXEL_LAYOUT_I422 - 1] = cfl_ac_422_c; 762| 3.46k| c->cfl_ac[DAV1D_PIXEL_LAYOUT_I444 - 1] = cfl_ac_444_c; 763| | 764| 3.46k| c->cfl_pred[DC_PRED ] = ipred_cfl_c; 765| 3.46k| c->cfl_pred[DC_128_PRED ] = ipred_cfl_128_c; 766| 3.46k| c->cfl_pred[TOP_DC_PRED ] = ipred_cfl_top_c; 767| 3.46k| c->cfl_pred[LEFT_DC_PRED] = ipred_cfl_left_c; 768| | 769| 3.46k| c->pal_pred = pal_pred_c; 770| | 771| 3.46k|#if HAVE_ASM 772| |#if ARCH_AARCH64 || ARCH_ARM 773| | intra_pred_dsp_init_arm(c); 774| |#elif ARCH_RISCV 775| | intra_pred_dsp_init_riscv(c); 776| |#elif ARCH_X86 777| | intra_pred_dsp_init_x86(c); 778| |#elif ARCH_LOONGARCH64 779| | intra_pred_dsp_init_loongarch(c); 780| |#endif 781| 3.46k|#endif 782| 3.46k|} dav1d_intra_pred_dsp_init_16bpc: 744| 5.11k|COLD void bitfn(dav1d_intra_pred_dsp_init)(Dav1dIntraPredDSPContext *const c) { 745| 5.11k| c->intra_pred[DC_PRED ] = ipred_dc_c; 746| 5.11k| c->intra_pred[DC_128_PRED ] = ipred_dc_128_c; 747| 5.11k| c->intra_pred[TOP_DC_PRED ] = ipred_dc_top_c; 748| 5.11k| c->intra_pred[LEFT_DC_PRED ] = ipred_dc_left_c; 749| 5.11k| c->intra_pred[HOR_PRED ] = ipred_h_c; 750| 5.11k| c->intra_pred[VERT_PRED ] = ipred_v_c; 751| 5.11k| c->intra_pred[PAETH_PRED ] = ipred_paeth_c; 752| 5.11k| c->intra_pred[SMOOTH_PRED ] = ipred_smooth_c; 753| 5.11k| c->intra_pred[SMOOTH_V_PRED] = ipred_smooth_v_c; 754| 5.11k| c->intra_pred[SMOOTH_H_PRED] = ipred_smooth_h_c; 755| 5.11k| c->intra_pred[Z1_PRED ] = ipred_z1_c; 756| 5.11k| c->intra_pred[Z2_PRED ] = ipred_z2_c; 757| 5.11k| c->intra_pred[Z3_PRED ] = ipred_z3_c; 758| 5.11k| c->intra_pred[FILTER_PRED ] = ipred_filter_c; 759| | 760| 5.11k| c->cfl_ac[DAV1D_PIXEL_LAYOUT_I420 - 1] = cfl_ac_420_c; 761| 5.11k| c->cfl_ac[DAV1D_PIXEL_LAYOUT_I422 - 1] = cfl_ac_422_c; 762| 5.11k| c->cfl_ac[DAV1D_PIXEL_LAYOUT_I444 - 1] = cfl_ac_444_c; 763| | 764| 5.11k| c->cfl_pred[DC_PRED ] = ipred_cfl_c; 765| 5.11k| c->cfl_pred[DC_128_PRED ] = ipred_cfl_128_c; 766| 5.11k| c->cfl_pred[TOP_DC_PRED ] = ipred_cfl_top_c; 767| 5.11k| c->cfl_pred[LEFT_DC_PRED] = ipred_cfl_left_c; 768| | 769| 5.11k| c->pal_pred = pal_pred_c; 770| | 771| 5.11k|#if HAVE_ASM 772| |#if ARCH_AARCH64 || ARCH_ARM 773| | intra_pred_dsp_init_arm(c); 774| |#elif ARCH_RISCV 775| | intra_pred_dsp_init_riscv(c); 776| |#elif ARCH_X86 777| | intra_pred_dsp_init_x86(c); 778| |#elif ARCH_LOONGARCH64 779| | intra_pred_dsp_init_loongarch(c); 780| |#endif 781| 5.11k|#endif 782| 5.11k|} itx_1d.c:inv_dct4_1d_internal_c: 68| 1.67M|{ 69| 1.67M| assert(stride > 0); ------------------ | Branch (69:5): [True: 1.67M, False: 2] ------------------ 70| 1.67M| const int in0 = c[0 * stride], in1 = c[1 * stride]; 71| | 72| 1.67M| int t0, t1, t2, t3; 73| 1.67M| if (tx64) { ------------------ | Branch (73:9): [True: 903k, False: 775k] ------------------ 74| 903k| t0 = t1 = (in0 * 181 + 128) >> 8; 75| 903k| t2 = (in1 * 1567 + 2048) >> 12; 76| 903k| t3 = (in1 * 3784 + 2048) >> 12; 77| 903k| } else { 78| 775k| const int in2 = c[2 * stride], in3 = c[3 * stride]; 79| | 80| 775k| t0 = ((in0 + in2) * 181 + 128) >> 8; 81| 775k| t1 = ((in0 - in2) * 181 + 128) >> 8; 82| 775k| t2 = ((in1 * 1567 - in3 * (3784 - 4096) + 2048) >> 12) - in3; 83| 775k| t3 = ((in1 * (3784 - 4096) + in3 * 1567 + 2048) >> 12) + in1; 84| 775k| } 85| | 86| 1.67M| c[0 * stride] = CLIP(t0 + t3); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 87| 1.67M| c[1 * stride] = CLIP(t1 + t2); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 88| 1.67M| c[2 * stride] = CLIP(t1 - t2); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 89| 1.67M| c[3 * stride] = CLIP(t0 - t3); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 90| 1.67M|} itx_1d.c:inv_dct8_1d_internal_c: 101| 1.67M|{ 102| 1.67M| assert(stride > 0); ------------------ | Branch (102:5): [True: 1.67M, False: 18.4E] ------------------ 103| 1.67M| inv_dct4_1d_internal_c(c, stride << 1, min, max, tx64); 104| | 105| 1.67M| const int in1 = c[1 * stride], in3 = c[3 * stride]; 106| | 107| 1.67M| int t4a, t5a, t6a, t7a; 108| 1.67M| if (tx64) { ------------------ | Branch (108:9): [True: 903k, False: 776k] ------------------ 109| 903k| t4a = (in1 * 799 + 2048) >> 12; 110| 903k| t5a = (in3 * -2276 + 2048) >> 12; 111| 903k| t6a = (in3 * 3406 + 2048) >> 12; 112| 903k| t7a = (in1 * 4017 + 2048) >> 12; 113| 903k| } else { 114| 776k| const int in5 = c[5 * stride], in7 = c[7 * stride]; 115| | 116| 776k| t4a = ((in1 * 799 - in7 * (4017 - 4096) + 2048) >> 12) - in7; 117| 776k| t5a = (in5 * 1703 - in3 * 1138 + 1024) >> 11; 118| 776k| t6a = (in5 * 1138 + in3 * 1703 + 1024) >> 11; 119| 776k| t7a = ((in1 * (4017 - 4096) + in7 * 799 + 2048) >> 12) + in1; 120| 776k| } 121| | 122| 1.67M| const int t4 = CLIP(t4a + t5a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 123| 1.67M| t5a = CLIP(t4a - t5a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 124| 1.67M| const int t7 = CLIP(t7a + t6a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 125| 1.67M| t6a = CLIP(t7a - t6a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 126| | 127| 1.67M| const int t5 = ((t6a - t5a) * 181 + 128) >> 8; 128| 1.67M| const int t6 = ((t6a + t5a) * 181 + 128) >> 8; 129| | 130| 1.67M| const int t0 = c[0 * stride]; 131| 1.67M| const int t1 = c[2 * stride]; 132| 1.67M| const int t2 = c[4 * stride]; 133| 1.67M| const int t3 = c[6 * stride]; 134| | 135| 1.67M| c[0 * stride] = CLIP(t0 + t7); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 136| 1.67M| c[1 * stride] = CLIP(t1 + t6); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 137| 1.67M| c[2 * stride] = CLIP(t2 + t5); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 138| 1.67M| c[3 * stride] = CLIP(t3 + t4); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 139| 1.67M| c[4 * stride] = CLIP(t3 - t4); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 140| 1.67M| c[5 * stride] = CLIP(t2 - t5); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 141| 1.67M| c[6 * stride] = CLIP(t1 - t6); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 142| 1.67M| c[7 * stride] = CLIP(t0 - t7); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 143| 1.67M|} itx_1d.c:inv_dct16_1d_c: 242| 200k|{ 243| 200k| inv_dct16_1d_internal_c(c, stride, min, max, 0); 244| 200k|} itx_1d.c:inv_dct16_1d_internal_c: 154| 1.67M|{ 155| 1.67M| assert(stride > 0); ------------------ | Branch (155:5): [True: 1.67M, False: 18.4E] ------------------ 156| 1.67M| inv_dct8_1d_internal_c(c, stride << 1, min, max, tx64); 157| | 158| 1.67M| const int in1 = c[1 * stride], in3 = c[3 * stride]; 159| 1.67M| const int in5 = c[5 * stride], in7 = c[7 * stride]; 160| | 161| 1.67M| int t8a, t9a, t10a, t11a, t12a, t13a, t14a, t15a; 162| 1.67M| if (tx64) { ------------------ | Branch (162:9): [True: 904k, False: 775k] ------------------ 163| 904k| t8a = (in1 * 401 + 2048) >> 12; 164| 904k| t9a = (in7 * -2598 + 2048) >> 12; 165| 904k| t10a = (in5 * 1931 + 2048) >> 12; 166| 904k| t11a = (in3 * -1189 + 2048) >> 12; 167| 904k| t12a = (in3 * 3920 + 2048) >> 12; 168| 904k| t13a = (in5 * 3612 + 2048) >> 12; 169| 904k| t14a = (in7 * 3166 + 2048) >> 12; 170| 904k| t15a = (in1 * 4076 + 2048) >> 12; 171| 904k| } else { 172| 775k| const int in9 = c[ 9 * stride], in11 = c[11 * stride]; 173| 775k| const int in13 = c[13 * stride], in15 = c[15 * stride]; 174| | 175| 775k| t8a = ((in1 * 401 - in15 * (4076 - 4096) + 2048) >> 12) - in15; 176| 775k| t9a = (in9 * 1583 - in7 * 1299 + 1024) >> 11; 177| 775k| t10a = ((in5 * 1931 - in11 * (3612 - 4096) + 2048) >> 12) - in11; 178| 775k| t11a = ((in13 * (3920 - 4096) - in3 * 1189 + 2048) >> 12) + in13; 179| 775k| t12a = ((in13 * 1189 + in3 * (3920 - 4096) + 2048) >> 12) + in3; 180| 775k| t13a = ((in5 * (3612 - 4096) + in11 * 1931 + 2048) >> 12) + in5; 181| 775k| t14a = (in9 * 1299 + in7 * 1583 + 1024) >> 11; 182| 775k| t15a = ((in1 * (4076 - 4096) + in15 * 401 + 2048) >> 12) + in1; 183| 775k| } 184| | 185| 1.67M| int t8 = CLIP(t8a + t9a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 186| 1.67M| int t9 = CLIP(t8a - t9a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 187| 1.67M| int t10 = CLIP(t11a - t10a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 188| 1.67M| int t11 = CLIP(t11a + t10a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 189| 1.67M| int t12 = CLIP(t12a + t13a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 190| 1.67M| int t13 = CLIP(t12a - t13a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 191| 1.67M| int t14 = CLIP(t15a - t14a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 192| 1.67M| int t15 = CLIP(t15a + t14a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 193| | 194| 1.67M| t9a = (( t14 * 1567 - t9 * (3784 - 4096) + 2048) >> 12) - t9; 195| 1.67M| t14a = (( t14 * (3784 - 4096) + t9 * 1567 + 2048) >> 12) + t14; 196| 1.67M| t10a = ((-(t13 * (3784 - 4096) + t10 * 1567) + 2048) >> 12) - t13; 197| 1.67M| t13a = (( t13 * 1567 - t10 * (3784 - 4096) + 2048) >> 12) - t10; 198| | 199| 1.67M| t8a = CLIP(t8 + t11); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 200| 1.67M| t9 = CLIP(t9a + t10a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 201| 1.67M| t10 = CLIP(t9a - t10a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 202| 1.67M| t11a = CLIP(t8 - t11); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 203| 1.67M| t12a = CLIP(t15 - t12); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 204| 1.67M| t13 = CLIP(t14a - t13a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 205| 1.67M| t14 = CLIP(t14a + t13a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 206| 1.67M| t15a = CLIP(t15 + t12); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 207| | 208| 1.67M| t10a = ((t13 - t10) * 181 + 128) >> 8; 209| 1.67M| t13a = ((t13 + t10) * 181 + 128) >> 8; 210| 1.67M| t11 = ((t12a - t11a) * 181 + 128) >> 8; 211| 1.67M| t12 = ((t12a + t11a) * 181 + 128) >> 8; 212| | 213| 1.67M| const int t0 = c[ 0 * stride]; 214| 1.67M| const int t1 = c[ 2 * stride]; 215| 1.67M| const int t2 = c[ 4 * stride]; 216| 1.67M| const int t3 = c[ 6 * stride]; 217| 1.67M| const int t4 = c[ 8 * stride]; 218| 1.67M| const int t5 = c[10 * stride]; 219| 1.67M| const int t6 = c[12 * stride]; 220| 1.67M| const int t7 = c[14 * stride]; 221| | 222| 1.67M| c[ 0 * stride] = CLIP(t0 + t15a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 223| 1.67M| c[ 1 * stride] = CLIP(t1 + t14); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 224| 1.67M| c[ 2 * stride] = CLIP(t2 + t13a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 225| 1.67M| c[ 3 * stride] = CLIP(t3 + t12); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 226| 1.67M| c[ 4 * stride] = CLIP(t4 + t11); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 227| 1.67M| c[ 5 * stride] = CLIP(t5 + t10a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 228| 1.67M| c[ 6 * stride] = CLIP(t6 + t9); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 229| 1.67M| c[ 7 * stride] = CLIP(t7 + t8a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 230| 1.67M| c[ 8 * stride] = CLIP(t7 - t8a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 231| 1.67M| c[ 9 * stride] = CLIP(t6 - t9); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 232| 1.67M| c[10 * stride] = CLIP(t5 - t10a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 233| 1.67M| c[11 * stride] = CLIP(t4 - t11); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 234| 1.67M| c[12 * stride] = CLIP(t3 - t12); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 235| 1.67M| c[13 * stride] = CLIP(t2 - t13a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 236| 1.67M| c[14 * stride] = CLIP(t1 - t14); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 237| 1.67M| c[15 * stride] = CLIP(t0 - t15a); ------------------ | | 37| 1.67M|#define CLIP(a) iclip(a, min, max) ------------------ 238| 1.67M|} itx_1d.c:inv_dct32_1d_c: 432| 575k|{ 433| 575k| inv_dct32_1d_internal_c(c, stride, min, max, 0); 434| 575k|} itx_1d.c:inv_dct32_1d_internal_c: 249| 1.47M|{ 250| 1.47M| assert(stride > 0); ------------------ | Branch (250:5): [True: 1.47M, False: 18.4E] ------------------ 251| 1.47M| inv_dct16_1d_internal_c(c, stride << 1, min, max, tx64); 252| | 253| 1.47M| const int in1 = c[ 1 * stride], in3 = c[ 3 * stride]; 254| 1.47M| const int in5 = c[ 5 * stride], in7 = c[ 7 * stride]; 255| 1.47M| const int in9 = c[ 9 * stride], in11 = c[11 * stride]; 256| 1.47M| const int in13 = c[13 * stride], in15 = c[15 * stride]; 257| | 258| 1.47M| int t16a, t17a, t18a, t19a, t20a, t21a, t22a, t23a; 259| 1.47M| int t24a, t25a, t26a, t27a, t28a, t29a, t30a, t31a; 260| 1.47M| if (tx64) { ------------------ | Branch (260:9): [True: 905k, False: 574k] ------------------ 261| 905k| t16a = (in1 * 201 + 2048) >> 12; 262| 905k| t17a = (in15 * -2751 + 2048) >> 12; 263| 905k| t18a = (in9 * 1751 + 2048) >> 12; 264| 905k| t19a = (in7 * -1380 + 2048) >> 12; 265| 905k| t20a = (in5 * 995 + 2048) >> 12; 266| 905k| t21a = (in11 * -2106 + 2048) >> 12; 267| 905k| t22a = (in13 * 2440 + 2048) >> 12; 268| 905k| t23a = (in3 * -601 + 2048) >> 12; 269| 905k| t24a = (in3 * 4052 + 2048) >> 12; 270| 905k| t25a = (in13 * 3290 + 2048) >> 12; 271| 905k| t26a = (in11 * 3513 + 2048) >> 12; 272| 905k| t27a = (in5 * 3973 + 2048) >> 12; 273| 905k| t28a = (in7 * 3857 + 2048) >> 12; 274| 905k| t29a = (in9 * 3703 + 2048) >> 12; 275| 905k| t30a = (in15 * 3035 + 2048) >> 12; 276| 905k| t31a = (in1 * 4091 + 2048) >> 12; 277| 905k| } else { 278| 574k| const int in17 = c[17 * stride], in19 = c[19 * stride]; 279| 574k| const int in21 = c[21 * stride], in23 = c[23 * stride]; 280| 574k| const int in25 = c[25 * stride], in27 = c[27 * stride]; 281| 574k| const int in29 = c[29 * stride], in31 = c[31 * stride]; 282| | 283| 574k| t16a = ((in1 * 201 - in31 * (4091 - 4096) + 2048) >> 12) - in31; 284| 574k| t17a = ((in17 * (3035 - 4096) - in15 * 2751 + 2048) >> 12) + in17; 285| 574k| t18a = ((in9 * 1751 - in23 * (3703 - 4096) + 2048) >> 12) - in23; 286| 574k| t19a = ((in25 * (3857 - 4096) - in7 * 1380 + 2048) >> 12) + in25; 287| 574k| t20a = ((in5 * 995 - in27 * (3973 - 4096) + 2048) >> 12) - in27; 288| 574k| t21a = ((in21 * (3513 - 4096) - in11 * 2106 + 2048) >> 12) + in21; 289| 574k| t22a = (in13 * 1220 - in19 * 1645 + 1024) >> 11; 290| 574k| t23a = ((in29 * (4052 - 4096) - in3 * 601 + 2048) >> 12) + in29; 291| 574k| t24a = ((in29 * 601 + in3 * (4052 - 4096) + 2048) >> 12) + in3; 292| 574k| t25a = (in13 * 1645 + in19 * 1220 + 1024) >> 11; 293| 574k| t26a = ((in21 * 2106 + in11 * (3513 - 4096) + 2048) >> 12) + in11; 294| 574k| t27a = ((in5 * (3973 - 4096) + in27 * 995 + 2048) >> 12) + in5; 295| 574k| t28a = ((in25 * 1380 + in7 * (3857 - 4096) + 2048) >> 12) + in7; 296| 574k| t29a = ((in9 * (3703 - 4096) + in23 * 1751 + 2048) >> 12) + in9; 297| 574k| t30a = ((in17 * 2751 + in15 * (3035 - 4096) + 2048) >> 12) + in15; 298| 574k| t31a = ((in1 * (4091 - 4096) + in31 * 201 + 2048) >> 12) + in1; 299| 574k| } 300| | 301| 1.47M| int t16 = CLIP(t16a + t17a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 302| 1.47M| int t17 = CLIP(t16a - t17a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 303| 1.47M| int t18 = CLIP(t19a - t18a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 304| 1.47M| int t19 = CLIP(t19a + t18a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 305| 1.47M| int t20 = CLIP(t20a + t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 306| 1.47M| int t21 = CLIP(t20a - t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 307| 1.47M| int t22 = CLIP(t23a - t22a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 308| 1.47M| int t23 = CLIP(t23a + t22a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 309| 1.47M| int t24 = CLIP(t24a + t25a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 310| 1.47M| int t25 = CLIP(t24a - t25a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 311| 1.47M| int t26 = CLIP(t27a - t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 312| 1.47M| int t27 = CLIP(t27a + t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 313| 1.47M| int t28 = CLIP(t28a + t29a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 314| 1.47M| int t29 = CLIP(t28a - t29a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 315| 1.47M| int t30 = CLIP(t31a - t30a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 316| 1.47M| int t31 = CLIP(t31a + t30a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 317| | 318| 1.47M| t17a = (( t30 * 799 - t17 * (4017 - 4096) + 2048) >> 12) - t17; 319| 1.47M| t30a = (( t30 * (4017 - 4096) + t17 * 799 + 2048) >> 12) + t30; 320| 1.47M| t18a = ((-(t29 * (4017 - 4096) + t18 * 799) + 2048) >> 12) - t29; 321| 1.47M| t29a = (( t29 * 799 - t18 * (4017 - 4096) + 2048) >> 12) - t18; 322| 1.47M| t21a = ( t26 * 1703 - t21 * 1138 + 1024) >> 11; 323| 1.47M| t26a = ( t26 * 1138 + t21 * 1703 + 1024) >> 11; 324| 1.47M| t22a = (-(t25 * 1138 + t22 * 1703 ) + 1024) >> 11; 325| 1.47M| t25a = ( t25 * 1703 - t22 * 1138 + 1024) >> 11; 326| | 327| 1.47M| t16a = CLIP(t16 + t19); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 328| 1.47M| t17 = CLIP(t17a + t18a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 329| 1.47M| t18 = CLIP(t17a - t18a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 330| 1.47M| t19a = CLIP(t16 - t19); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 331| 1.47M| t20a = CLIP(t23 - t20); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 332| 1.47M| t21 = CLIP(t22a - t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 333| 1.47M| t22 = CLIP(t22a + t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 334| 1.47M| t23a = CLIP(t23 + t20); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 335| 1.47M| t24a = CLIP(t24 + t27); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 336| 1.47M| t25 = CLIP(t25a + t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 337| 1.47M| t26 = CLIP(t25a - t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 338| 1.47M| t27a = CLIP(t24 - t27); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 339| 1.47M| t28a = CLIP(t31 - t28); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 340| 1.47M| t29 = CLIP(t30a - t29a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 341| 1.47M| t30 = CLIP(t30a + t29a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 342| 1.47M| t31a = CLIP(t31 + t28); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 343| | 344| 1.47M| t18a = (( t29 * 1567 - t18 * (3784 - 4096) + 2048) >> 12) - t18; 345| 1.47M| t29a = (( t29 * (3784 - 4096) + t18 * 1567 + 2048) >> 12) + t29; 346| 1.47M| t19 = (( t28a * 1567 - t19a * (3784 - 4096) + 2048) >> 12) - t19a; 347| 1.47M| t28 = (( t28a * (3784 - 4096) + t19a * 1567 + 2048) >> 12) + t28a; 348| 1.47M| t20 = ((-(t27a * (3784 - 4096) + t20a * 1567) + 2048) >> 12) - t27a; 349| 1.47M| t27 = (( t27a * 1567 - t20a * (3784 - 4096) + 2048) >> 12) - t20a; 350| 1.47M| t21a = ((-(t26 * (3784 - 4096) + t21 * 1567) + 2048) >> 12) - t26; 351| 1.47M| t26a = (( t26 * 1567 - t21 * (3784 - 4096) + 2048) >> 12) - t21; 352| | 353| 1.47M| t16 = CLIP(t16a + t23a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 354| 1.47M| t17a = CLIP(t17 + t22); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 355| 1.47M| t18 = CLIP(t18a + t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 356| 1.47M| t19a = CLIP(t19 + t20); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 357| 1.47M| t20a = CLIP(t19 - t20); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 358| 1.47M| t21 = CLIP(t18a - t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 359| 1.47M| t22a = CLIP(t17 - t22); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 360| 1.47M| t23 = CLIP(t16a - t23a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 361| 1.47M| t24 = CLIP(t31a - t24a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 362| 1.47M| t25a = CLIP(t30 - t25); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 363| 1.47M| t26 = CLIP(t29a - t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 364| 1.47M| t27a = CLIP(t28 - t27); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 365| 1.47M| t28a = CLIP(t28 + t27); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 366| 1.47M| t29 = CLIP(t29a + t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 367| 1.47M| t30a = CLIP(t30 + t25); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 368| 1.47M| t31 = CLIP(t31a + t24a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 369| | 370| 1.47M| t20 = ((t27a - t20a) * 181 + 128) >> 8; 371| 1.47M| t27 = ((t27a + t20a) * 181 + 128) >> 8; 372| 1.47M| t21a = ((t26 - t21 ) * 181 + 128) >> 8; 373| 1.47M| t26a = ((t26 + t21 ) * 181 + 128) >> 8; 374| 1.47M| t22 = ((t25a - t22a) * 181 + 128) >> 8; 375| 1.47M| t25 = ((t25a + t22a) * 181 + 128) >> 8; 376| 1.47M| t23a = ((t24 - t23 ) * 181 + 128) >> 8; 377| 1.47M| t24a = ((t24 + t23 ) * 181 + 128) >> 8; 378| | 379| 1.47M| const int t0 = c[ 0 * stride]; 380| 1.47M| const int t1 = c[ 2 * stride]; 381| 1.47M| const int t2 = c[ 4 * stride]; 382| 1.47M| const int t3 = c[ 6 * stride]; 383| 1.47M| const int t4 = c[ 8 * stride]; 384| 1.47M| const int t5 = c[10 * stride]; 385| 1.47M| const int t6 = c[12 * stride]; 386| 1.47M| const int t7 = c[14 * stride]; 387| 1.47M| const int t8 = c[16 * stride]; 388| 1.47M| const int t9 = c[18 * stride]; 389| 1.47M| const int t10 = c[20 * stride]; 390| 1.47M| const int t11 = c[22 * stride]; 391| 1.47M| const int t12 = c[24 * stride]; 392| 1.47M| const int t13 = c[26 * stride]; 393| 1.47M| const int t14 = c[28 * stride]; 394| 1.47M| const int t15 = c[30 * stride]; 395| | 396| 1.47M| c[ 0 * stride] = CLIP(t0 + t31); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 397| 1.47M| c[ 1 * stride] = CLIP(t1 + t30a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 398| 1.47M| c[ 2 * stride] = CLIP(t2 + t29); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 399| 1.47M| c[ 3 * stride] = CLIP(t3 + t28a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 400| 1.47M| c[ 4 * stride] = CLIP(t4 + t27); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 401| 1.47M| c[ 5 * stride] = CLIP(t5 + t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 402| 1.47M| c[ 6 * stride] = CLIP(t6 + t25); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 403| 1.47M| c[ 7 * stride] = CLIP(t7 + t24a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 404| 1.47M| c[ 8 * stride] = CLIP(t8 + t23a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 405| 1.47M| c[ 9 * stride] = CLIP(t9 + t22); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 406| 1.47M| c[10 * stride] = CLIP(t10 + t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 407| 1.47M| c[11 * stride] = CLIP(t11 + t20); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 408| 1.47M| c[12 * stride] = CLIP(t12 + t19a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 409| 1.47M| c[13 * stride] = CLIP(t13 + t18); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 410| 1.47M| c[14 * stride] = CLIP(t14 + t17a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 411| 1.47M| c[15 * stride] = CLIP(t15 + t16); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 412| 1.47M| c[16 * stride] = CLIP(t15 - t16); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 413| 1.47M| c[17 * stride] = CLIP(t14 - t17a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 414| 1.47M| c[18 * stride] = CLIP(t13 - t18); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 415| 1.47M| c[19 * stride] = CLIP(t12 - t19a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 416| 1.47M| c[20 * stride] = CLIP(t11 - t20); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 417| 1.47M| c[21 * stride] = CLIP(t10 - t21a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 418| 1.47M| c[22 * stride] = CLIP(t9 - t22); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 419| 1.47M| c[23 * stride] = CLIP(t8 - t23a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 420| 1.47M| c[24 * stride] = CLIP(t7 - t24a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 421| 1.47M| c[25 * stride] = CLIP(t6 - t25); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 422| 1.47M| c[26 * stride] = CLIP(t5 - t26a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 423| 1.47M| c[27 * stride] = CLIP(t4 - t27); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 424| 1.47M| c[28 * stride] = CLIP(t3 - t28a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 425| 1.47M| c[29 * stride] = CLIP(t2 - t29); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 426| 1.47M| c[30 * stride] = CLIP(t1 - t30a); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 427| 1.47M| c[31 * stride] = CLIP(t0 - t31); ------------------ | | 37| 1.47M|#define CLIP(a) iclip(a, min, max) ------------------ 428| 1.47M|} itx_1d.c:inv_dct64_1d_c: 438| 904k|{ 439| 904k| assert(stride > 0); ------------------ | Branch (439:5): [True: 904k, False: 18.4E] ------------------ 440| 904k| inv_dct32_1d_internal_c(c, stride << 1, min, max, 1); 441| | 442| 904k| const int in1 = c[ 1 * stride], in3 = c[ 3 * stride]; 443| 904k| const int in5 = c[ 5 * stride], in7 = c[ 7 * stride]; 444| 904k| const int in9 = c[ 9 * stride], in11 = c[11 * stride]; 445| 904k| const int in13 = c[13 * stride], in15 = c[15 * stride]; 446| 904k| const int in17 = c[17 * stride], in19 = c[19 * stride]; 447| 904k| const int in21 = c[21 * stride], in23 = c[23 * stride]; 448| 904k| const int in25 = c[25 * stride], in27 = c[27 * stride]; 449| 904k| const int in29 = c[29 * stride], in31 = c[31 * stride]; 450| | 451| 904k| int t32a = (in1 * 101 + 2048) >> 12; 452| 904k| int t33a = (in31 * -2824 + 2048) >> 12; 453| 904k| int t34a = (in17 * 1660 + 2048) >> 12; 454| 904k| int t35a = (in15 * -1474 + 2048) >> 12; 455| 904k| int t36a = (in9 * 897 + 2048) >> 12; 456| 904k| int t37a = (in23 * -2191 + 2048) >> 12; 457| 904k| int t38a = (in25 * 2359 + 2048) >> 12; 458| 904k| int t39a = (in7 * -700 + 2048) >> 12; 459| 904k| int t40a = (in5 * 501 + 2048) >> 12; 460| 904k| int t41a = (in27 * -2520 + 2048) >> 12; 461| 904k| int t42a = (in21 * 2019 + 2048) >> 12; 462| 904k| int t43a = (in11 * -1092 + 2048) >> 12; 463| 904k| int t44a = (in13 * 1285 + 2048) >> 12; 464| 904k| int t45a = (in19 * -1842 + 2048) >> 12; 465| 904k| int t46a = (in29 * 2675 + 2048) >> 12; 466| 904k| int t47a = (in3 * -301 + 2048) >> 12; 467| 904k| int t48a = (in3 * 4085 + 2048) >> 12; 468| 904k| int t49a = (in29 * 3102 + 2048) >> 12; 469| 904k| int t50a = (in19 * 3659 + 2048) >> 12; 470| 904k| int t51a = (in13 * 3889 + 2048) >> 12; 471| 904k| int t52a = (in11 * 3948 + 2048) >> 12; 472| 904k| int t53a = (in21 * 3564 + 2048) >> 12; 473| 904k| int t54a = (in27 * 3229 + 2048) >> 12; 474| 904k| int t55a = (in5 * 4065 + 2048) >> 12; 475| 904k| int t56a = (in7 * 4036 + 2048) >> 12; 476| 904k| int t57a = (in25 * 3349 + 2048) >> 12; 477| 904k| int t58a = (in23 * 3461 + 2048) >> 12; 478| 904k| int t59a = (in9 * 3996 + 2048) >> 12; 479| 904k| int t60a = (in15 * 3822 + 2048) >> 12; 480| 904k| int t61a = (in17 * 3745 + 2048) >> 12; 481| 904k| int t62a = (in31 * 2967 + 2048) >> 12; 482| 904k| int t63a = (in1 * 4095 + 2048) >> 12; 483| | 484| 904k| int t32 = CLIP(t32a + t33a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 485| 904k| int t33 = CLIP(t32a - t33a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 486| 904k| int t34 = CLIP(t35a - t34a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 487| 904k| int t35 = CLIP(t35a + t34a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 488| 904k| int t36 = CLIP(t36a + t37a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 489| 904k| int t37 = CLIP(t36a - t37a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 490| 904k| int t38 = CLIP(t39a - t38a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 491| 904k| int t39 = CLIP(t39a + t38a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 492| 904k| int t40 = CLIP(t40a + t41a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 493| 904k| int t41 = CLIP(t40a - t41a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 494| 904k| int t42 = CLIP(t43a - t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 495| 904k| int t43 = CLIP(t43a + t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 496| 904k| int t44 = CLIP(t44a + t45a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 497| 904k| int t45 = CLIP(t44a - t45a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 498| 904k| int t46 = CLIP(t47a - t46a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 499| 904k| int t47 = CLIP(t47a + t46a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 500| 904k| int t48 = CLIP(t48a + t49a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 501| 904k| int t49 = CLIP(t48a - t49a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 502| 904k| int t50 = CLIP(t51a - t50a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 503| 904k| int t51 = CLIP(t51a + t50a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 504| 904k| int t52 = CLIP(t52a + t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 505| 904k| int t53 = CLIP(t52a - t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 506| 904k| int t54 = CLIP(t55a - t54a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 507| 904k| int t55 = CLIP(t55a + t54a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 508| 904k| int t56 = CLIP(t56a + t57a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 509| 904k| int t57 = CLIP(t56a - t57a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 510| 904k| int t58 = CLIP(t59a - t58a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 511| 904k| int t59 = CLIP(t59a + t58a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 512| 904k| int t60 = CLIP(t60a + t61a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 513| 904k| int t61 = CLIP(t60a - t61a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 514| 904k| int t62 = CLIP(t63a - t62a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 515| 904k| int t63 = CLIP(t63a + t62a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 516| | 517| 904k| t33a = ((t33 * (4096 - 4076) + t62 * 401 + 2048) >> 12) - t33; 518| 904k| t34a = ((t34 * -401 + t61 * (4096 - 4076) + 2048) >> 12) - t61; 519| 904k| t37a = (t37 * -1299 + t58 * 1583 + 1024) >> 11; 520| 904k| t38a = (t38 * -1583 + t57 * -1299 + 1024) >> 11; 521| 904k| t41a = ((t41 * (4096 - 3612) + t54 * 1931 + 2048) >> 12) - t41; 522| 904k| t42a = ((t42 * -1931 + t53 * (4096 - 3612) + 2048) >> 12) - t53; 523| 904k| t45a = ((t45 * -1189 + t50 * (3920 - 4096) + 2048) >> 12) + t50; 524| 904k| t46a = ((t46 * (4096 - 3920) + t49 * -1189 + 2048) >> 12) - t46; 525| 904k| t49a = ((t46 * -1189 + t49 * (3920 - 4096) + 2048) >> 12) + t49; 526| 904k| t50a = ((t45 * (3920 - 4096) + t50 * 1189 + 2048) >> 12) + t45; 527| 904k| t53a = ((t42 * (4096 - 3612) + t53 * 1931 + 2048) >> 12) - t42; 528| 904k| t54a = ((t41 * 1931 + t54 * (3612 - 4096) + 2048) >> 12) + t54; 529| 904k| t57a = (t38 * -1299 + t57 * 1583 + 1024) >> 11; 530| 904k| t58a = (t37 * 1583 + t58 * 1299 + 1024) >> 11; 531| 904k| t61a = ((t34 * (4096 - 4076) + t61 * 401 + 2048) >> 12) - t34; 532| 904k| t62a = ((t33 * 401 + t62 * (4076 - 4096) + 2048) >> 12) + t62; 533| | 534| 904k| t32a = CLIP(t32 + t35); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 535| 904k| t33 = CLIP(t33a + t34a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 536| 904k| t34 = CLIP(t33a - t34a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 537| 904k| t35a = CLIP(t32 - t35); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 538| 904k| t36a = CLIP(t39 - t36); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 539| 904k| t37 = CLIP(t38a - t37a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 540| 904k| t38 = CLIP(t38a + t37a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 541| 904k| t39a = CLIP(t39 + t36); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 542| 904k| t40a = CLIP(t40 + t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 543| 904k| t41 = CLIP(t41a + t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 544| 904k| t42 = CLIP(t41a - t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 545| 904k| t43a = CLIP(t40 - t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 546| 904k| t44a = CLIP(t47 - t44); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 547| 904k| t45 = CLIP(t46a - t45a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 548| 904k| t46 = CLIP(t46a + t45a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 549| 904k| t47a = CLIP(t47 + t44); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 550| 904k| t48a = CLIP(t48 + t51); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 551| 904k| t49 = CLIP(t49a + t50a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 552| 904k| t50 = CLIP(t49a - t50a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 553| 904k| t51a = CLIP(t48 - t51); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 554| 904k| t52a = CLIP(t55 - t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 555| 904k| t53 = CLIP(t54a - t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 556| 904k| t54 = CLIP(t54a + t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 557| 904k| t55a = CLIP(t55 + t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 558| 904k| t56a = CLIP(t56 + t59); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 559| 904k| t57 = CLIP(t57a + t58a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 560| 904k| t58 = CLIP(t57a - t58a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 561| 904k| t59a = CLIP(t56 - t59); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 562| 904k| t60a = CLIP(t63 - t60); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 563| 904k| t61 = CLIP(t62a - t61a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 564| 904k| t62 = CLIP(t62a + t61a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 565| 904k| t63a = CLIP(t63 + t60); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 566| | 567| 904k| t34a = ((t34 * (4096 - 4017) + t61 * 799 + 2048) >> 12) - t34; 568| 904k| t35 = ((t35a * (4096 - 4017) + t60a * 799 + 2048) >> 12) - t35a; 569| 904k| t36 = ((t36a * -799 + t59a * (4096 - 4017) + 2048) >> 12) - t59a; 570| 904k| t37a = ((t37 * -799 + t58 * (4096 - 4017) + 2048) >> 12) - t58; 571| 904k| t42a = (t42 * -1138 + t53 * 1703 + 1024) >> 11; 572| 904k| t43 = (t43a * -1138 + t52a * 1703 + 1024) >> 11; 573| 904k| t44 = (t44a * -1703 + t51a * -1138 + 1024) >> 11; 574| 904k| t45a = (t45 * -1703 + t50 * -1138 + 1024) >> 11; 575| 904k| t50a = (t45 * -1138 + t50 * 1703 + 1024) >> 11; 576| 904k| t51 = (t44a * -1138 + t51a * 1703 + 1024) >> 11; 577| 904k| t52 = (t43a * 1703 + t52a * 1138 + 1024) >> 11; 578| 904k| t53a = (t42 * 1703 + t53 * 1138 + 1024) >> 11; 579| 904k| t58a = ((t37 * (4096 - 4017) + t58 * 799 + 2048) >> 12) - t37; 580| 904k| t59 = ((t36a * (4096 - 4017) + t59a * 799 + 2048) >> 12) - t36a; 581| 904k| t60 = ((t35a * 799 + t60a * (4017 - 4096) + 2048) >> 12) + t60a; 582| 904k| t61a = ((t34 * 799 + t61 * (4017 - 4096) + 2048) >> 12) + t61; 583| | 584| 904k| t32 = CLIP(t32a + t39a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 585| 904k| t33a = CLIP(t33 + t38); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 586| 904k| t34 = CLIP(t34a + t37a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 587| 904k| t35a = CLIP(t35 + t36); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 588| 904k| t36a = CLIP(t35 - t36); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 589| 904k| t37 = CLIP(t34a - t37a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 590| 904k| t38a = CLIP(t33 - t38); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 591| 904k| t39 = CLIP(t32a - t39a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 592| 904k| t40 = CLIP(t47a - t40a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 593| 904k| t41a = CLIP(t46 - t41); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 594| 904k| t42 = CLIP(t45a - t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 595| 904k| t43a = CLIP(t44 - t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 596| 904k| t44a = CLIP(t44 + t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 597| 904k| t45 = CLIP(t45a + t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 598| 904k| t46a = CLIP(t46 + t41); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 599| 904k| t47 = CLIP(t47a + t40a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 600| 904k| t48 = CLIP(t48a + t55a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 601| 904k| t49a = CLIP(t49 + t54); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 602| 904k| t50 = CLIP(t50a + t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 603| 904k| t51a = CLIP(t51 + t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 604| 904k| t52a = CLIP(t51 - t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 605| 904k| t53 = CLIP(t50a - t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 606| 904k| t54a = CLIP(t49 - t54); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 607| 904k| t55 = CLIP(t48a - t55a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 608| 904k| t56 = CLIP(t63a - t56a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 609| 904k| t57a = CLIP(t62 - t57); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 610| 904k| t58 = CLIP(t61a - t58a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 611| 904k| t59a = CLIP(t60 - t59); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 612| 904k| t60a = CLIP(t60 + t59); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 613| 904k| t61 = CLIP(t61a + t58a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 614| 904k| t62a = CLIP(t62 + t57); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 615| 904k| t63 = CLIP(t63a + t56a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 616| | 617| 904k| t36 = ((t36a * (4096 - 3784) + t59a * 1567 + 2048) >> 12) - t36a; 618| 904k| t37a = ((t37 * (4096 - 3784) + t58 * 1567 + 2048) >> 12) - t37; 619| 904k| t38 = ((t38a * (4096 - 3784) + t57a * 1567 + 2048) >> 12) - t38a; 620| 904k| t39a = ((t39 * (4096 - 3784) + t56 * 1567 + 2048) >> 12) - t39; 621| 904k| t40a = ((t40 * -1567 + t55 * (4096 - 3784) + 2048) >> 12) - t55; 622| 904k| t41 = ((t41a * -1567 + t54a * (4096 - 3784) + 2048) >> 12) - t54a; 623| 904k| t42a = ((t42 * -1567 + t53 * (4096 - 3784) + 2048) >> 12) - t53; 624| 904k| t43 = ((t43a * -1567 + t52a * (4096 - 3784) + 2048) >> 12) - t52a; 625| 904k| t52 = ((t43a * (4096 - 3784) + t52a * 1567 + 2048) >> 12) - t43a; 626| 904k| t53a = ((t42 * (4096 - 3784) + t53 * 1567 + 2048) >> 12) - t42; 627| 904k| t54 = ((t41a * (4096 - 3784) + t54a * 1567 + 2048) >> 12) - t41a; 628| 904k| t55a = ((t40 * (4096 - 3784) + t55 * 1567 + 2048) >> 12) - t40; 629| 904k| t56a = ((t39 * 1567 + t56 * (3784 - 4096) + 2048) >> 12) + t56; 630| 904k| t57 = ((t38a * 1567 + t57a * (3784 - 4096) + 2048) >> 12) + t57a; 631| 904k| t58a = ((t37 * 1567 + t58 * (3784 - 4096) + 2048) >> 12) + t58; 632| 904k| t59 = ((t36a * 1567 + t59a * (3784 - 4096) + 2048) >> 12) + t59a; 633| | 634| 904k| t32a = CLIP(t32 + t47); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 635| 904k| t33 = CLIP(t33a + t46a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 636| 904k| t34a = CLIP(t34 + t45); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 637| 904k| t35 = CLIP(t35a + t44a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 638| 904k| t36a = CLIP(t36 + t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 639| 904k| t37 = CLIP(t37a + t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 640| 904k| t38a = CLIP(t38 + t41); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 641| 904k| t39 = CLIP(t39a + t40a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 642| 904k| t40 = CLIP(t39a - t40a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 643| 904k| t41a = CLIP(t38 - t41); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 644| 904k| t42 = CLIP(t37a - t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 645| 904k| t43a = CLIP(t36 - t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 646| 904k| t44 = CLIP(t35a - t44a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 647| 904k| t45a = CLIP(t34 - t45); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 648| 904k| t46 = CLIP(t33a - t46a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 649| 904k| t47a = CLIP(t32 - t47); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 650| 904k| t48a = CLIP(t63 - t48); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 651| 904k| t49 = CLIP(t62a - t49a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 652| 904k| t50a = CLIP(t61 - t50); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 653| 904k| t51 = CLIP(t60a - t51a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 654| 904k| t52a = CLIP(t59 - t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 655| 904k| t53 = CLIP(t58a - t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 656| 904k| t54a = CLIP(t57 - t54); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 657| 904k| t55 = CLIP(t56a - t55a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 658| 904k| t56 = CLIP(t56a + t55a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 659| 904k| t57a = CLIP(t57 + t54); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 660| 904k| t58 = CLIP(t58a + t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 661| 904k| t59a = CLIP(t59 + t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 662| 904k| t60 = CLIP(t60a + t51a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 663| 904k| t61a = CLIP(t61 + t50); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 664| 904k| t62 = CLIP(t62a + t49a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 665| 904k| t63a = CLIP(t63 + t48); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 666| | 667| 904k| t40a = ((t55 - t40 ) * 181 + 128) >> 8; 668| 904k| t41 = ((t54a - t41a) * 181 + 128) >> 8; 669| 904k| t42a = ((t53 - t42 ) * 181 + 128) >> 8; 670| 904k| t43 = ((t52a - t43a) * 181 + 128) >> 8; 671| 904k| t44a = ((t51 - t44 ) * 181 + 128) >> 8; 672| 904k| t45 = ((t50a - t45a) * 181 + 128) >> 8; 673| 904k| t46a = ((t49 - t46 ) * 181 + 128) >> 8; 674| 904k| t47 = ((t48a - t47a) * 181 + 128) >> 8; 675| 904k| t48 = ((t47a + t48a) * 181 + 128) >> 8; 676| 904k| t49a = ((t46 + t49 ) * 181 + 128) >> 8; 677| 904k| t50 = ((t45a + t50a) * 181 + 128) >> 8; 678| 904k| t51a = ((t44 + t51 ) * 181 + 128) >> 8; 679| 904k| t52 = ((t43a + t52a) * 181 + 128) >> 8; 680| 904k| t53a = ((t42 + t53 ) * 181 + 128) >> 8; 681| 904k| t54 = ((t41a + t54a) * 181 + 128) >> 8; 682| 904k| t55a = ((t40 + t55 ) * 181 + 128) >> 8; 683| | 684| 904k| const int t0 = c[ 0 * stride]; 685| 904k| const int t1 = c[ 2 * stride]; 686| 904k| const int t2 = c[ 4 * stride]; 687| 904k| const int t3 = c[ 6 * stride]; 688| 904k| const int t4 = c[ 8 * stride]; 689| 904k| const int t5 = c[10 * stride]; 690| 904k| const int t6 = c[12 * stride]; 691| 904k| const int t7 = c[14 * stride]; 692| 904k| const int t8 = c[16 * stride]; 693| 904k| const int t9 = c[18 * stride]; 694| 904k| const int t10 = c[20 * stride]; 695| 904k| const int t11 = c[22 * stride]; 696| 904k| const int t12 = c[24 * stride]; 697| 904k| const int t13 = c[26 * stride]; 698| 904k| const int t14 = c[28 * stride]; 699| 904k| const int t15 = c[30 * stride]; 700| 904k| const int t16 = c[32 * stride]; 701| 904k| const int t17 = c[34 * stride]; 702| 904k| const int t18 = c[36 * stride]; 703| 904k| const int t19 = c[38 * stride]; 704| 904k| const int t20 = c[40 * stride]; 705| 904k| const int t21 = c[42 * stride]; 706| 904k| const int t22 = c[44 * stride]; 707| 904k| const int t23 = c[46 * stride]; 708| 904k| const int t24 = c[48 * stride]; 709| 904k| const int t25 = c[50 * stride]; 710| 904k| const int t26 = c[52 * stride]; 711| 904k| const int t27 = c[54 * stride]; 712| 904k| const int t28 = c[56 * stride]; 713| 904k| const int t29 = c[58 * stride]; 714| 904k| const int t30 = c[60 * stride]; 715| 904k| const int t31 = c[62 * stride]; 716| | 717| 904k| c[ 0 * stride] = CLIP(t0 + t63a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 718| 904k| c[ 1 * stride] = CLIP(t1 + t62); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 719| 904k| c[ 2 * stride] = CLIP(t2 + t61a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 720| 904k| c[ 3 * stride] = CLIP(t3 + t60); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 721| 904k| c[ 4 * stride] = CLIP(t4 + t59a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 722| 904k| c[ 5 * stride] = CLIP(t5 + t58); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 723| 904k| c[ 6 * stride] = CLIP(t6 + t57a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 724| 904k| c[ 7 * stride] = CLIP(t7 + t56); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 725| 904k| c[ 8 * stride] = CLIP(t8 + t55a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 726| 904k| c[ 9 * stride] = CLIP(t9 + t54); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 727| 904k| c[10 * stride] = CLIP(t10 + t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 728| 904k| c[11 * stride] = CLIP(t11 + t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 729| 904k| c[12 * stride] = CLIP(t12 + t51a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 730| 904k| c[13 * stride] = CLIP(t13 + t50); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 731| 904k| c[14 * stride] = CLIP(t14 + t49a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 732| 904k| c[15 * stride] = CLIP(t15 + t48); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 733| 904k| c[16 * stride] = CLIP(t16 + t47); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 734| 904k| c[17 * stride] = CLIP(t17 + t46a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 735| 904k| c[18 * stride] = CLIP(t18 + t45); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 736| 904k| c[19 * stride] = CLIP(t19 + t44a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 737| 904k| c[20 * stride] = CLIP(t20 + t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 738| 904k| c[21 * stride] = CLIP(t21 + t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 739| 904k| c[22 * stride] = CLIP(t22 + t41); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 740| 904k| c[23 * stride] = CLIP(t23 + t40a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 741| 904k| c[24 * stride] = CLIP(t24 + t39); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 742| 904k| c[25 * stride] = CLIP(t25 + t38a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 743| 904k| c[26 * stride] = CLIP(t26 + t37); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 744| 904k| c[27 * stride] = CLIP(t27 + t36a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 745| 904k| c[28 * stride] = CLIP(t28 + t35); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 746| 904k| c[29 * stride] = CLIP(t29 + t34a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 747| 904k| c[30 * stride] = CLIP(t30 + t33); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 748| 904k| c[31 * stride] = CLIP(t31 + t32a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 749| 904k| c[32 * stride] = CLIP(t31 - t32a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 750| 904k| c[33 * stride] = CLIP(t30 - t33); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 751| 904k| c[34 * stride] = CLIP(t29 - t34a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 752| 904k| c[35 * stride] = CLIP(t28 - t35); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 753| 904k| c[36 * stride] = CLIP(t27 - t36a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 754| 904k| c[37 * stride] = CLIP(t26 - t37); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 755| 904k| c[38 * stride] = CLIP(t25 - t38a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 756| 904k| c[39 * stride] = CLIP(t24 - t39); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 757| 904k| c[40 * stride] = CLIP(t23 - t40a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 758| 904k| c[41 * stride] = CLIP(t22 - t41); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 759| 904k| c[42 * stride] = CLIP(t21 - t42a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 760| 904k| c[43 * stride] = CLIP(t20 - t43); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 761| 904k| c[44 * stride] = CLIP(t19 - t44a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 762| 904k| c[45 * stride] = CLIP(t18 - t45); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 763| 904k| c[46 * stride] = CLIP(t17 - t46a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 764| 904k| c[47 * stride] = CLIP(t16 - t47); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 765| 904k| c[48 * stride] = CLIP(t15 - t48); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 766| 904k| c[49 * stride] = CLIP(t14 - t49a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 767| 904k| c[50 * stride] = CLIP(t13 - t50); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 768| 904k| c[51 * stride] = CLIP(t12 - t51a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 769| 904k| c[52 * stride] = CLIP(t11 - t52); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 770| 904k| c[53 * stride] = CLIP(t10 - t53a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 771| 904k| c[54 * stride] = CLIP(t9 - t54); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 772| 904k| c[55 * stride] = CLIP(t8 - t55a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 773| 904k| c[56 * stride] = CLIP(t7 - t56); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 774| 904k| c[57 * stride] = CLIP(t6 - t57a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 775| 904k| c[58 * stride] = CLIP(t5 - t58); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 776| 904k| c[59 * stride] = CLIP(t4 - t59a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 777| 904k| c[60 * stride] = CLIP(t3 - t60); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 778| 904k| c[61 * stride] = CLIP(t2 - t61a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 779| 904k| c[62 * stride] = CLIP(t1 - t62); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 780| 904k| c[63 * stride] = CLIP(t0 - t63a); ------------------ | | 37| 904k|#define CLIP(a) iclip(a, min, max) ------------------ 781| 904k|} dav1d_itx_dsp_init_8bpc: 220| 3.46k|COLD void bitfn(dav1d_itx_dsp_init)(Dav1dInvTxfmDSPContext *const c, int bpc) { 221| 3.46k|#define assign_itx_all_fn64(w, h, pfx) \ 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c 224| | 225| 3.46k|#define assign_itx_all_fn32(w, h, pfx) \ 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c 229| | 230| 3.46k|#define assign_itx_all_fn16(w, h, pfx) \ 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c 252| | 253| 3.46k|#define assign_itx_all_fn84(w, h, pfx) \ 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ 263| 3.46k| 264| 3.46k|#if !(HAVE_ASM && TRIM_DSP_FUNCTIONS && ( \ 265| 3.46k| ARCH_AARCH64 || \ 266| 3.46k| (ARCH_ARM && (defined(__ARM_NEON) || defined(__APPLE__) || defined(_WIN32))) \ 267| 3.46k|)) 268| 3.46k| c->itxfm_add[TX_4X4][WHT_WHT] = inv_txfm_add_wht_wht_4x4_c; 269| 3.46k|#endif 270| 3.46k| assign_itx_all_fn84( 4, 4, ); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 271| 3.46k| assign_itx_all_fn84( 4, 8, R); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 272| 3.46k| assign_itx_all_fn84( 4, 16, R); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 273| 3.46k| assign_itx_all_fn84( 8, 4, R); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 274| 3.46k| assign_itx_all_fn84( 8, 8, ); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 275| 3.46k| assign_itx_all_fn84( 8, 16, R); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 276| 3.46k| assign_itx_all_fn32( 8, 32, R); ------------------ | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 277| 3.46k| assign_itx_all_fn84(16, 4, R); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 278| 3.46k| assign_itx_all_fn84(16, 8, R); ------------------ | | 254| 3.46k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 3.46k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 3.46k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 3.46k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 3.46k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 279| 3.46k| assign_itx_all_fn16(16, 16, ); ------------------ | | 231| 3.46k| assign_itx_all_fn32(w, h, pfx); \ | | ------------------ | | | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | | | ------------------ | | | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | ------------------ | | | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c | | ------------------ | | 232| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | 233| 3.46k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | 234| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | 235| 3.46k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | 236| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | 237| 3.46k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | 238| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | 239| 3.46k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | 240| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | 241| 3.46k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | 242| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | 243| 3.46k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | 244| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | 245| 3.46k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | 246| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | 247| 3.46k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | 248| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | 249| 3.46k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | 250| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | 251| 3.46k| inv_txfm_add_identity_dct_##w##x##h##_c ------------------ 280| 3.46k| assign_itx_all_fn32(16, 32, R); ------------------ | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 281| 3.46k| assign_itx_all_fn64(16, 64, R); ------------------ | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 282| 3.46k| assign_itx_all_fn32(32, 8, R); ------------------ | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 283| 3.46k| assign_itx_all_fn32(32, 16, R); ------------------ | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 284| 3.46k| assign_itx_all_fn32(32, 32, ); ------------------ | | 226| 3.46k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 3.46k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 285| 3.46k| assign_itx_all_fn64(32, 64, R); ------------------ | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 286| 3.46k| assign_itx_all_fn64(64, 16, R); ------------------ | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 287| 3.46k| assign_itx_all_fn64(64, 32, R); ------------------ | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 288| 3.46k| assign_itx_all_fn64(64, 64, ); ------------------ | | 222| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 3.46k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 289| | 290| 3.46k| int all_simd = 0; 291| 3.46k|#if HAVE_ASM 292| |#if ARCH_AARCH64 || ARCH_ARM 293| | itx_dsp_init_arm(c, bpc, &all_simd); 294| |#endif 295| |#if ARCH_LOONGARCH64 296| | itx_dsp_init_loongarch(c, bpc); 297| |#endif 298| |#if ARCH_PPC64LE 299| | itx_dsp_init_ppc(c, bpc); 300| |#endif 301| |#if ARCH_RISCV 302| | itx_dsp_init_riscv(c, bpc); 303| |#endif 304| 3.46k|#if ARCH_X86 305| 3.46k| itx_dsp_init_x86(c, bpc, &all_simd); 306| 3.46k|#endif 307| 3.46k|#endif 308| | 309| 3.46k| if (!all_simd) ------------------ | Branch (309:9): [True: 0, False: 3.46k] ------------------ 310| 0| dav1d_init_last_nonzero_col_from_eob_tables(); 311| 3.46k|} itx_tmpl.c:inv_txfm_add_c: 47| 55.6k|{ 48| 55.6k| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[tx]; 49| 55.6k| const int w = 4 * t_dim->w, h = 4 * t_dim->h; 50| 55.6k| const int has_dconly = txtp == DCT_DCT; 51| 55.6k| assert(w >= 4 && w <= 64); ------------------ | Branch (51:5): [True: 55.6k, False: 18.4E] | Branch (51:5): [True: 55.6k, False: 18.4E] ------------------ 52| 55.6k| assert(h >= 4 && h <= 64); ------------------ | Branch (52:5): [True: 55.6k, False: 1] | Branch (52:5): [True: 55.6k, False: 0] ------------------ 53| 55.6k| assert(eob >= 0); ------------------ | Branch (53:5): [True: 55.6k, False: 0] ------------------ 54| | 55| 55.6k| const int is_rect2 = w * 2 == h || h * 2 == w; ------------------ | Branch (55:26): [True: 7.40k, False: 48.2k] | Branch (55:40): [True: 8.99k, False: 39.2k] ------------------ 56| 55.6k| const int rnd = (1 << shift) >> 1; 57| | 58| 55.6k| if (eob < has_dconly) { ------------------ | Branch (58:9): [True: 24.7k, False: 30.9k] ------------------ 59| 24.7k| int dc = coeff[0]; 60| 24.7k| coeff[0] = 0; 61| 24.7k| if (is_rect2) ------------------ | Branch (61:13): [True: 6.99k, False: 17.7k] ------------------ 62| 6.99k| dc = (dc * 181 + 128) >> 8; 63| 24.7k| dc = (dc * 181 + 128) >> 8; 64| 24.7k| dc = (dc + rnd) >> shift; 65| 24.7k| dc = (dc * 181 + 128 + 2048) >> 12; 66| 1.03M| for (int y = 0; y < h; y++, dst += PXSTRIDE(stride)) ------------------ | | 53| 1.01M|#define PXSTRIDE(x) (x) ------------------ | Branch (66:25): [True: 1.01M, False: 24.7k] ------------------ 67| 49.5M| for (int x = 0; x < w; x++) ------------------ | Branch (67:29): [True: 48.5M, False: 1.01M] ------------------ 68| 48.5M| dst[x] = iclip_pixel(dst[x] + dc); ------------------ | | 49| 48.5M|#define iclip_pixel iclip_u8 ------------------ 69| 24.7k| return; 70| 24.7k| } 71| | 72| 30.9k| const uint8_t *const txtps = dav1d_tx1d_types[txtp]; 73| 30.9k| const itx_1d_fn first_1d_fn = dav1d_tx1d_fns[t_dim->lw][txtps[0]]; 74| 30.9k| const itx_1d_fn second_1d_fn = dav1d_tx1d_fns[t_dim->lh][txtps[1]]; 75| 30.9k| const int sh = imin(h, 32), sw = imin(w, 32); 76| 30.9k|#if BITDEPTH == 8 77| 30.9k| const int row_clip_min = INT16_MIN; 78| 30.9k| const int col_clip_min = INT16_MIN; 79| |#else 80| | const int row_clip_min = (int) ((unsigned) ~bitdepth_max << 7); 81| | const int col_clip_min = (int) ((unsigned) ~bitdepth_max << 5); 82| |#endif 83| 30.9k| const int row_clip_max = ~row_clip_min; 84| 30.9k| const int col_clip_max = ~col_clip_min; 85| | 86| 30.9k| int32_t tmp[64 * 64], *c = tmp; 87| 30.9k| int last_nonzero_col; // in first 1d itx 88| 30.9k| if (txtps[1] == IDENTITY && txtps[0] != IDENTITY) { ------------------ | Branch (88:9): [True: 0, False: 30.9k] | Branch (88:33): [True: 0, False: 0] ------------------ 89| 0| last_nonzero_col = imin(sh - 1, eob); 90| 30.9k| } else if (txtps[0] == IDENTITY && txtps[1] != IDENTITY) { ------------------ | Branch (90:16): [True: 0, False: 30.9k] | Branch (90:40): [True: 0, False: 0] ------------------ 91| 0| last_nonzero_col = eob >> (t_dim->lw + 2); 92| 30.9k| } else { 93| 30.9k| last_nonzero_col = dav1d_last_nonzero_col_from_eob[tx][eob]; 94| 30.9k| } 95| 30.9k| assert(last_nonzero_col < sh); ------------------ | Branch (95:5): [True: 30.9k, False: 18.4E] ------------------ 96| 401k| for (int y = 0; y <= last_nonzero_col; y++, c += w) { ------------------ | Branch (96:21): [True: 370k, False: 30.9k] ------------------ 97| 370k| if (is_rect2) ------------------ | Branch (97:13): [True: 101k, False: 269k] ------------------ 98| 2.90M| for (int x = 0; x < sw; x++) ------------------ | Branch (98:29): [True: 2.80M, False: 101k] ------------------ 99| 2.80M| c[x] = (coeff[y + x * sh] * 181 + 128) >> 8; 100| 269k| else 101| 8.67M| for (int x = 0; x < sw; x++) ------------------ | Branch (101:29): [True: 8.40M, False: 269k] ------------------ 102| 8.40M| c[x] = coeff[y + x * sh]; 103| 370k| first_1d_fn(c, 1, row_clip_min, row_clip_max); 104| 370k| } 105| 30.9k| if (last_nonzero_col + 1 < sh) ------------------ | Branch (105:9): [True: 25.9k, False: 5.04k] ------------------ 106| 25.9k| memset(c, 0, sizeof(*c) * (sh - last_nonzero_col - 1) * w); 107| | 108| 30.9k| memset(coeff, 0, sizeof(*coeff) * sw * sh); 109| 39.3M| for (int i = 0; i < w * sh; i++) ------------------ | Branch (109:21): [True: 39.2M, False: 30.9k] ------------------ 110| 39.2M| tmp[i] = iclip((tmp[i] + rnd) >> shift, col_clip_min, col_clip_max); 111| | 112| 1.34M| for (int x = 0; x < w; x++) ------------------ | Branch (112:21): [True: 1.30M, False: 30.9k] ------------------ 113| 1.30M| second_1d_fn(&tmp[x], w, col_clip_min, col_clip_max); 114| | 115| 30.9k| c = tmp; 116| 1.38M| for (int y = 0; y < h; y++, dst += PXSTRIDE(stride)) ------------------ | | 53| 1.35M|#define PXSTRIDE(x) (x) ------------------ | Branch (116:21): [True: 1.35M, False: 30.9k] ------------------ 117| 63.5M| for (int x = 0; x < w; x++) ------------------ | Branch (117:25): [True: 62.1M, False: 1.35M] ------------------ 118| 62.1M| dst[x] = iclip_pixel(dst[x] + ((*c++ + 8) >> 4)); ------------------ | | 49| 62.1M|#define iclip_pixel iclip_u8 ------------------ 119| 30.9k|} itx_tmpl.c:inv_txfm_add_dct_dct_16x32_c: 127| 4.45k| HIGHBD_DECL_SUFFIX) \ 128| 4.45k|{ \ 129| 4.45k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 4.45k| HIGHBD_TAIL_SUFFIX); \ 131| 4.45k|} itx_tmpl.c:inv_txfm_add_dct_dct_16x64_c: 127| 1.26k| HIGHBD_DECL_SUFFIX) \ 128| 1.26k|{ \ 129| 1.26k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 1.26k| HIGHBD_TAIL_SUFFIX); \ 131| 1.26k|} itx_tmpl.c:inv_txfm_add_dct_dct_32x16_c: 127| 6.32k| HIGHBD_DECL_SUFFIX) \ 128| 6.32k|{ \ 129| 6.32k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 6.32k| HIGHBD_TAIL_SUFFIX); \ 131| 6.32k|} itx_tmpl.c:inv_txfm_add_dct_dct_32x32_c: 127| 18.8k| HIGHBD_DECL_SUFFIX) \ 128| 18.8k|{ \ 129| 18.8k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 18.8k| HIGHBD_TAIL_SUFFIX); \ 131| 18.8k|} itx_tmpl.c:inv_txfm_add_dct_dct_32x64_c: 127| 2.94k| HIGHBD_DECL_SUFFIX) \ 128| 2.94k|{ \ 129| 2.94k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 2.94k| HIGHBD_TAIL_SUFFIX); \ 131| 2.94k|} itx_tmpl.c:inv_txfm_add_dct_dct_64x16_c: 127| 1.02k| HIGHBD_DECL_SUFFIX) \ 128| 1.02k|{ \ 129| 1.02k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 1.02k| HIGHBD_TAIL_SUFFIX); \ 131| 1.02k|} itx_tmpl.c:inv_txfm_add_dct_dct_64x32_c: 127| 2.68k| HIGHBD_DECL_SUFFIX) \ 128| 2.68k|{ \ 129| 2.68k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 2.68k| HIGHBD_TAIL_SUFFIX); \ 131| 2.68k|} itx_tmpl.c:inv_txfm_add_dct_dct_64x64_c: 127| 18.1k| HIGHBD_DECL_SUFFIX) \ 128| 18.1k|{ \ 129| 18.1k| inv_txfm_add_c(dst, stride, coeff, eob, pfx##TX_##w##X##h, shift, type \ 130| 18.1k| HIGHBD_TAIL_SUFFIX); \ 131| 18.1k|} dav1d_itx_dsp_init_16bpc: 220| 5.11k|COLD void bitfn(dav1d_itx_dsp_init)(Dav1dInvTxfmDSPContext *const c, int bpc) { 221| 5.11k|#define assign_itx_all_fn64(w, h, pfx) \ 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c 224| | 225| 5.11k|#define assign_itx_all_fn32(w, h, pfx) \ 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c 229| | 230| 5.11k|#define assign_itx_all_fn16(w, h, pfx) \ 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c 252| | 253| 5.11k|#define assign_itx_all_fn84(w, h, pfx) \ 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ 263| 5.11k| 264| 5.11k|#if !(HAVE_ASM && TRIM_DSP_FUNCTIONS && ( \ 265| 5.11k| ARCH_AARCH64 || \ 266| 5.11k| (ARCH_ARM && (defined(__ARM_NEON) || defined(__APPLE__) || defined(_WIN32))) \ 267| 5.11k|)) 268| 5.11k| c->itxfm_add[TX_4X4][WHT_WHT] = inv_txfm_add_wht_wht_4x4_c; 269| 5.11k|#endif 270| 5.11k| assign_itx_all_fn84( 4, 4, ); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 271| 5.11k| assign_itx_all_fn84( 4, 8, R); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 272| 5.11k| assign_itx_all_fn84( 4, 16, R); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 273| 5.11k| assign_itx_all_fn84( 8, 4, R); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 274| 5.11k| assign_itx_all_fn84( 8, 8, ); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 275| 5.11k| assign_itx_all_fn84( 8, 16, R); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 276| 5.11k| assign_itx_all_fn32( 8, 32, R); ------------------ | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 277| 5.11k| assign_itx_all_fn84(16, 4, R); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 278| 5.11k| assign_itx_all_fn84(16, 8, R); ------------------ | | 254| 5.11k| assign_itx_all_fn16(w, h, pfx); \ | | ------------------ | | | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | | | ------------------ | | | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | | | ------------------ | | | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | | | ------------------ | | | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | | | ------------------ | | | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c | | ------------------ | | 255| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_FLIPADST] = \ | | 256| 5.11k| inv_txfm_add_flipadst_identity_##w##x##h##_c; \ | | 257| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_FLIPADST] = \ | | 258| 5.11k| inv_txfm_add_identity_flipadst_##w##x##h##_c; \ | | 259| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_ADST] = \ | | 260| 5.11k| inv_txfm_add_adst_identity_##w##x##h##_c; \ | | 261| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_ADST] = \ | | 262| 5.11k| inv_txfm_add_identity_adst_##w##x##h##_c; \ ------------------ 279| 5.11k| assign_itx_all_fn16(16, 16, ); ------------------ | | 231| 5.11k| assign_itx_all_fn32(w, h, pfx); \ | | ------------------ | | | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | | | ------------------ | | | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | | | ------------------ | | | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c | | ------------------ | | 232| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_ADST ] = \ | | 233| 5.11k| inv_txfm_add_adst_dct_##w##x##h##_c; \ | | 234| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_DCT ] = \ | | 235| 5.11k| inv_txfm_add_dct_adst_##w##x##h##_c; \ | | 236| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_ADST] = \ | | 237| 5.11k| inv_txfm_add_adst_adst_##w##x##h##_c; \ | | 238| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][ADST_FLIPADST] = \ | | 239| 5.11k| inv_txfm_add_flipadst_adst_##w##x##h##_c; \ | | 240| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_ADST] = \ | | 241| 5.11k| inv_txfm_add_adst_flipadst_##w##x##h##_c; \ | | 242| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_FLIPADST] = \ | | 243| 5.11k| inv_txfm_add_flipadst_dct_##w##x##h##_c; \ | | 244| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_DCT] = \ | | 245| 5.11k| inv_txfm_add_dct_flipadst_##w##x##h##_c; \ | | 246| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][FLIPADST_FLIPADST] = \ | | 247| 5.11k| inv_txfm_add_flipadst_flipadst_##w##x##h##_c; \ | | 248| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][H_DCT] = \ | | 249| 5.11k| inv_txfm_add_dct_identity_##w##x##h##_c; \ | | 250| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][V_DCT] = \ | | 251| 5.11k| inv_txfm_add_identity_dct_##w##x##h##_c ------------------ 280| 5.11k| assign_itx_all_fn32(16, 32, R); ------------------ | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 281| 5.11k| assign_itx_all_fn64(16, 64, R); ------------------ | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 282| 5.11k| assign_itx_all_fn32(32, 8, R); ------------------ | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 283| 5.11k| assign_itx_all_fn32(32, 16, R); ------------------ | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 284| 5.11k| assign_itx_all_fn32(32, 32, ); ------------------ | | 226| 5.11k| assign_itx_all_fn64(w, h, pfx); \ | | ------------------ | | | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c | | ------------------ | | 227| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][IDTX] = \ | | 228| 5.11k| inv_txfm_add_identity_identity_##w##x##h##_c ------------------ 285| 5.11k| assign_itx_all_fn64(32, 64, R); ------------------ | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 286| 5.11k| assign_itx_all_fn64(64, 16, R); ------------------ | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 287| 5.11k| assign_itx_all_fn64(64, 32, R); ------------------ | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 288| 5.11k| assign_itx_all_fn64(64, 64, ); ------------------ | | 222| 5.11k| c->itxfm_add[pfx##TX_##w##X##h][DCT_DCT ] = \ | | 223| 5.11k| inv_txfm_add_dct_dct_##w##x##h##_c ------------------ 289| | 290| 5.11k| int all_simd = 0; 291| 5.11k|#if HAVE_ASM 292| |#if ARCH_AARCH64 || ARCH_ARM 293| | itx_dsp_init_arm(c, bpc, &all_simd); 294| |#endif 295| |#if ARCH_LOONGARCH64 296| | itx_dsp_init_loongarch(c, bpc); 297| |#endif 298| |#if ARCH_PPC64LE 299| | itx_dsp_init_ppc(c, bpc); 300| |#endif 301| |#if ARCH_RISCV 302| | itx_dsp_init_riscv(c, bpc); 303| |#endif 304| 5.11k|#if ARCH_X86 305| 5.11k| itx_dsp_init_x86(c, bpc, &all_simd); 306| 5.11k|#endif 307| 5.11k|#endif 308| | 309| 5.11k| if (!all_simd) ------------------ | Branch (309:9): [True: 2.99k, False: 2.11k] ------------------ 310| 2.99k| dav1d_init_last_nonzero_col_from_eob_tables(); 311| 5.11k|} dav1d_copy_lpf_8bpc: 106| 241k|{ 107| 241k| const int have_tt = f->c->n_tc > 1; 108| 241k| const int resize = f->frame_hdr->width[0] != f->frame_hdr->width[1]; 109| 241k| const int offset = 8 * !!sby; 110| 241k| const ptrdiff_t *const src_stride = f->cur.stride; 111| 241k| const ptrdiff_t *const lr_stride = f->sr_cur.p.stride; 112| 241k| const int tt_off = have_tt * sby * (4 << f->seq_hdr->sb128); 113| 241k| pixel *const dst[3] = { 114| 241k| f->lf.lr_lpf_line[0] + tt_off * PXSTRIDE(lr_stride[0]), ------------------ | | 53| 241k|#define PXSTRIDE(x) (x) ------------------ 115| 241k| f->lf.lr_lpf_line[1] + tt_off * PXSTRIDE(lr_stride[1]), ------------------ | | 53| 241k|#define PXSTRIDE(x) (x) ------------------ 116| 241k| f->lf.lr_lpf_line[2] + tt_off * PXSTRIDE(lr_stride[1]) ------------------ | | 53| 241k|#define PXSTRIDE(x) (x) ------------------ 117| 241k| }; 118| | 119| | // TODO Also check block level restore type to reduce copying. 120| 241k| const int restore_planes = f->lf.restore_planes; 121| | 122| 241k| if (f->seq_hdr->cdef || restore_planes & LR_RESTORE_Y) { ------------------ | Branch (122:9): [True: 72.3k, False: 168k] | Branch (122:29): [True: 166k, False: 2.40k] ------------------ 123| 238k| const int h = f->cur.p.h; 124| 238k| const int w = f->bw << 2; 125| 238k| const int row_h = imin((sby + 1) << (6 + f->seq_hdr->sb128), h - 1); 126| 238k| const int y_stripe = (sby << (6 + f->seq_hdr->sb128)) - offset; 127| 238k| if (restore_planes & LR_RESTORE_Y || !resize) ------------------ | Branch (127:13): [True: 171k, False: 66.8k] | Branch (127:46): [True: 59.4k, False: 7.35k] ------------------ 128| 231k| backup_lpf(f, dst[0], lr_stride[0], 129| 231k| src[0] - offset * PXSTRIDE(src_stride[0]), src_stride[0], ------------------ | | 53| 231k|#define PXSTRIDE(x) (x) ------------------ 130| 231k| 0, f->seq_hdr->sb128, y_stripe, row_h, w, h, 0, 1); 131| 238k| if (have_tt && resize) { ------------------ | Branch (131:13): [True: 238k, False: 158] | Branch (131:24): [True: 9.98k, False: 228k] ------------------ 132| 9.98k| const ptrdiff_t cdef_off_y = sby * 4 * PXSTRIDE(src_stride[0]); ------------------ | | 53| 9.98k|#define PXSTRIDE(x) (x) ------------------ 133| 9.98k| backup_lpf(f, f->lf.cdef_lpf_line[0] + cdef_off_y, src_stride[0], 134| 9.98k| src[0] - offset * PXSTRIDE(src_stride[0]), src_stride[0], ------------------ | | 53| 9.98k|#define PXSTRIDE(x) (x) ------------------ 135| 9.98k| 0, f->seq_hdr->sb128, y_stripe, row_h, w, h, 0, 0); 136| 9.98k| } 137| 238k| } 138| 241k| if ((f->seq_hdr->cdef || restore_planes & (LR_RESTORE_U | LR_RESTORE_V)) && ------------------ | Branch (138:10): [True: 72.3k, False: 168k] | Branch (138:30): [True: 4.56k, False: 164k] ------------------ 139| 76.7k| f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) ------------------ | Branch (139:9): [True: 48.1k, False: 28.5k] ------------------ 140| 48.1k| { 141| 48.1k| const int ss_ver = f->sr_cur.p.p.layout == DAV1D_PIXEL_LAYOUT_I420; 142| 48.1k| const int ss_hor = f->sr_cur.p.p.layout != DAV1D_PIXEL_LAYOUT_I444; 143| 48.1k| const int h = (f->cur.p.h + ss_ver) >> ss_ver; 144| 48.1k| const int w = f->bw << (2 - ss_hor); 145| 48.1k| const int row_h = imin((sby + 1) << ((6 - ss_ver) + f->seq_hdr->sb128), h - 1); 146| 48.1k| const int offset_uv = offset >> ss_ver; 147| 48.1k| const int y_stripe = (sby << ((6 - ss_ver) + f->seq_hdr->sb128)) - offset_uv; 148| 48.1k| const ptrdiff_t cdef_off_uv = sby * 4 * PXSTRIDE(src_stride[1]); ------------------ | | 53| 48.1k|#define PXSTRIDE(x) (x) ------------------ 149| 48.1k| if (f->seq_hdr->cdef || restore_planes & LR_RESTORE_U) { ------------------ | Branch (149:13): [True: 43.6k, False: 4.56k] | Branch (149:33): [True: 3.39k, False: 1.16k] ------------------ 150| 47.0k| if (restore_planes & LR_RESTORE_U || !resize) ------------------ | Branch (150:17): [True: 6.22k, False: 40.7k] | Branch (150:50): [True: 34.1k, False: 6.64k] ------------------ 151| 40.3k| backup_lpf(f, dst[1], lr_stride[1], 152| 40.3k| src[1] - offset_uv * PXSTRIDE(src_stride[1]), ------------------ | | 53| 40.3k|#define PXSTRIDE(x) (x) ------------------ 153| 40.3k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 154| 40.3k| row_h, w, h, ss_hor, 1); 155| 47.0k| if (have_tt && resize) ------------------ | Branch (155:17): [True: 47.0k, False: 18.4E] | Branch (155:28): [True: 10.7k, False: 36.2k] ------------------ 156| 10.7k| backup_lpf(f, f->lf.cdef_lpf_line[1] + cdef_off_uv, src_stride[1], 157| 10.7k| src[1] - offset_uv * PXSTRIDE(src_stride[1]), ------------------ | | 53| 10.7k|#define PXSTRIDE(x) (x) ------------------ 158| 10.7k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 159| 10.7k| row_h, w, h, ss_hor, 0); 160| 47.0k| } 161| 48.1k| if (f->seq_hdr->cdef || restore_planes & LR_RESTORE_V) { ------------------ | Branch (161:13): [True: 43.6k, False: 4.56k] | Branch (161:33): [True: 3.59k, False: 971] ------------------ 162| 47.2k| if (restore_planes & LR_RESTORE_V || !resize) ------------------ | Branch (162:17): [True: 6.21k, False: 41.0k] | Branch (162:50): [True: 33.8k, False: 7.15k] ------------------ 163| 40.0k| backup_lpf(f, dst[2], lr_stride[1], 164| 40.0k| src[2] - offset_uv * PXSTRIDE(src_stride[1]), ------------------ | | 53| 40.0k|#define PXSTRIDE(x) (x) ------------------ 165| 40.0k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 166| 40.0k| row_h, w, h, ss_hor, 1); 167| 47.2k| if (have_tt && resize) ------------------ | Branch (167:17): [True: 47.2k, False: 0] | Branch (167:28): [True: 10.7k, False: 36.4k] ------------------ 168| 10.7k| backup_lpf(f, f->lf.cdef_lpf_line[2] + cdef_off_uv, src_stride[1], 169| 10.7k| src[2] - offset_uv * PXSTRIDE(src_stride[1]), ------------------ | | 53| 10.7k|#define PXSTRIDE(x) (x) ------------------ 170| 10.7k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 171| 10.7k| row_h, w, h, ss_hor, 0); 172| 47.2k| } 173| 48.1k| } 174| 241k|} dav1d_loopfilter_sbrow_cols_8bpc: 316| 518k|{ 317| 518k| int x, have_left; 318| | // Don't filter outside the frame 319| 518k| const int is_sb64 = !f->seq_hdr->sb128; 320| 518k| const int starty4 = (sby & is_sb64) << 4; 321| 518k| const int sbsz = 32 >> is_sb64; 322| 518k| const int sbl2 = 5 - is_sb64; 323| 518k| const int halign = (f->bh + 31) & ~31; 324| 518k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 325| 518k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 326| 518k| const int vmask = 16 >> ss_ver, hmask = 16 >> ss_hor; 327| 518k| const unsigned vmax = 1U << vmask, hmax = 1U << hmask; 328| 518k| const unsigned endy4 = starty4 + imin(f->h4 - sby * sbsz, sbsz); 329| 518k| const unsigned uv_endy4 = (endy4 + ss_ver) >> ss_ver; 330| | 331| | // fix lpf strength at tile col boundaries 332| 518k| const uint8_t *lpf_y = &f->lf.tx_lpf_right_edge[0][sby << sbl2]; 333| 518k| const uint8_t *lpf_uv = &f->lf.tx_lpf_right_edge[1][sby << (sbl2 - ss_ver)]; 334| 822k| for (int tile_col = 1;; tile_col++) { 335| 822k| x = f->frame_hdr->tiling.col_start_sb[tile_col]; 336| 822k| if ((x << sbl2) >= f->bw) break; ------------------ | Branch (336:13): [True: 518k, False: 303k] ------------------ 337| 303k| const int bx4 = x & is_sb64 ? 16 : 0, cbx4 = bx4 >> ss_hor; ------------------ | Branch (337:25): [True: 299k, False: 4.03k] ------------------ 338| 303k| x >>= is_sb64; 339| | 340| 303k| uint16_t (*const y_hmask)[2] = lflvl[x].filter_y[0][bx4]; 341| 5.17M| for (unsigned y = starty4, mask = 1 << y; y < endy4; y++, mask <<= 1) { ------------------ | Branch (341:51): [True: 4.87M, False: 303k] ------------------ 342| 4.87M| const int sidx = mask >= 0x10000U; 343| 4.87M| const unsigned smask = mask >> (sidx << 4); 344| 4.87M| const int idx = 2 * !!(y_hmask[2][sidx] & smask) + 345| 4.87M| !!(y_hmask[1][sidx] & smask); 346| 4.87M| y_hmask[2][sidx] &= ~smask; 347| 4.87M| y_hmask[1][sidx] &= ~smask; 348| 4.87M| y_hmask[0][sidx] &= ~smask; 349| 4.87M| y_hmask[imin(idx, lpf_y[y - starty4])][sidx] |= smask; 350| 4.87M| } 351| | 352| 303k| if (f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) { ------------------ | Branch (352:13): [True: 5.43k, False: 298k] ------------------ 353| 5.43k| uint16_t (*const uv_hmask)[2] = lflvl[x].filter_uv[0][cbx4]; 354| 62.4k| for (unsigned y = starty4 >> ss_ver, uv_mask = 1 << y; y < uv_endy4; ------------------ | Branch (354:68): [True: 56.9k, False: 5.43k] ------------------ 355| 56.9k| y++, uv_mask <<= 1) 356| 56.9k| { 357| 56.9k| const int sidx = uv_mask >= vmax; 358| 56.9k| const unsigned smask = uv_mask >> (sidx << (4 - ss_ver)); 359| 56.9k| const int idx = !!(uv_hmask[1][sidx] & smask); 360| 56.9k| uv_hmask[1][sidx] &= ~smask; 361| 56.9k| uv_hmask[0][sidx] &= ~smask; 362| 56.9k| uv_hmask[imin(idx, lpf_uv[y - (starty4 >> ss_ver)])][sidx] |= smask; 363| 56.9k| } 364| 5.43k| } 365| 303k| lpf_y += halign; 366| 303k| lpf_uv += halign >> ss_ver; 367| 303k| } 368| | 369| | // fix lpf strength at tile row boundaries 370| 518k| if (start_of_tile_row) { ------------------ | Branch (370:9): [True: 544, False: 518k] ------------------ 371| 544| const BlockContext *a; 372| 544| for (x = 0, a = &f->a[f->sb128w * (start_of_tile_row - 1)]; 373| 2.55k| x < f->sb128w; x++, a++) ------------------ | Branch (373:14): [True: 2.01k, False: 544] ------------------ 374| 2.01k| { 375| 2.01k| uint16_t (*const y_vmask)[2] = lflvl[x].filter_y[1][starty4]; 376| 2.01k| const unsigned w = imin(32, f->w4 - (x << 5)); 377| 61.0k| for (unsigned mask = 1, i = 0; i < w; mask <<= 1, i++) { ------------------ | Branch (377:44): [True: 59.0k, False: 2.01k] ------------------ 378| 59.0k| const int sidx = mask >= 0x10000U; 379| 59.0k| const unsigned smask = mask >> (sidx << 4); 380| 59.0k| const int idx = 2 * !!(y_vmask[2][sidx] & smask) + 381| 59.0k| !!(y_vmask[1][sidx] & smask); 382| 59.0k| y_vmask[2][sidx] &= ~smask; 383| 59.0k| y_vmask[1][sidx] &= ~smask; 384| 59.0k| y_vmask[0][sidx] &= ~smask; 385| 59.0k| y_vmask[imin(idx, a->tx_lpf_y[i])][sidx] |= smask; 386| 59.0k| } 387| | 388| 2.01k| if (f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) { ------------------ | Branch (388:17): [True: 1.24k, False: 764] ------------------ 389| 1.24k| const unsigned cw = (w + ss_hor) >> ss_hor; 390| 1.24k| uint16_t (*const uv_vmask)[2] = lflvl[x].filter_uv[1][starty4 >> ss_ver]; 391| 20.1k| for (unsigned uv_mask = 1, i = 0; i < cw; uv_mask <<= 1, i++) { ------------------ | Branch (391:51): [True: 18.9k, False: 1.24k] ------------------ 392| 18.9k| const int sidx = uv_mask >= hmax; 393| 18.9k| const unsigned smask = uv_mask >> (sidx << (4 - ss_hor)); 394| 18.9k| const int idx = !!(uv_vmask[1][sidx] & smask); 395| 18.9k| uv_vmask[1][sidx] &= ~smask; 396| 18.9k| uv_vmask[0][sidx] &= ~smask; 397| 18.9k| uv_vmask[imin(idx, a->tx_lpf_uv[i])][sidx] |= smask; 398| 18.9k| } 399| 1.24k| } 400| 2.01k| } 401| 544| } 402| | 403| 518k| pixel *ptr; 404| 518k| uint8_t (*level_ptr)[4] = f->lf.level + f->b4_stride * sby * sbsz; 405| 1.07M| for (ptr = p[0], have_left = 0, x = 0; x < f->sb128w; ------------------ | Branch (405:44): [True: 551k, False: 518k] ------------------ 406| 551k| x++, have_left = 1, ptr += 128, level_ptr += 32) 407| 551k| { 408| 551k| filter_plane_cols_y(f, have_left, level_ptr, f->b4_stride, 409| 551k| lflvl[x].filter_y[0], ptr, f->cur.stride[0], 410| 551k| imin(32, f->w4 - x * 32), starty4, endy4); 411| 551k| } 412| | 413| 518k| if (!f->frame_hdr->loopfilter.level_u && !f->frame_hdr->loopfilter.level_v) ------------------ | Branch (413:9): [True: 489k, False: 29.2k] | Branch (413:46): [True: 485k, False: 4.22k] ------------------ 414| 485k| return; 415| | 416| 33.4k| ptrdiff_t uv_off; 417| 33.4k| level_ptr = f->lf.level + f->b4_stride * (sby * sbsz >> ss_ver); 418| 92.5k| for (uv_off = 0, have_left = 0, x = 0; x < f->sb128w; ------------------ | Branch (418:44): [True: 59.1k, False: 33.4k] ------------------ 419| 59.1k| x++, have_left = 1, uv_off += 128 >> ss_hor, level_ptr += 32 >> ss_hor) 420| 59.1k| { 421| 59.1k| filter_plane_cols_uv(f, have_left, level_ptr, f->b4_stride, 422| 59.1k| lflvl[x].filter_uv[0], 423| 59.1k| &p[1][uv_off], &p[2][uv_off], f->cur.stride[1], 424| 59.1k| (imin(32, f->w4 - x * 32) + ss_hor) >> ss_hor, 425| 59.1k| starty4 >> ss_ver, uv_endy4, ss_ver); 426| 59.1k| } 427| 33.4k|} dav1d_loopfilter_sbrow_rows_8bpc: 432| 518k|{ 433| 518k| int x; 434| | // Don't filter outside the frame 435| 518k| const int have_top = sby > 0; 436| 518k| const int is_sb64 = !f->seq_hdr->sb128; 437| 518k| const int starty4 = (sby & is_sb64) << 4; 438| 518k| const int sbsz = 32 >> is_sb64; 439| 518k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 440| 518k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 441| 518k| const unsigned endy4 = starty4 + imin(f->h4 - sby * sbsz, sbsz); 442| 518k| const unsigned uv_endy4 = (endy4 + ss_ver) >> ss_ver; 443| | 444| 518k| pixel *ptr; 445| 518k| uint8_t (*level_ptr)[4] = f->lf.level + f->b4_stride * sby * sbsz; 446| 1.07M| for (ptr = p[0], x = 0; x < f->sb128w; x++, ptr += 128, level_ptr += 32) { ------------------ | Branch (446:29): [True: 551k, False: 518k] ------------------ 447| 551k| filter_plane_rows_y(f, have_top, level_ptr, f->b4_stride, 448| 551k| lflvl[x].filter_y[1], ptr, f->cur.stride[0], 449| 551k| imin(32, f->w4 - x * 32), starty4, endy4); 450| 551k| } 451| | 452| 518k| if (!f->frame_hdr->loopfilter.level_u && !f->frame_hdr->loopfilter.level_v) ------------------ | Branch (452:9): [True: 489k, False: 29.2k] | Branch (452:46): [True: 485k, False: 4.20k] ------------------ 453| 485k| return; 454| | 455| 33.4k| ptrdiff_t uv_off; 456| 33.4k| level_ptr = f->lf.level + f->b4_stride * (sby * sbsz >> ss_ver); 457| 92.5k| for (uv_off = 0, x = 0; x < f->sb128w; ------------------ | Branch (457:29): [True: 59.0k, False: 33.4k] ------------------ 458| 59.0k| x++, uv_off += 128 >> ss_hor, level_ptr += 32 >> ss_hor) 459| 59.0k| { 460| 59.0k| filter_plane_rows_uv(f, have_top, level_ptr, f->b4_stride, 461| 59.0k| lflvl[x].filter_uv[1], 462| 59.0k| &p[1][uv_off], &p[2][uv_off], f->cur.stride[1], 463| 59.0k| (imin(32, f->w4 - x * 32) + ss_hor) >> ss_hor, 464| 59.0k| starty4 >> ss_ver, uv_endy4, ss_hor); 465| 59.0k| } 466| 33.4k|} lf_apply_tmpl.c:backup_lpf: 47| 1.56M|{ 48| 1.56M| const int cdef_backup = !lr_backup; 49| 1.56M| const int dst_w = f->frame_hdr->super_res.enabled ? ------------------ | Branch (49:23): [True: 109k, False: 1.45M] ------------------ 50| 1.45M| (f->frame_hdr->width[1] + ss_hor) >> ss_hor : src_w; 51| | 52| | // The first stripe of the frame is shorter by 8 luma pixel rows. 53| 1.56M| int stripe_h = ((64 << (cdef_backup & sb128)) - 8 * !row) >> ss_ver; 54| 1.56M| src += (stripe_h - 2) * PXSTRIDE(src_stride); ------------------ | | 53| 1.56M|#define PXSTRIDE(x) (x) ------------------ 55| | 56| 1.56M| if (f->c->n_tc == 1) { ------------------ | Branch (56:9): [True: 0, False: 1.56M] ------------------ 57| 0| if (row) { ------------------ | Branch (57:13): [True: 0, False: 0] ------------------ 58| 0| const int top = 4 << sb128; 59| | // Copy the top part of the stored loop filtered pixels from the 60| | // previous sb row needed above the first stripe of this sb row. 61| 0| pixel_copy(&dst[PXSTRIDE(dst_stride) * 0], ------------------ | | 47| 0|#define pixel_copy memcpy ------------------ pixel_copy(&dst[PXSTRIDE(dst_stride) * 0], ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 62| 0| &dst[PXSTRIDE(dst_stride) * top], dst_w); ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 63| 0| pixel_copy(&dst[PXSTRIDE(dst_stride) * 1], ------------------ | | 47| 0|#define pixel_copy memcpy ------------------ pixel_copy(&dst[PXSTRIDE(dst_stride) * 1], ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 64| 0| &dst[PXSTRIDE(dst_stride) * (top + 1)], dst_w); ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 65| 0| pixel_copy(&dst[PXSTRIDE(dst_stride) * 2], ------------------ | | 47| 0|#define pixel_copy memcpy ------------------ pixel_copy(&dst[PXSTRIDE(dst_stride) * 2], ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 66| 0| &dst[PXSTRIDE(dst_stride) * (top + 2)], dst_w); ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 67| 0| pixel_copy(&dst[PXSTRIDE(dst_stride) * 3], ------------------ | | 47| 0|#define pixel_copy memcpy ------------------ pixel_copy(&dst[PXSTRIDE(dst_stride) * 3], ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 68| 0| &dst[PXSTRIDE(dst_stride) * (top + 3)], dst_w); ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 69| 0| } 70| 0| dst += 4 * PXSTRIDE(dst_stride); ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 71| 0| } 72| | 73| 1.56M| if (lr_backup && (f->frame_hdr->width[0] != f->frame_hdr->width[1])) { ------------------ | Branch (73:9): [True: 1.48M, False: 74.9k] | Branch (73:22): [True: 29.7k, False: 1.45M] ------------------ 74| 68.2k| while (row + stripe_h <= row_h) { ------------------ | Branch (74:16): [True: 38.4k, False: 29.7k] ------------------ 75| 38.4k| const int n_lines = 4 - (row + stripe_h + 1 == h); 76| 38.4k| f->dsp->mc.resize(dst, dst_stride, src, src_stride, 77| 38.4k| dst_w, n_lines, src_w, f->resize_step[ss_hor], 78| 38.4k| f->resize_start[ss_hor] HIGHBD_CALL_SUFFIX); 79| 38.4k| row += stripe_h; // unmodified stripe_h for the 1st stripe 80| 38.4k| stripe_h = 64 >> ss_ver; 81| 38.4k| src += stripe_h * PXSTRIDE(src_stride); ------------------ | | 53| 38.4k|#define PXSTRIDE(x) (x) ------------------ 82| 38.4k| dst += n_lines * PXSTRIDE(dst_stride); ------------------ | | 53| 38.4k|#define PXSTRIDE(x) (x) ------------------ 83| 38.4k| if (n_lines == 3) { ------------------ | Branch (83:17): [True: 2.87k, False: 35.6k] ------------------ 84| 2.87k| pixel_copy(dst, &dst[-PXSTRIDE(dst_stride)], dst_w); ------------------ | | 47| 2.87k|#define pixel_copy memcpy ------------------ pixel_copy(dst, &dst[-PXSTRIDE(dst_stride)], dst_w); ------------------ | | 53| 2.87k|#define PXSTRIDE(x) (x) ------------------ 85| 2.87k| dst += PXSTRIDE(dst_stride); ------------------ | | 53| 2.87k|#define PXSTRIDE(x) (x) ------------------ 86| 2.87k| } 87| 38.4k| } 88| 1.53M| } else { 89| 3.38M| while (row + stripe_h <= row_h) { ------------------ | Branch (89:16): [True: 1.85M, False: 1.53M] ------------------ 90| 1.85M| const int n_lines = 4 - (row + stripe_h + 1 == h); 91| 9.23M| for (int i = 0; i < 4; i++) { ------------------ | Branch (91:29): [True: 7.38M, False: 1.85M] ------------------ 92| 7.38M| pixel_copy(dst, i == n_lines ? &dst[-PXSTRIDE(dst_stride)] : ------------------ | | 47| 7.38M|#define pixel_copy memcpy ------------------ pixel_copy(dst, i == n_lines ? &dst[-PXSTRIDE(dst_stride)] : ------------------ | | 53| 8.52k|#define PXSTRIDE(x) (x) ------------------ | Branch (92:33): [True: 8.52k, False: 7.37M] ------------------ 93| 7.38M| src, src_w); 94| 7.38M| dst += PXSTRIDE(dst_stride); ------------------ | | 53| 7.38M|#define PXSTRIDE(x) (x) ------------------ 95| 7.38M| src += PXSTRIDE(src_stride); ------------------ | | 53| 7.38M|#define PXSTRIDE(x) (x) ------------------ 96| 7.38M| } 97| 1.85M| row += stripe_h; // unmodified stripe_h for the 1st stripe 98| 1.85M| stripe_h = 64 >> ss_ver; 99| 1.85M| src += (stripe_h - 4) * PXSTRIDE(src_stride); ------------------ | | 53| 1.85M|#define PXSTRIDE(x) (x) ------------------ 100| 1.85M| } 101| 1.53M| } 102| 1.56M|} lf_apply_tmpl.c:filter_plane_cols_y: 184| 1.27M|{ 185| 1.27M| const Dav1dDSPContext *const dsp = f->dsp; 186| | 187| | // filter edges between columns (e.g. block1 | block2) 188| 34.5M| for (int x = 0; x < w; x++) { ------------------ | Branch (188:21): [True: 33.2M, False: 1.27M] ------------------ 189| 33.2M| if (!have_left && !x) continue; ------------------ | Branch (189:13): [True: 31.6M, False: 1.68M] | Branch (189:27): [True: 1.22M, False: 30.3M] ------------------ 190| 32.0M| uint32_t hmask[4]; 191| 32.0M| if (!starty4) { ------------------ | Branch (191:13): [True: 25.1M, False: 6.88M] ------------------ 192| 25.1M| hmask[0] = mask[x][0][0]; 193| 25.1M| hmask[1] = mask[x][1][0]; 194| 25.1M| hmask[2] = mask[x][2][0]; 195| 25.1M| if (endy4 > 16) { ------------------ | Branch (195:17): [True: 18.0M, False: 7.18M] ------------------ 196| 18.0M| hmask[0] |= (unsigned) mask[x][0][1] << 16; 197| 18.0M| hmask[1] |= (unsigned) mask[x][1][1] << 16; 198| 18.0M| hmask[2] |= (unsigned) mask[x][2][1] << 16; 199| 18.0M| } 200| 25.1M| } else { 201| 6.88M| hmask[0] = mask[x][0][1]; 202| 6.88M| hmask[1] = mask[x][1][1]; 203| 6.88M| hmask[2] = mask[x][2][1]; 204| 6.88M| } 205| 32.0M| hmask[3] = 0; 206| 32.0M| dsp->lf.loop_filter_sb[0][0](&dst[x * 4], ls, hmask, 207| 32.0M| (const uint8_t(*)[4]) &lvl[x][0], b4_stride, 208| 32.0M| &f->lf.lim_lut, endy4 - starty4 HIGHBD_CALL_SUFFIX); 209| 32.0M| } 210| 1.27M|} lf_apply_tmpl.c:filter_plane_cols_uv: 251| 134k|{ 252| 134k| const Dav1dDSPContext *const dsp = f->dsp; 253| | 254| | // filter edges between columns (e.g. block1 | block2) 255| 2.10M| for (int x = 0; x < w; x++) { ------------------ | Branch (255:21): [True: 1.96M, False: 134k] ------------------ 256| 1.96M| if (!have_left && !x) continue; ------------------ | Branch (256:13): [True: 1.29M, False: 671k] | Branch (256:27): [True: 97.7k, False: 1.19M] ------------------ 257| 1.86M| uint32_t hmask[3]; 258| 1.86M| if (!starty4) { ------------------ | Branch (258:13): [True: 1.73M, False: 131k] ------------------ 259| 1.73M| hmask[0] = mask[x][0][0]; 260| 1.73M| hmask[1] = mask[x][1][0]; 261| 1.73M| if (endy4 > (16 >> ss_ver)) { ------------------ | Branch (261:17): [True: 1.50M, False: 227k] ------------------ 262| 1.50M| hmask[0] |= (unsigned) mask[x][0][1] << (16 >> ss_ver); 263| 1.50M| hmask[1] |= (unsigned) mask[x][1][1] << (16 >> ss_ver); 264| 1.50M| } 265| 1.73M| } else { 266| 131k| hmask[0] = mask[x][0][1]; 267| 131k| hmask[1] = mask[x][1][1]; 268| 131k| } 269| 1.86M| hmask[2] = 0; 270| 1.86M| dsp->lf.loop_filter_sb[1][0](&u[x * 4], ls, hmask, 271| 1.86M| (const uint8_t(*)[4]) &lvl[x][2], b4_stride, 272| 1.86M| &f->lf.lim_lut, endy4 - starty4 HIGHBD_CALL_SUFFIX); 273| 1.86M| dsp->lf.loop_filter_sb[1][0](&v[x * 4], ls, hmask, 274| 1.86M| (const uint8_t(*)[4]) &lvl[x][3], b4_stride, 275| 1.86M| &f->lf.lim_lut, endy4 - starty4 HIGHBD_CALL_SUFFIX); 276| 1.86M| } 277| 134k|} lf_apply_tmpl.c:filter_plane_rows_y: 220| 1.27M|{ 221| 1.27M| const Dav1dDSPContext *const dsp = f->dsp; 222| | 223| | // block1 224| | // filter edges between rows (e.g. ------) 225| | // block2 226| 33.3M| for (int y = starty4; y < endy4; ------------------ | Branch (226:27): [True: 32.0M, False: 1.27M] ------------------ 227| 32.0M| y++, dst += 4 * PXSTRIDE(ls), lvl += b4_stride) ------------------ | | 53| 32.0M|#define PXSTRIDE(x) (x) ------------------ 228| 32.0M| { 229| 32.0M| if (!have_top && !y) continue; ------------------ | Branch (229:13): [True: 740k, False: 31.3M] | Branch (229:26): [True: 33.5k, False: 706k] ------------------ 230| 32.0M| const uint32_t vmask[4] = { 231| 32.0M| mask[y][0][0] | ((unsigned) mask[y][0][1] << 16), 232| 32.0M| mask[y][1][0] | ((unsigned) mask[y][1][1] << 16), 233| 32.0M| mask[y][2][0] | ((unsigned) mask[y][2][1] << 16), 234| 32.0M| 0, 235| 32.0M| }; 236| 32.0M| dsp->lf.loop_filter_sb[0][1](dst, ls, vmask, 237| 32.0M| (const uint8_t(*)[4]) &lvl[0][1], b4_stride, 238| 32.0M| &f->lf.lim_lut, w HIGHBD_CALL_SUFFIX); 239| 32.0M| } 240| 1.27M|} lf_apply_tmpl.c:filter_plane_rows_uv: 288| 134k|{ 289| 134k| const Dav1dDSPContext *const dsp = f->dsp; 290| 134k| ptrdiff_t off_l = 0; 291| | 292| | // block1 293| | // filter edges between rows (e.g. ------) 294| | // block2 295| 2.31M| for (int y = starty4; y < endy4; ------------------ | Branch (295:27): [True: 2.18M, False: 134k] ------------------ 296| 2.18M| y++, off_l += 4 * PXSTRIDE(ls), lvl += b4_stride) ------------------ | | 53| 2.18M|#define PXSTRIDE(x) (x) ------------------ 297| 2.18M| { 298| 2.18M| if (!have_top && !y) continue; ------------------ | Branch (298:13): [True: 241k, False: 1.94M] | Branch (298:26): [True: 15.5k, False: 226k] ------------------ 299| 2.16M| const uint32_t vmask[3] = { 300| 2.16M| mask[y][0][0] | ((unsigned) mask[y][0][1] << (16 >> ss_hor)), 301| 2.16M| mask[y][1][0] | ((unsigned) mask[y][1][1] << (16 >> ss_hor)), 302| 2.16M| 0, 303| 2.16M| }; 304| 2.16M| dsp->lf.loop_filter_sb[1][1](&u[off_l], ls, vmask, 305| 2.16M| (const uint8_t(*)[4]) &lvl[0][2], b4_stride, 306| 2.16M| &f->lf.lim_lut, w HIGHBD_CALL_SUFFIX); 307| 2.16M| dsp->lf.loop_filter_sb[1][1](&v[off_l], ls, vmask, 308| 2.16M| (const uint8_t(*)[4]) &lvl[0][3], b4_stride, 309| 2.16M| &f->lf.lim_lut, w HIGHBD_CALL_SUFFIX); 310| 2.16M| } 311| 134k|} dav1d_copy_lpf_16bpc: 106| 871k|{ 107| 871k| const int have_tt = f->c->n_tc > 1; 108| 871k| const int resize = f->frame_hdr->width[0] != f->frame_hdr->width[1]; 109| 871k| const int offset = 8 * !!sby; 110| 871k| const ptrdiff_t *const src_stride = f->cur.stride; 111| 871k| const ptrdiff_t *const lr_stride = f->sr_cur.p.stride; 112| 871k| const int tt_off = have_tt * sby * (4 << f->seq_hdr->sb128); 113| 871k| pixel *const dst[3] = { 114| 871k| f->lf.lr_lpf_line[0] + tt_off * PXSTRIDE(lr_stride[0]), 115| 871k| f->lf.lr_lpf_line[1] + tt_off * PXSTRIDE(lr_stride[1]), 116| 871k| f->lf.lr_lpf_line[2] + tt_off * PXSTRIDE(lr_stride[1]) 117| 871k| }; 118| | 119| | // TODO Also check block level restore type to reduce copying. 120| 871k| const int restore_planes = f->lf.restore_planes; 121| | 122| 871k| if (f->seq_hdr->cdef || restore_planes & LR_RESTORE_Y) { ------------------ | Branch (122:9): [True: 859k, False: 11.5k] | Branch (122:29): [True: 10.8k, False: 683] ------------------ 123| 870k| const int h = f->cur.p.h; 124| 870k| const int w = f->bw << 2; 125| 870k| const int row_h = imin((sby + 1) << (6 + f->seq_hdr->sb128), h - 1); 126| 870k| const int y_stripe = (sby << (6 + f->seq_hdr->sb128)) - offset; 127| 870k| if (restore_planes & LR_RESTORE_Y || !resize) ------------------ | Branch (127:13): [True: 41.4k, False: 828k] | Branch (127:46): [True: 810k, False: 18.3k] ------------------ 128| 851k| backup_lpf(f, dst[0], lr_stride[0], 129| 851k| src[0] - offset * PXSTRIDE(src_stride[0]), src_stride[0], 130| 851k| 0, f->seq_hdr->sb128, y_stripe, row_h, w, h, 0, 1); 131| 870k| if (have_tt && resize) { ------------------ | Branch (131:13): [True: 869k, False: 211] | Branch (131:24): [True: 28.1k, False: 841k] ------------------ 132| 28.1k| const ptrdiff_t cdef_off_y = sby * 4 * PXSTRIDE(src_stride[0]); 133| 28.1k| backup_lpf(f, f->lf.cdef_lpf_line[0] + cdef_off_y, src_stride[0], 134| 28.1k| src[0] - offset * PXSTRIDE(src_stride[0]), src_stride[0], 135| 28.1k| 0, f->seq_hdr->sb128, y_stripe, row_h, w, h, 0, 0); 136| 28.1k| } 137| 870k| } 138| 871k| if ((f->seq_hdr->cdef || restore_planes & (LR_RESTORE_U | LR_RESTORE_V)) && ------------------ | Branch (138:10): [True: 859k, False: 11.5k] | Branch (138:30): [True: 2.03k, False: 9.52k] ------------------ 139| 861k| f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) ------------------ | Branch (139:9): [True: 166k, False: 694k] ------------------ 140| 166k| { 141| 166k| const int ss_ver = f->sr_cur.p.p.layout == DAV1D_PIXEL_LAYOUT_I420; 142| 166k| const int ss_hor = f->sr_cur.p.p.layout != DAV1D_PIXEL_LAYOUT_I444; 143| 166k| const int h = (f->cur.p.h + ss_ver) >> ss_ver; 144| 166k| const int w = f->bw << (2 - ss_hor); 145| 166k| const int row_h = imin((sby + 1) << ((6 - ss_ver) + f->seq_hdr->sb128), h - 1); 146| 166k| const int offset_uv = offset >> ss_ver; 147| 166k| const int y_stripe = (sby << ((6 - ss_ver) + f->seq_hdr->sb128)) - offset_uv; 148| 166k| const ptrdiff_t cdef_off_uv = sby * 4 * PXSTRIDE(src_stride[1]); 149| 166k| if (f->seq_hdr->cdef || restore_planes & LR_RESTORE_U) { ------------------ | Branch (149:13): [True: 164k, False: 2.03k] | Branch (149:33): [True: 1.45k, False: 574] ------------------ 150| 165k| if (restore_planes & LR_RESTORE_U || !resize) ------------------ | Branch (150:17): [True: 6.60k, False: 159k] | Branch (150:50): [True: 155k, False: 3.60k] ------------------ 151| 162k| backup_lpf(f, dst[1], lr_stride[1], 152| 162k| src[1] - offset_uv * PXSTRIDE(src_stride[1]), 153| 162k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 154| 162k| row_h, w, h, ss_hor, 1); 155| 165k| if (have_tt && resize) ------------------ | Branch (155:17): [True: 165k, False: 18.4E] | Branch (155:28): [True: 7.98k, False: 157k] ------------------ 156| 7.98k| backup_lpf(f, f->lf.cdef_lpf_line[1] + cdef_off_uv, src_stride[1], 157| 7.98k| src[1] - offset_uv * PXSTRIDE(src_stride[1]), 158| 7.98k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 159| 7.98k| row_h, w, h, ss_hor, 0); 160| 165k| } 161| 166k| if (f->seq_hdr->cdef || restore_planes & LR_RESTORE_V) { ------------------ | Branch (161:13): [True: 164k, False: 2.03k] | Branch (161:33): [True: 1.26k, False: 771] ------------------ 162| 165k| if (restore_planes & LR_RESTORE_V || !resize) ------------------ | Branch (162:17): [True: 7.84k, False: 157k] | Branch (162:50): [True: 155k, False: 2.53k] ------------------ 163| 163k| backup_lpf(f, dst[2], lr_stride[1], 164| 163k| src[2] - offset_uv * PXSTRIDE(src_stride[1]), 165| 163k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 166| 163k| row_h, w, h, ss_hor, 1); 167| 165k| if (have_tt && resize) ------------------ | Branch (167:17): [True: 165k, False: 18.4E] | Branch (167:28): [True: 7.80k, False: 157k] ------------------ 168| 7.80k| backup_lpf(f, f->lf.cdef_lpf_line[2] + cdef_off_uv, src_stride[1], 169| 7.80k| src[2] - offset_uv * PXSTRIDE(src_stride[1]), 170| 7.80k| src_stride[1], ss_ver, f->seq_hdr->sb128, y_stripe, 171| 7.80k| row_h, w, h, ss_hor, 0); 172| 165k| } 173| 166k| } 174| 871k|} dav1d_loopfilter_sbrow_cols_16bpc: 316| 701k|{ 317| 701k| int x, have_left; 318| | // Don't filter outside the frame 319| 701k| const int is_sb64 = !f->seq_hdr->sb128; 320| 701k| const int starty4 = (sby & is_sb64) << 4; 321| 701k| const int sbsz = 32 >> is_sb64; 322| 701k| const int sbl2 = 5 - is_sb64; 323| 701k| const int halign = (f->bh + 31) & ~31; 324| 701k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 325| 701k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 326| 701k| const int vmask = 16 >> ss_ver, hmask = 16 >> ss_hor; 327| 701k| const unsigned vmax = 1U << vmask, hmax = 1U << hmask; 328| 701k| const unsigned endy4 = starty4 + imin(f->h4 - sby * sbsz, sbsz); 329| 701k| const unsigned uv_endy4 = (endy4 + ss_ver) >> ss_ver; 330| | 331| | // fix lpf strength at tile col boundaries 332| 701k| const uint8_t *lpf_y = &f->lf.tx_lpf_right_edge[0][sby << sbl2]; 333| 701k| const uint8_t *lpf_uv = &f->lf.tx_lpf_right_edge[1][sby << (sbl2 - ss_ver)]; 334| 709k| for (int tile_col = 1;; tile_col++) { 335| 709k| x = f->frame_hdr->tiling.col_start_sb[tile_col]; 336| 709k| if ((x << sbl2) >= f->bw) break; ------------------ | Branch (336:13): [True: 701k, False: 8.14k] ------------------ 337| 8.14k| const int bx4 = x & is_sb64 ? 16 : 0, cbx4 = bx4 >> ss_hor; ------------------ | Branch (337:25): [True: 6.33k, False: 1.81k] ------------------ 338| 8.14k| x >>= is_sb64; 339| | 340| 8.14k| uint16_t (*const y_hmask)[2] = lflvl[x].filter_y[0][bx4]; 341| 162k| for (unsigned y = starty4, mask = 1 << y; y < endy4; y++, mask <<= 1) { ------------------ | Branch (341:51): [True: 154k, False: 8.14k] ------------------ 342| 154k| const int sidx = mask >= 0x10000U; 343| 154k| const unsigned smask = mask >> (sidx << 4); 344| 154k| const int idx = 2 * !!(y_hmask[2][sidx] & smask) + 345| 154k| !!(y_hmask[1][sidx] & smask); 346| 154k| y_hmask[2][sidx] &= ~smask; 347| 154k| y_hmask[1][sidx] &= ~smask; 348| 154k| y_hmask[0][sidx] &= ~smask; 349| 154k| y_hmask[imin(idx, lpf_y[y - starty4])][sidx] |= smask; 350| 154k| } 351| | 352| 8.14k| if (f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) { ------------------ | Branch (352:13): [True: 1.06k, False: 7.08k] ------------------ 353| 1.06k| uint16_t (*const uv_hmask)[2] = lflvl[x].filter_uv[0][cbx4]; 354| 28.2k| for (unsigned y = starty4 >> ss_ver, uv_mask = 1 << y; y < uv_endy4; ------------------ | Branch (354:68): [True: 27.2k, False: 1.06k] ------------------ 355| 27.2k| y++, uv_mask <<= 1) 356| 27.2k| { 357| 27.2k| const int sidx = uv_mask >= vmax; 358| 27.2k| const unsigned smask = uv_mask >> (sidx << (4 - ss_ver)); 359| 27.2k| const int idx = !!(uv_hmask[1][sidx] & smask); 360| 27.2k| uv_hmask[1][sidx] &= ~smask; 361| 27.2k| uv_hmask[0][sidx] &= ~smask; 362| 27.2k| uv_hmask[imin(idx, lpf_uv[y - (starty4 >> ss_ver)])][sidx] |= smask; 363| 27.2k| } 364| 1.06k| } 365| 8.14k| lpf_y += halign; 366| 8.14k| lpf_uv += halign >> ss_ver; 367| 8.14k| } 368| | 369| | // fix lpf strength at tile row boundaries 370| 701k| if (start_of_tile_row) { ------------------ | Branch (370:9): [True: 1.46k, False: 700k] ------------------ 371| 1.46k| const BlockContext *a; 372| 1.46k| for (x = 0, a = &f->a[f->sb128w * (start_of_tile_row - 1)]; 373| 5.19k| x < f->sb128w; x++, a++) ------------------ | Branch (373:14): [True: 3.73k, False: 1.46k] ------------------ 374| 3.73k| { 375| 3.73k| uint16_t (*const y_vmask)[2] = lflvl[x].filter_y[1][starty4]; 376| 3.73k| const unsigned w = imin(32, f->w4 - (x << 5)); 377| 113k| for (unsigned mask = 1, i = 0; i < w; mask <<= 1, i++) { ------------------ | Branch (377:44): [True: 109k, False: 3.73k] ------------------ 378| 109k| const int sidx = mask >= 0x10000U; 379| 109k| const unsigned smask = mask >> (sidx << 4); 380| 109k| const int idx = 2 * !!(y_vmask[2][sidx] & smask) + 381| 109k| !!(y_vmask[1][sidx] & smask); 382| 109k| y_vmask[2][sidx] &= ~smask; 383| 109k| y_vmask[1][sidx] &= ~smask; 384| 109k| y_vmask[0][sidx] &= ~smask; 385| 109k| y_vmask[imin(idx, a->tx_lpf_y[i])][sidx] |= smask; 386| 109k| } 387| | 388| 3.73k| if (f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) { ------------------ | Branch (388:17): [True: 2.39k, False: 1.33k] ------------------ 389| 2.39k| const unsigned cw = (w + ss_hor) >> ss_hor; 390| 2.39k| uint16_t (*const uv_vmask)[2] = lflvl[x].filter_uv[1][starty4 >> ss_ver]; 391| 49.2k| for (unsigned uv_mask = 1, i = 0; i < cw; uv_mask <<= 1, i++) { ------------------ | Branch (391:51): [True: 46.8k, False: 2.39k] ------------------ 392| 46.8k| const int sidx = uv_mask >= hmax; 393| 46.8k| const unsigned smask = uv_mask >> (sidx << (4 - ss_hor)); 394| 46.8k| const int idx = !!(uv_vmask[1][sidx] & smask); 395| 46.8k| uv_vmask[1][sidx] &= ~smask; 396| 46.8k| uv_vmask[0][sidx] &= ~smask; 397| 46.8k| uv_vmask[imin(idx, a->tx_lpf_uv[i])][sidx] |= smask; 398| 46.8k| } 399| 2.39k| } 400| 3.73k| } 401| 1.46k| } 402| | 403| 701k| pixel *ptr; 404| 701k| uint8_t (*level_ptr)[4] = f->lf.level + f->b4_stride * sby * sbsz; 405| 1.42M| for (ptr = p[0], have_left = 0, x = 0; x < f->sb128w; ------------------ | Branch (405:44): [True: 724k, False: 701k] ------------------ 406| 724k| x++, have_left = 1, ptr += 128, level_ptr += 32) 407| 724k| { 408| 724k| filter_plane_cols_y(f, have_left, level_ptr, f->b4_stride, 409| 724k| lflvl[x].filter_y[0], ptr, f->cur.stride[0], 410| 724k| imin(32, f->w4 - x * 32), starty4, endy4); 411| 724k| } 412| | 413| 701k| if (!f->frame_hdr->loopfilter.level_u && !f->frame_hdr->loopfilter.level_v) ------------------ | Branch (413:9): [True: 640k, False: 61.6k] | Branch (413:46): [True: 637k, False: 2.07k] ------------------ 414| 637k| return; 415| | 416| 63.7k| ptrdiff_t uv_off; 417| 63.7k| level_ptr = f->lf.level + f->b4_stride * (sby * sbsz >> ss_ver); 418| 139k| for (uv_off = 0, have_left = 0, x = 0; x < f->sb128w; ------------------ | Branch (418:44): [True: 75.6k, False: 63.7k] ------------------ 419| 75.6k| x++, have_left = 1, uv_off += 128 >> ss_hor, level_ptr += 32 >> ss_hor) 420| 75.6k| { 421| 75.6k| filter_plane_cols_uv(f, have_left, level_ptr, f->b4_stride, 422| 75.6k| lflvl[x].filter_uv[0], 423| 75.6k| &p[1][uv_off], &p[2][uv_off], f->cur.stride[1], 424| 75.6k| (imin(32, f->w4 - x * 32) + ss_hor) >> ss_hor, 425| 75.6k| starty4 >> ss_ver, uv_endy4, ss_ver); 426| 75.6k| } 427| 63.7k|} dav1d_loopfilter_sbrow_rows_16bpc: 432| 702k|{ 433| 702k| int x; 434| | // Don't filter outside the frame 435| 702k| const int have_top = sby > 0; 436| 702k| const int is_sb64 = !f->seq_hdr->sb128; 437| 702k| const int starty4 = (sby & is_sb64) << 4; 438| 702k| const int sbsz = 32 >> is_sb64; 439| 702k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 440| 702k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 441| 702k| const unsigned endy4 = starty4 + imin(f->h4 - sby * sbsz, sbsz); 442| 702k| const unsigned uv_endy4 = (endy4 + ss_ver) >> ss_ver; 443| | 444| 702k| pixel *ptr; 445| 702k| uint8_t (*level_ptr)[4] = f->lf.level + f->b4_stride * sby * sbsz; 446| 1.42M| for (ptr = p[0], x = 0; x < f->sb128w; x++, ptr += 128, level_ptr += 32) { ------------------ | Branch (446:29): [True: 725k, False: 702k] ------------------ 447| 725k| filter_plane_rows_y(f, have_top, level_ptr, f->b4_stride, 448| 725k| lflvl[x].filter_y[1], ptr, f->cur.stride[0], 449| 725k| imin(32, f->w4 - x * 32), starty4, endy4); 450| 725k| } 451| | 452| 702k| if (!f->frame_hdr->loopfilter.level_u && !f->frame_hdr->loopfilter.level_v) ------------------ | Branch (452:9): [True: 639k, False: 62.1k] | Branch (452:46): [True: 637k, False: 2.07k] ------------------ 453| 637k| return; 454| | 455| 64.2k| ptrdiff_t uv_off; 456| 64.2k| level_ptr = f->lf.level + f->b4_stride * (sby * sbsz >> ss_ver); 457| 139k| for (uv_off = 0, x = 0; x < f->sb128w; ------------------ | Branch (457:29): [True: 75.6k, False: 64.2k] ------------------ 458| 75.6k| x++, uv_off += 128 >> ss_hor, level_ptr += 32 >> ss_hor) 459| 75.6k| { 460| 75.6k| filter_plane_rows_uv(f, have_top, level_ptr, f->b4_stride, 461| 75.6k| lflvl[x].filter_uv[1], 462| 75.6k| &p[1][uv_off], &p[2][uv_off], f->cur.stride[1], 463| 75.6k| (imin(32, f->w4 - x * 32) + ss_hor) >> ss_hor, 464| 75.6k| starty4 >> ss_ver, uv_endy4, ss_hor); 465| 75.6k| } 466| 64.2k|} dav1d_create_lf_mask_intra: 271| 1.53M|{ 272| 1.53M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 273| 1.53M| const int bw4 = imin(iw - bx, b_dim[0]); 274| 1.53M| const int bh4 = imin(ih - by, b_dim[1]); 275| 1.53M| const int bx4 = bx & 31; 276| 1.53M| const int by4 = by & 31; 277| 1.53M| assert(bw4 >= 0 && bh4 >= 0); ------------------ | Branch (277:5): [True: 1.53M, False: 4.72k] | Branch (277:5): [True: 1.53M, False: 18.4E] ------------------ 278| | 279| 1.53M| if (bw4 && bh4) { ------------------ | Branch (279:9): [True: 1.52M, False: 3.68k] | Branch (279:16): [True: 1.50M, False: 28.4k] ------------------ 280| 1.50M| uint8_t (*level_cache_ptr)[4] = level_cache + by * b4_stride + bx; 281| 8.47M| for (int y = 0; y < bh4; y++) { ------------------ | Branch (281:25): [True: 6.97M, False: 1.50M] ------------------ 282| 66.4M| for (int x = 0; x < bw4; x++) { ------------------ | Branch (282:29): [True: 59.4M, False: 6.97M] ------------------ 283| 59.4M| level_cache_ptr[x][0] = filter_level[0][0][0]; 284| 59.4M| level_cache_ptr[x][1] = filter_level[1][0][0]; 285| 59.4M| } 286| 6.97M| level_cache_ptr += b4_stride; 287| 6.97M| } 288| | 289| 1.50M| mask_edges_intra(lflvl->filter_y, by4, bx4, bw4, bh4, ytx, ay, ly); 290| 1.50M| } 291| | 292| 1.53M| if (!auv) return; ------------------ | Branch (292:9): [True: 453k, False: 1.07M] ------------------ 293| | 294| 1.07M| const int ss_ver = layout == DAV1D_PIXEL_LAYOUT_I420; 295| 1.07M| const int ss_hor = layout != DAV1D_PIXEL_LAYOUT_I444; 296| 1.07M| const int cbw4 = imin(((iw + ss_hor) >> ss_hor) - (bx >> ss_hor), 297| 1.07M| (b_dim[0] + ss_hor) >> ss_hor); 298| 1.07M| const int cbh4 = imin(((ih + ss_ver) >> ss_ver) - (by >> ss_ver), 299| 1.07M| (b_dim[1] + ss_ver) >> ss_ver); 300| 1.07M| assert(cbw4 >= 0 && cbh4 >= 0); ------------------ | Branch (300:5): [True: 1.08M, False: 18.4E] | Branch (300:5): [True: 1.08M, False: 18.4E] ------------------ 301| | 302| 1.08M| if (!cbw4 || !cbh4) return; ------------------ | Branch (302:9): [True: 1.92k, False: 1.08M] | Branch (302:18): [True: 8.92k, False: 1.07M] ------------------ 303| | 304| 1.07M| const int cbx4 = bx4 >> ss_hor; 305| 1.07M| const int cby4 = by4 >> ss_ver; 306| | 307| 1.07M| uint8_t (*level_cache_ptr)[4] = 308| 1.07M| level_cache + (by >> ss_ver) * b4_stride + (bx >> ss_hor); 309| 4.56M| for (int y = 0; y < cbh4; y++) { ------------------ | Branch (309:21): [True: 3.48M, False: 1.07M] ------------------ 310| 24.4M| for (int x = 0; x < cbw4; x++) { ------------------ | Branch (310:25): [True: 20.9M, False: 3.48M] ------------------ 311| 20.9M| level_cache_ptr[x][2] = filter_level[2][0][0]; 312| 20.9M| level_cache_ptr[x][3] = filter_level[3][0][0]; 313| 20.9M| } 314| 3.48M| level_cache_ptr += b4_stride; 315| 3.48M| } 316| | 317| 1.07M| mask_edges_chroma(lflvl->filter_uv, cby4, cbx4, cbw4, cbh4, 0, uvtx, 318| 1.07M| auv, luv, ss_hor, ss_ver); 319| 1.07M|} dav1d_create_lf_mask_inter: 334| 3.63M|{ 335| 3.63M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 336| 3.63M| const int bw4 = imin(iw - bx, b_dim[0]); 337| 3.63M| const int bh4 = imin(ih - by, b_dim[1]); 338| 3.63M| const int bx4 = bx & 31; 339| 3.63M| const int by4 = by & 31; 340| 3.63M| assert(bw4 >= 0 && bh4 >= 0); ------------------ | Branch (340:5): [True: 3.63M, False: 7.49k] | Branch (340:5): [True: 3.63M, False: 18.4E] ------------------ 341| | 342| 3.63M| if (bw4 && bh4) { ------------------ | Branch (342:9): [True: 3.62M, False: 3.85k] | Branch (342:16): [True: 3.62M, False: 7.83k] ------------------ 343| 3.62M| uint8_t (*level_cache_ptr)[4] = level_cache + by * b4_stride + bx; 344| 49.6M| for (int y = 0; y < bh4; y++) { ------------------ | Branch (344:25): [True: 46.0M, False: 3.62M] ------------------ 345| 866M| for (int x = 0; x < bw4; x++) { ------------------ | Branch (345:29): [True: 820M, False: 46.0M] ------------------ 346| 820M| level_cache_ptr[x][0] = filter_level[0][0][0]; 347| 820M| level_cache_ptr[x][1] = filter_level[1][0][0]; 348| 820M| } 349| 46.0M| level_cache_ptr += b4_stride; 350| 46.0M| } 351| | 352| 3.62M| mask_edges_inter(lflvl->filter_y, by4, bx4, bw4, bh4, skip, 353| 3.62M| max_ytx, tx_masks, ay, ly); 354| 3.62M| } 355| | 356| 3.63M| if (!auv) return; ------------------ | Branch (356:9): [True: 2.51M, False: 1.11M] ------------------ 357| | 358| 1.11M| const int ss_ver = layout == DAV1D_PIXEL_LAYOUT_I420; 359| 1.11M| const int ss_hor = layout != DAV1D_PIXEL_LAYOUT_I444; 360| 1.11M| const int cbw4 = imin(((iw + ss_hor) >> ss_hor) - (bx >> ss_hor), 361| 1.11M| (b_dim[0] + ss_hor) >> ss_hor); 362| 1.11M| const int cbh4 = imin(((ih + ss_ver) >> ss_ver) - (by >> ss_ver), 363| 1.11M| (b_dim[1] + ss_ver) >> ss_ver); 364| 1.11M| assert(cbw4 >= 0 && cbh4 >= 0); ------------------ | Branch (364:5): [True: 1.11M, False: 18.4E] | Branch (364:5): [True: 1.11M, False: 104] ------------------ 365| | 366| 1.11M| if (!cbw4 || !cbh4) return; ------------------ | Branch (366:9): [True: 831, False: 1.11M] | Branch (366:18): [True: 1.66k, False: 1.11M] ------------------ 367| | 368| 1.11M| const int cbx4 = bx4 >> ss_hor; 369| 1.11M| const int cby4 = by4 >> ss_ver; 370| | 371| 1.11M| uint8_t (*level_cache_ptr)[4] = 372| 1.11M| level_cache + (by >> ss_ver) * b4_stride + (bx >> ss_hor); 373| 5.46M| for (int y = 0; y < cbh4; y++) { ------------------ | Branch (373:21): [True: 4.35M, False: 1.11M] ------------------ 374| 38.5M| for (int x = 0; x < cbw4; x++) { ------------------ | Branch (374:25): [True: 34.1M, False: 4.35M] ------------------ 375| 34.1M| level_cache_ptr[x][2] = filter_level[2][0][0]; 376| 34.1M| level_cache_ptr[x][3] = filter_level[3][0][0]; 377| 34.1M| } 378| 4.35M| level_cache_ptr += b4_stride; 379| 4.35M| } 380| | 381| 1.11M| mask_edges_chroma(lflvl->filter_uv, cby4, cbx4, cbw4, cbh4, skip, uvtx, 382| 1.11M| auv, luv, ss_hor, ss_ver); 383| 1.11M|} dav1d_calc_eih: 385| 34.1k|void dav1d_calc_eih(Av1FilterLUT *const lim_lut, const int filter_sharpness) { 386| | // set E/I/H values from loopfilter level 387| 34.1k| const int sharp = filter_sharpness; 388| 2.21M| for (int level = 0; level < 64; level++) { ------------------ | Branch (388:25): [True: 2.17M, False: 34.1k] ------------------ 389| 2.17M| int limit = level; 390| | 391| 2.17M| if (sharp > 0) { ------------------ | Branch (391:13): [True: 1.11M, False: 1.06M] ------------------ 392| 1.11M| limit >>= (sharp + 3) >> 2; 393| 1.11M| limit = imin(limit, 9 - sharp); 394| 1.11M| } 395| 2.17M| limit = imax(limit, 1); 396| | 397| 2.17M| lim_lut->i[level] = limit; 398| 2.17M| lim_lut->e[level] = 2 * (level + 2) + limit; 399| 2.17M| } 400| 34.1k| lim_lut->sharp[0] = (sharp + 3) >> 2; 401| 34.1k| lim_lut->sharp[1] = sharp ? 9 - sharp : 0xff; ------------------ | Branch (401:25): [True: 17.3k, False: 16.7k] ------------------ 402| 34.1k|} dav1d_calc_lf_values: 441| 371k|{ 442| 371k| const int n_seg = hdr->segmentation.enabled ? 8 : 1; ------------------ | Branch (442:23): [True: 25.2k, False: 346k] ------------------ 443| | 444| 371k| if (!hdr->loopfilter.level_y[0] && !hdr->loopfilter.level_y[1]) { ------------------ | Branch (444:9): [True: 282k, False: 89.3k] | Branch (444:40): [True: 272k, False: 9.52k] ------------------ 445| 272k| memset(lflvl_values, 0, sizeof(*lflvl_values) * n_seg); 446| 272k| return; 447| 272k| } 448| | 449| 98.8k| const Dav1dLoopfilterModeRefDeltas *const mr_deltas = 450| 98.8k| hdr->loopfilter.mode_ref_delta_enabled ? ------------------ | Branch (450:9): [True: 67.1k, False: 31.7k] ------------------ 451| 98.8k| &hdr->loopfilter.mode_ref_deltas : NULL; 452| 349k| for (int s = 0; s < n_seg; s++) { ------------------ | Branch (452:21): [True: 250k, False: 98.8k] ------------------ 453| 250k| const Dav1dSegmentationData *const segd = 454| 250k| hdr->segmentation.enabled ? &hdr->segmentation.seg_data.d[s] : NULL; ------------------ | Branch (454:13): [True: 173k, False: 77.1k] ------------------ 455| | 456| 250k| calc_lf_value(lflvl_values[s][0], hdr->loopfilter.level_y[0], 457| 250k| lf_delta[0], segd ? segd->delta_lf_y_v : 0, mr_deltas); ------------------ | Branch (457:36): [True: 173k, False: 77.1k] ------------------ 458| 250k| calc_lf_value(lflvl_values[s][1], hdr->loopfilter.level_y[1], 459| 250k| lf_delta[hdr->delta.lf.multi ? 1 : 0], ------------------ | Branch (459:32): [True: 89.5k, False: 161k] ------------------ 460| 250k| segd ? segd->delta_lf_y_h : 0, mr_deltas); ------------------ | Branch (460:23): [True: 173k, False: 77.2k] ------------------ 461| 250k| calc_lf_value_chroma(lflvl_values[s][2], hdr->loopfilter.level_u, 462| 250k| lf_delta[hdr->delta.lf.multi ? 2 : 0], ------------------ | Branch (462:39): [True: 89.5k, False: 161k] ------------------ 463| 250k| segd ? segd->delta_lf_u : 0, mr_deltas); ------------------ | Branch (463:30): [True: 173k, False: 77.1k] ------------------ 464| 250k| calc_lf_value_chroma(lflvl_values[s][3], hdr->loopfilter.level_v, 465| 250k| lf_delta[hdr->delta.lf.multi ? 3 : 0], ------------------ | Branch (465:39): [True: 89.5k, False: 161k] ------------------ 466| 250k| segd ? segd->delta_lf_v : 0, mr_deltas); ------------------ | Branch (466:30): [True: 173k, False: 77.1k] ------------------ 467| 250k| } 468| 98.8k|} lf_mask.c:mask_edges_intra: 152| 1.50M|{ 153| 1.50M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[tx]; 154| 1.50M| const int twl4 = t_dim->lw, thl4 = t_dim->lh; 155| 1.50M| const int twl4c = imin(2, twl4), thl4c = imin(2, thl4); 156| 1.50M| int y, x; 157| | 158| | // left block edge 159| 1.50M| unsigned mask = 1U << by4; 160| 8.51M| for (y = 0; y < h4; y++, mask <<= 1) { ------------------ | Branch (160:17): [True: 7.01M, False: 1.50M] ------------------ 161| 7.01M| const int sidx = mask >= 0x10000; 162| 7.01M| const unsigned smask = mask >> (sidx << 4); 163| 7.01M| masks[0][bx4][imin(twl4c, l[y])][sidx] |= smask; 164| 7.01M| } 165| | 166| | // top block edge 167| 8.43M| for (x = 0, mask = 1U << bx4; x < w4; x++, mask <<= 1) { ------------------ | Branch (167:35): [True: 6.93M, False: 1.50M] ------------------ 168| 6.93M| const int sidx = mask >= 0x10000; 169| 6.93M| const unsigned smask = mask >> (sidx << 4); 170| 6.93M| masks[1][by4][imin(thl4c, a[x])][sidx] |= smask; 171| 6.93M| } 172| | 173| | // inner (tx) left|right edges 174| 1.50M| const int hstep = t_dim->w; 175| 1.50M| unsigned t = 1U << by4; 176| 1.50M| unsigned inner = (unsigned) ((((uint64_t) t) << h4) - t); 177| 1.50M| unsigned inner1 = inner & 0xffff, inner2 = inner >> 16; 178| 1.94M| for (x = hstep; x < w4; x += hstep) { ------------------ | Branch (178:21): [True: 441k, False: 1.50M] ------------------ 179| 441k| if (inner1) masks[0][bx4 + x][twl4c][0] |= inner1; ------------------ | Branch (179:13): [True: 340k, False: 100k] ------------------ 180| 441k| if (inner2) masks[0][bx4 + x][twl4c][1] |= inner2; ------------------ | Branch (180:13): [True: 247k, False: 193k] ------------------ 181| 441k| } 182| | 183| | // top 184| | // inner (tx) --- edges 185| | // bottom 186| 1.50M| const int vstep = t_dim->h; 187| 1.50M| t = 1U << bx4; 188| 1.50M| inner = (unsigned) ((((uint64_t) t) << w4) - t); 189| 1.50M| inner1 = inner & 0xffff; 190| 1.50M| inner2 = inner >> 16; 191| 2.02M| for (y = vstep; y < h4; y += vstep) { ------------------ | Branch (191:21): [True: 520k, False: 1.50M] ------------------ 192| 520k| if (inner1) masks[1][by4 + y][thl4c][0] |= inner1; ------------------ | Branch (192:13): [True: 386k, False: 133k] ------------------ 193| 520k| if (inner2) masks[1][by4 + y][thl4c][1] |= inner2; ------------------ | Branch (193:13): [True: 360k, False: 159k] ------------------ 194| 520k| } 195| | 196| 1.50M| dav1d_memset_likely_pow2(a, thl4c, w4); 197| 1.50M| dav1d_memset_likely_pow2(l, twl4c, h4); 198| 1.50M|} lf_mask.c:mask_edges_chroma: 207| 2.18M|{ 208| 2.18M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[tx]; 209| 2.18M| const int twl4 = t_dim->lw, thl4 = t_dim->lh; 210| 2.18M| const int twl4c = !!twl4, thl4c = !!thl4; 211| 2.18M| int y, x; 212| 2.18M| const int vbits = 4 - ss_ver, hbits = 4 - ss_hor; 213| 2.18M| const int vmask = 16 >> ss_ver, hmask = 16 >> ss_hor; 214| 2.18M| const unsigned vmax = 1 << vmask, hmax = 1 << hmask; 215| | 216| | // left block edge 217| 2.18M| unsigned mask = 1U << cby4; 218| 10.0M| for (y = 0; y < ch4; y++, mask <<= 1) { ------------------ | Branch (218:17): [True: 7.86M, False: 2.18M] ------------------ 219| 7.86M| const int sidx = mask >= vmax; 220| 7.86M| const unsigned smask = mask >> (sidx << vbits); 221| 7.86M| masks[0][cbx4][imin(twl4c, l[y])][sidx] |= smask; 222| 7.86M| } 223| | 224| | // top block edge 225| 9.27M| for (x = 0, mask = 1U << cbx4; x < cw4; x++, mask <<= 1) { ------------------ | Branch (225:36): [True: 7.08M, False: 2.18M] ------------------ 226| 7.08M| const int sidx = mask >= hmax; 227| 7.08M| const unsigned smask = mask >> (sidx << hbits); 228| 7.08M| masks[1][cby4][imin(thl4c, a[x])][sidx] |= smask; 229| 7.08M| } 230| | 231| 2.18M| if (!skip_inter) { ------------------ | Branch (231:9): [True: 1.64M, False: 545k] ------------------ 232| | // inner (tx) left|right edges 233| 1.64M| const int hstep = t_dim->w; 234| 1.64M| unsigned t = 1U << cby4; 235| 1.64M| unsigned inner = (unsigned) ((((uint64_t) t) << ch4) - t); 236| 1.64M| unsigned inner1 = inner & ((1 << vmask) - 1), inner2 = inner >> vmask; 237| 1.78M| for (x = hstep; x < cw4; x += hstep) { ------------------ | Branch (237:25): [True: 144k, False: 1.64M] ------------------ 238| 144k| if (inner1) masks[0][cbx4 + x][twl4c][0] |= inner1; ------------------ | Branch (238:17): [True: 127k, False: 17.2k] ------------------ 239| 144k| if (inner2) masks[0][cbx4 + x][twl4c][1] |= inner2; ------------------ | Branch (239:17): [True: 116k, False: 27.8k] ------------------ 240| 144k| } 241| | 242| | // top 243| | // inner (tx) --- edges 244| | // bottom 245| 1.64M| const int vstep = t_dim->h; 246| 1.64M| t = 1U << cbx4; 247| 1.64M| inner = (unsigned) ((((uint64_t) t) << cw4) - t); 248| 1.64M| inner1 = inner & ((1 << hmask) - 1), inner2 = inner >> hmask; 249| 1.92M| for (y = vstep; y < ch4; y += vstep) { ------------------ | Branch (249:25): [True: 280k, False: 1.64M] ------------------ 250| 280k| if (inner1) masks[1][cby4 + y][thl4c][0] |= inner1; ------------------ | Branch (250:17): [True: 253k, False: 26.9k] ------------------ 251| 280k| if (inner2) masks[1][cby4 + y][thl4c][1] |= inner2; ------------------ | Branch (251:17): [True: 236k, False: 43.9k] ------------------ 252| 280k| } 253| 1.64M| } 254| | 255| 2.18M| dav1d_memset_likely_pow2(a, thl4c, cw4); 256| 2.18M| dav1d_memset_likely_pow2(l, twl4c, ch4); 257| 2.18M|} lf_mask.c:mask_edges_inter: 85| 3.62M|{ 86| 3.62M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[max_tx]; 87| 3.62M| int y, x; 88| | 89| 3.62M| ALIGN_STK_16(uint8_t, txa, 2 /* edge */, [2 /* txsz, step */][32 /* y */][32 /* x */]); ------------------ | | 100| 3.62M| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 3.62M| line __attribute__((aligned(align))) | | ------------------ ------------------ 90| 8.30M| for (int y_off = 0, y = 0; y < h4; y += t_dim->h, y_off++) ------------------ | Branch (90:32): [True: 4.67M, False: 3.62M] ------------------ 91| 16.4M| for (int x_off = 0, x = 0; x < w4; x += t_dim->w, x_off++) ------------------ | Branch (91:36): [True: 11.7M, False: 4.67M] ------------------ 92| 11.7M| decomp_tx((uint8_t(*)[2][32][32]) &txa[0][0][y][x], 93| 11.7M| max_tx, 0, y_off, x_off, tx_masks); 94| | 95| | // left block edge 96| 3.62M| unsigned mask = 1U << by4; 97| 50.1M| for (y = 0; y < h4; y++, mask <<= 1) { ------------------ | Branch (97:17): [True: 46.5M, False: 3.62M] ------------------ 98| 46.5M| const int sidx = mask >= 0x10000; 99| 46.5M| const unsigned smask = mask >> (sidx << 4); 100| 46.5M| masks[0][bx4][imin(txa[0][0][y][0], l[y])][sidx] |= smask; 101| 46.5M| } 102| | 103| | // top block edge 104| 42.9M| for (x = 0, mask = 1U << bx4; x < w4; x++, mask <<= 1) { ------------------ | Branch (104:35): [True: 39.3M, False: 3.62M] ------------------ 105| 39.3M| const int sidx = mask >= 0x10000; 106| 39.3M| const unsigned smask = mask >> (sidx << 4); 107| 39.3M| masks[1][by4][imin(txa[1][0][0][x], a[x])][sidx] |= smask; 108| 39.3M| } 109| | 110| 3.62M| if (!skip) { ------------------ | Branch (110:9): [True: 1.09M, False: 2.52M] ------------------ 111| | // inner (tx) left|right edges 112| 5.41M| for (y = 0, mask = 1U << by4; y < h4; y++, mask <<= 1) { ------------------ | Branch (112:39): [True: 4.32M, False: 1.09M] ------------------ 113| 4.32M| const int sidx = mask >= 0x10000U; 114| 4.32M| const unsigned smask = mask >> (sidx << 4); 115| 4.32M| int ltx = txa[0][0][y][0]; 116| 4.32M| int step = txa[0][1][y][0]; 117| 5.30M| for (x = step; x < w4; x += step) { ------------------ | Branch (117:28): [True: 983k, False: 4.32M] ------------------ 118| 983k| const int rtx = txa[0][0][y][x]; 119| 983k| masks[0][bx4 + x][imin(rtx, ltx)][sidx] |= smask; 120| 983k| ltx = rtx; 121| 983k| step = txa[0][1][y][x]; 122| 983k| } 123| 4.32M| } 124| | 125| | // top 126| | // inner (tx) --- edges 127| | // bottom 128| 5.80M| for (x = 0, mask = 1U << bx4; x < w4; x++, mask <<= 1) { ------------------ | Branch (128:39): [True: 4.70M, False: 1.09M] ------------------ 129| 4.70M| const int sidx = mask >= 0x10000U; 130| 4.70M| const unsigned smask = mask >> (sidx << 4); 131| 4.70M| int ttx = txa[1][0][0][x]; 132| 4.70M| int step = txa[1][1][0][x]; 133| 5.63M| for (y = step; y < h4; y += step) { ------------------ | Branch (133:28): [True: 925k, False: 4.70M] ------------------ 134| 925k| const int btx = txa[1][0][y][x]; 135| 925k| masks[1][by4 + y][imin(ttx, btx)][sidx] |= smask; 136| 925k| ttx = btx; 137| 925k| step = txa[1][1][y][x]; 138| 925k| } 139| 4.70M| } 140| 1.09M| } 141| | 142| 50.1M| for (y = 0; y < h4; y++) ------------------ | Branch (142:17): [True: 46.5M, False: 3.62M] ------------------ 143| 46.5M| l[y] = txa[0][0][y][w4 - 1]; 144| 3.62M| memcpy(a, txa[1][0][h4 - 1], w4); 145| 3.62M|} lf_mask.c:decomp_tx: 44| 12.3M|{ 45| 12.3M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[from]; 46| 12.3M| const int is_split = (from == (int) TX_4X4 || depth > 1) ? 0 : ------------------ | Branch (46:27): [True: 6.38M, False: 5.97M] | Branch (46:51): [True: 153k, False: 5.82M] ------------------ 47| 12.3M| (tx_masks[depth] >> (y_off * 4 + x_off)) & 1; 48| | 49| 12.3M| if (is_split) { ------------------ | Branch (49:9): [True: 209k, False: 12.1M] ------------------ 50| 209k| const enum RectTxfmSize sub = t_dim->sub; 51| 209k| const int htw4 = t_dim->w >> 1, hth4 = t_dim->h >> 1; 52| | 53| 209k| decomp_tx(txa, sub, depth + 1, y_off * 2 + 0, x_off * 2 + 0, tx_masks); 54| 209k| if (t_dim->w >= t_dim->h) ------------------ | Branch (54:13): [True: 164k, False: 45.0k] ------------------ 55| 164k| decomp_tx((uint8_t(*)[2][32][32]) &txa[0][0][0][htw4], 56| 164k| sub, depth + 1, y_off * 2 + 0, x_off * 2 + 1, tx_masks); 57| 209k| if (t_dim->h >= t_dim->w) { ------------------ | Branch (57:13): [True: 154k, False: 54.7k] ------------------ 58| 154k| decomp_tx((uint8_t(*)[2][32][32]) &txa[0][0][hth4][0], 59| 154k| sub, depth + 1, y_off * 2 + 1, x_off * 2 + 0, tx_masks); 60| 154k| if (t_dim->w >= t_dim->h) ------------------ | Branch (60:17): [True: 109k, False: 44.9k] ------------------ 61| 109k| decomp_tx((uint8_t(*)[2][32][32]) &txa[0][0][hth4][htw4], 62| 109k| sub, depth + 1, y_off * 2 + 1, x_off * 2 + 1, tx_masks); 63| 154k| } 64| 12.1M| } else { 65| 12.1M| const int lw = imin(2, t_dim->lw), lh = imin(2, t_dim->lh); 66| | 67| 12.1M|#define set_ctx(rep_macro) \ 68| 12.1M| for (int y = 0; y < t_dim->h; y++) { \ 69| 12.1M| rep_macro(txa[0][0][y], 0, lw); \ 70| 12.1M| rep_macro(txa[1][0][y], 0, lh); \ 71| 12.1M| txa[0][1][y][0] = t_dim->w; \ 72| 12.1M| } 73| 12.1M| case_set_upto16(t_dim->lw); ------------------ | | 80| 12.1M| switch (var) { \ | | 81| 6.57M| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 68| 13.4M| for (int y = 0; y < t_dim->h; y++) { \ | | | | ------------------ | | | | | Branch (68:25): [True: 6.91M, False: 6.57M] | | | | ------------------ | | | | 69| 6.91M| rep_macro(txa[0][0][y], 0, lw); \ | | | | ------------------ | | | | | | 81| 6.91M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 6.91M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 70| 6.91M| rep_macro(txa[1][0][y], 0, lh); \ | | | | ------------------ | | | | | | 81| 6.91M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 6.91M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 71| 6.91M| txa[0][1][y][0] = t_dim->w; \ | | | | 72| 6.91M| } | | ------------------ | | | Branch (81:5): [True: 6.57M, False: 5.57M] | | ------------------ | | 82| 783k| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 68| 2.79M| for (int y = 0; y < t_dim->h; y++) { \ | | | | ------------------ | | | | | Branch (68:25): [True: 2.00M, False: 783k] | | | | ------------------ | | | | 69| 2.00M| rep_macro(txa[0][0][y], 0, lw); \ | | | | ------------------ | | | | | | 82| 2.00M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 2.00M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 70| 2.00M| rep_macro(txa[1][0][y], 0, lh); \ | | | | ------------------ | | | | | | 82| 2.00M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 2.00M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 71| 2.00M| txa[0][1][y][0] = t_dim->w; \ | | | | 72| 2.00M| } | | ------------------ | | | Branch (82:5): [True: 783k, False: 11.3M] | | ------------------ | | 83| 707k| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 68| 3.58M| for (int y = 0; y < t_dim->h; y++) { \ | | | | ------------------ | | | | | Branch (68:25): [True: 2.87M, False: 707k] | | | | ------------------ | | | | 69| 2.87M| rep_macro(txa[0][0][y], 0, lw); \ | | | | ------------------ | | | | | | 83| 2.87M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 2.87M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 70| 2.87M| rep_macro(txa[1][0][y], 0, lh); \ | | | | ------------------ | | | | | | 83| 2.87M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 2.87M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 71| 2.87M| txa[0][1][y][0] = t_dim->w; \ | | | | 72| 2.87M| } | | ------------------ | | | Branch (83:5): [True: 707k, False: 11.4M] | | ------------------ | | 84| 227k| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 68| 1.60M| for (int y = 0; y < t_dim->h; y++) { \ | | | | ------------------ | | | | | Branch (68:25): [True: 1.38M, False: 227k] | | | | ------------------ | | | | 69| 1.38M| rep_macro(txa[0][0][y], 0, lw); \ | | | | ------------------ | | | | | | 84| 1.38M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.38M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 70| 1.38M| rep_macro(txa[1][0][y], 0, lh); \ | | | | ------------------ | | | | | | 84| 1.38M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.38M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 71| 1.38M| txa[0][1][y][0] = t_dim->w; \ | | | | 72| 1.38M| } | | ------------------ | | | Branch (84:5): [True: 227k, False: 11.9M] | | ------------------ | | 85| 3.86M| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 68| 65.3M| for (int y = 0; y < t_dim->h; y++) { \ | | | | ------------------ | | | | | Branch (68:25): [True: 61.4M, False: 3.86M] | | | | ------------------ | | | | 69| 61.4M| rep_macro(txa[0][0][y], 0, lw); \ | | | | ------------------ | | | | | | 85| 61.4M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 61.4M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 61.4M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 61.4M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 61.4M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 70| 61.4M| rep_macro(txa[1][0][y], 0, lh); \ | | | | ------------------ | | | | | | 85| 61.4M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 61.4M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 61.4M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 61.4M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 61.4M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 71| 61.4M| txa[0][1][y][0] = t_dim->w; \ | | | | 72| 61.4M| } | | ------------------ | | | Branch (85:5): [True: 3.86M, False: 8.28M] | | ------------------ | | 86| 0| default: assert(0); \ | | ------------------ | | | Branch (86:5): [True: 0, False: 12.1M] | | ------------------ | | 87| 12.1M| } ------------------ | Branch (73:9): [Folded, False: 0] ------------------ 74| 12.1M|#undef set_ctx 75| 12.1M| dav1d_memset_pow2[t_dim->lw](txa[1][1][0], t_dim->h); 76| 12.1M| } 77| 12.3M|} lf_mask.c:calc_lf_value: 408| 658k|{ 409| 658k| const int base = iclip(iclip(base_lvl + lf_delta, 0, 63) + seg_delta, 0, 63); 410| | 411| 658k| if (!mr_delta) { ------------------ | Branch (411:9): [True: 258k, False: 400k] ------------------ 412| 258k| memset(lflvl_values, base, sizeof(*lflvl_values) * 8); 413| 400k| } else { 414| 400k| const int sh = base >= 32; 415| 400k| lflvl_values[0][0] = lflvl_values[0][1] = 416| 400k| iclip(base + (mr_delta->ref_delta[0] * (1 << sh)), 0, 63); 417| 3.19M| for (int r = 1; r < 8; r++) { ------------------ | Branch (417:25): [True: 2.79M, False: 400k] ------------------ 418| 8.38M| for (int m = 0; m < 2; m++) { ------------------ | Branch (418:29): [True: 5.58M, False: 2.79M] ------------------ 419| 5.58M| const int delta = 420| 5.58M| mr_delta->mode_delta[m] + mr_delta->ref_delta[r]; 421| 5.58M| lflvl_values[r][m] = iclip(base + (delta * (1 << sh)), 0, 63); 422| 5.58M| } 423| 2.79M| } 424| 400k| } 425| 658k|} lf_mask.c:calc_lf_value_chroma: 431| 501k|{ 432| 501k| if (!base_lvl) ------------------ | Branch (432:9): [True: 343k, False: 157k] ------------------ 433| 343k| memset(lflvl_values, 0, sizeof(*lflvl_values) * 8); 434| 157k| else 435| 157k| calc_lf_value(lflvl_values, base_lvl, lf_delta, seg_delta, mr_delta); 436| 501k|} dav1d_version: 61| 9.42k|COLD const char *dav1d_version(void) { 62| 9.42k| return DAV1D_VERSION; ------------------ | | 2| 9.42k|#define DAV1D_VERSION "1.5.4-0-g54706fc" ------------------ 63| 9.42k|} dav1d_default_settings: 71| 9.41k|COLD void dav1d_default_settings(Dav1dSettings *const s) { 72| 9.41k| s->n_threads = 0; 73| 9.41k| s->max_frame_delay = 0; 74| 9.41k| s->apply_grain = 1; 75| 9.41k| s->allocator.cookie = NULL; 76| 9.41k| s->allocator.alloc_picture_callback = dav1d_default_picture_alloc; 77| 9.41k| s->allocator.release_picture_callback = dav1d_default_picture_release; 78| 9.41k| s->logger.cookie = NULL; 79| | s->logger.callback = dav1d_log_default_callback; ------------------ | | 43| 9.41k|#define dav1d_log_default_callback NULL ------------------ 80| 9.41k| s->operating_point = 0; 81| 9.41k| s->all_layers = 1; // just until the tests are adjusted 82| 9.41k| s->frame_size_limit = 0; 83| 9.41k| s->strict_std_compliance = 0; 84| 9.41k| s->output_invisible_frames = 0; 85| 9.41k| s->inloop_filters = DAV1D_INLOOPFILTER_ALL; 86| 9.41k| s->decode_frame_type = DAV1D_DECODEFRAMETYPE_ALL; 87| 9.41k|} dav1d_open: 140| 9.41k|COLD int dav1d_open(Dav1dContext **const c_out, const Dav1dSettings *const s) { 141| 9.41k| static pthread_once_t initted = PTHREAD_ONCE_INIT; 142| 9.41k| pthread_once(&initted, init_internal); 143| | 144| 9.41k| validate_input_or_ret(c_out != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 9.41k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 9.41k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 145| 9.41k| validate_input_or_ret(s != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 9.41k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 9.41k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 146| 9.41k| validate_input_or_ret(s->n_threads >= 0 && ------------------ | | 52| 18.8k| if (!(x)) { \ | | ------------------ | | | Branch (52:11): [True: 9.41k, False: 0] | | | Branch (52:11): [True: 9.41k, False: 0] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 147| 9.41k| s->n_threads <= DAV1D_MAX_THREADS, DAV1D_ERR(EINVAL)); 148| 9.41k| validate_input_or_ret(s->max_frame_delay >= 0 && ------------------ | | 52| 18.8k| if (!(x)) { \ | | ------------------ | | | Branch (52:11): [True: 9.41k, False: 0] | | | Branch (52:11): [True: 9.41k, False: 0] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 149| 9.41k| s->max_frame_delay <= DAV1D_MAX_FRAME_DELAY, DAV1D_ERR(EINVAL)); 150| 9.41k| validate_input_or_ret(s->allocator.alloc_picture_callback != NULL, ------------------ | | 52| 9.41k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 9.41k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 151| 9.41k| DAV1D_ERR(EINVAL)); 152| 9.41k| validate_input_or_ret(s->allocator.release_picture_callback != NULL, ------------------ | | 52| 9.41k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 9.41k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 153| 9.41k| DAV1D_ERR(EINVAL)); 154| 9.41k| validate_input_or_ret(s->operating_point >= 0 && ------------------ | | 52| 18.8k| if (!(x)) { \ | | ------------------ | | | Branch (52:11): [True: 9.41k, False: 0] | | | Branch (52:11): [True: 9.41k, False: 0] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 155| 9.41k| s->operating_point <= 31, DAV1D_ERR(EINVAL)); 156| 9.41k| validate_input_or_ret(s->decode_frame_type >= DAV1D_DECODEFRAMETYPE_ALL && ------------------ | | 52| 18.8k| if (!(x)) { \ | | ------------------ | | | Branch (52:11): [True: 9.41k, False: 0] | | | Branch (52:11): [True: 9.41k, False: 0] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 157| 9.41k| s->decode_frame_type <= DAV1D_DECODEFRAMETYPE_KEY, DAV1D_ERR(EINVAL)); 158| | 159| 9.41k| pthread_attr_t thread_attr; 160| 9.41k| if (pthread_attr_init(&thread_attr)) return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (160:9): [True: 0, False: 9.41k] ------------------ 161| 9.41k| size_t stack_size = 1024 * 1024 + get_stack_size_internal(&thread_attr); 162| | 163| 9.41k| pthread_attr_setstacksize(&thread_attr, stack_size); 164| | 165| 9.41k| Dav1dContext *const c = *c_out = dav1d_alloc_aligned(ALLOC_COMMON_CTX, sizeof(*c), 64); ------------------ | | 134| 9.41k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 166| 9.41k| if (!c) goto error; ------------------ | Branch (166:9): [True: 0, False: 9.41k] ------------------ 167| 9.41k| memset(c, 0, sizeof(*c)); 168| | 169| 9.41k| c->allocator = s->allocator; 170| 9.41k| c->logger = s->logger; 171| 9.41k| c->apply_grain = s->apply_grain; 172| 9.41k| c->operating_point = s->operating_point; 173| 9.41k| c->all_layers = s->all_layers; 174| 9.41k| c->frame_size_limit = s->frame_size_limit; 175| 9.41k| c->strict_std_compliance = s->strict_std_compliance; 176| 9.41k| c->output_invisible_frames = s->output_invisible_frames; 177| 9.41k| c->inloop_filters = s->inloop_filters; 178| 9.41k| c->decode_frame_type = s->decode_frame_type; 179| | 180| 9.41k| dav1d_data_props_set_defaults(&c->cached_error_props); 181| | 182| 9.41k| if (dav1d_mem_pool_init(ALLOC_OBU_HDR, &c->seq_hdr_pool) || ------------------ | | 131| 18.8k|#define dav1d_mem_pool_init(type, pool) dav1d_mem_pool_init(pool) | | ------------------ | | | Branch (131:41): [True: 0, False: 9.41k] | | ------------------ ------------------ 183| 9.41k| dav1d_mem_pool_init(ALLOC_OBU_HDR, &c->frame_hdr_pool) || ------------------ | | 131| 18.8k|#define dav1d_mem_pool_init(type, pool) dav1d_mem_pool_init(pool) | | ------------------ | | | Branch (131:41): [True: 0, False: 9.41k] | | ------------------ ------------------ 184| 9.41k| dav1d_mem_pool_init(ALLOC_SEGMAP, &c->segmap_pool) || ------------------ | | 131| 18.8k|#define dav1d_mem_pool_init(type, pool) dav1d_mem_pool_init(pool) | | ------------------ | | | Branch (131:41): [True: 0, False: 9.41k] | | ------------------ ------------------ 185| 9.41k| dav1d_mem_pool_init(ALLOC_REFMVS, &c->refmvs_pool) || ------------------ | | 131| 18.8k|#define dav1d_mem_pool_init(type, pool) dav1d_mem_pool_init(pool) | | ------------------ | | | Branch (131:41): [True: 0, False: 9.41k] | | ------------------ ------------------ 186| 9.41k| dav1d_mem_pool_init(ALLOC_PIC_CTX, &c->pic_ctx_pool) || ------------------ | | 131| 18.8k|#define dav1d_mem_pool_init(type, pool) dav1d_mem_pool_init(pool) | | ------------------ | | | Branch (131:41): [True: 0, False: 9.41k] | | ------------------ ------------------ 187| 9.41k| dav1d_mem_pool_init(ALLOC_CDF, &c->cdf_pool)) ------------------ | | 131| 9.41k|#define dav1d_mem_pool_init(type, pool) dav1d_mem_pool_init(pool) | | ------------------ | | | Branch (131:41): [True: 0, False: 9.41k] | | ------------------ ------------------ 188| 0| { 189| 0| goto error; 190| 0| } 191| | 192| 9.41k| if (c->allocator.alloc_picture_callback == dav1d_default_picture_alloc && ------------------ | Branch (192:9): [True: 9.41k, False: 0] ------------------ 193| 9.41k| c->allocator.release_picture_callback == dav1d_default_picture_release) ------------------ | Branch (193:9): [True: 9.41k, False: 0] ------------------ 194| 9.41k| { 195| 9.41k| if (c->allocator.cookie) goto error; ------------------ | Branch (195:13): [True: 0, False: 9.41k] ------------------ 196| 9.41k| if (dav1d_mem_pool_init(ALLOC_PIC, &c->picture_pool)) goto error; ------------------ | | 131| 9.41k|#define dav1d_mem_pool_init(type, pool) dav1d_mem_pool_init(pool) | | ------------------ | | | Branch (131:41): [True: 0, False: 9.41k] | | ------------------ ------------------ 197| 9.41k| c->allocator.cookie = c->picture_pool; 198| 9.41k| } else if (c->allocator.alloc_picture_callback == dav1d_default_picture_alloc || ------------------ | Branch (198:16): [True: 0, False: 0] ------------------ 199| 0| c->allocator.release_picture_callback == dav1d_default_picture_release) ------------------ | Branch (199:16): [True: 0, False: 0] ------------------ 200| 0| { 201| 0| goto error; 202| 0| } 203| | 204| | /* On 32-bit systems extremely large frame sizes can cause overflows in 205| | * dav1d_decode_frame() malloc size calculations. Prevent that from occuring 206| | * by enforcing a maximum frame size limit, chosen to roughly correspond to 207| | * the largest size possible to decode without exhausting virtual memory. */ 208| 9.41k| if (sizeof(size_t) < 8 && s->frame_size_limit - 1 >= 8192 * 8192) { ------------------ | Branch (208:9): [Folded, False: 9.41k] | Branch (208:31): [True: 0, False: 0] ------------------ 209| 0| c->frame_size_limit = 8192 * 8192; 210| 0| if (s->frame_size_limit) ------------------ | Branch (210:13): [True: 0, False: 0] ------------------ 211| 0| dav1d_log(c, "Frame size limit reduced from %u to %u.\n", ------------------ | | 44| 0|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 0] | | ------------------ ------------------ 212| 0| s->frame_size_limit, c->frame_size_limit); 213| 0| } 214| | 215| 9.41k| c->flush = &c->flush_mem; 216| 9.41k| atomic_init(c->flush, 0); 217| | 218| 9.41k| get_num_threads(c, s, &c->n_tc, &c->n_fc); 219| | 220| 9.41k| c->fc = dav1d_alloc_aligned(ALLOC_THREAD_CTX, sizeof(*c->fc) * c->n_fc, 32); ------------------ | | 134| 9.41k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 221| 9.41k| if (!c->fc) goto error; ------------------ | Branch (221:9): [True: 0, False: 9.41k] ------------------ 222| 9.41k| memset(c->fc, 0, sizeof(*c->fc) * c->n_fc); 223| | 224| 9.41k| c->tc = dav1d_alloc_aligned(ALLOC_THREAD_CTX, sizeof(*c->tc) * c->n_tc, 64); ------------------ | | 134| 9.41k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 225| 9.41k| if (!c->tc) goto error; ------------------ | Branch (225:9): [True: 0, False: 9.41k] ------------------ 226| 9.41k| memset(c->tc, 0, sizeof(*c->tc) * c->n_tc); 227| 9.41k| if (c->n_tc > 1) { ------------------ | Branch (227:9): [True: 9.41k, False: 0] ------------------ 228| 9.41k| if (pthread_mutex_init(&c->task_thread.lock, NULL)) goto error; ------------------ | Branch (228:13): [True: 0, False: 9.41k] ------------------ 229| 9.41k| if (pthread_cond_init(&c->task_thread.cond, NULL)) { ------------------ | Branch (229:13): [True: 0, False: 9.41k] ------------------ 230| 0| pthread_mutex_destroy(&c->task_thread.lock); 231| 0| goto error; 232| 0| } 233| 9.41k| if (pthread_cond_init(&c->task_thread.delayed_fg.cond, NULL)) { ------------------ | Branch (233:13): [True: 0, False: 9.41k] ------------------ 234| 0| pthread_cond_destroy(&c->task_thread.cond); 235| 0| pthread_mutex_destroy(&c->task_thread.lock); 236| 0| goto error; 237| 0| } 238| 9.41k| c->task_thread.cur = c->n_fc; 239| 9.41k| atomic_init(&c->task_thread.reset_task_cur, UINT_MAX); 240| 9.41k| atomic_init(&c->task_thread.cond_signaled, 0); 241| 9.41k| c->task_thread.inited = 1; 242| 9.41k| } 243| | 244| 9.41k| if (c->n_fc > 1) { ------------------ | Branch (244:9): [True: 9.41k, False: 0] ------------------ 245| 9.41k| const size_t out_delayed_sz = sizeof(*c->frame_thread.out_delayed) * c->n_fc; 246| 9.41k| c->frame_thread.out_delayed = 247| 9.41k| dav1d_malloc(ALLOC_THREAD_CTX, out_delayed_sz); ------------------ | | 132| 9.41k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 248| 9.41k| if (!c->frame_thread.out_delayed) goto error; ------------------ | Branch (248:13): [True: 0, False: 9.41k] ------------------ 249| 9.41k| memset(c->frame_thread.out_delayed, 0, out_delayed_sz); 250| 9.41k| } 251| 47.0k| for (unsigned n = 0; n < c->n_fc; n++) { ------------------ | Branch (251:26): [True: 37.6k, False: 9.41k] ------------------ 252| 37.6k| Dav1dFrameContext *const f = &c->fc[n]; 253| 37.6k| if (c->n_tc > 1) { ------------------ | Branch (253:13): [True: 37.6k, False: 0] ------------------ 254| 37.6k| if (pthread_mutex_init(&f->task_thread.lock, NULL)) goto error; ------------------ | Branch (254:17): [True: 0, False: 37.6k] ------------------ 255| 37.6k| if (pthread_cond_init(&f->task_thread.cond, NULL)) { ------------------ | Branch (255:17): [True: 0, False: 37.6k] ------------------ 256| 0| pthread_mutex_destroy(&f->task_thread.lock); 257| 0| goto error; 258| 0| } 259| 37.6k| if (pthread_mutex_init(&f->task_thread.pending_tasks.lock, NULL)) { ------------------ | Branch (259:17): [True: 0, False: 37.6k] ------------------ 260| 0| pthread_cond_destroy(&f->task_thread.cond); 261| 0| pthread_mutex_destroy(&f->task_thread.lock); 262| 0| goto error; 263| 0| } 264| 37.6k| } 265| 37.6k| f->c = c; 266| 37.6k| f->task_thread.ttd = &c->task_thread; 267| 37.6k| f->lf.last_sharpness = -1; 268| 37.6k| } 269| | 270| 47.0k| for (unsigned m = 0; m < c->n_tc; m++) { ------------------ | Branch (270:26): [True: 37.6k, False: 9.41k] ------------------ 271| 37.6k| Dav1dTaskContext *const t = &c->tc[m]; 272| 37.6k| t->f = &c->fc[0]; 273| 37.6k| t->task_thread.ttd = &c->task_thread; 274| 37.6k| t->c = c; 275| 37.6k| memset(t->cf_16bpc, 0, sizeof(t->cf_16bpc)); 276| 37.6k| if (c->n_tc > 1) { ------------------ | Branch (276:13): [True: 37.6k, False: 0] ------------------ 277| 37.6k| if (pthread_mutex_init(&t->task_thread.td.lock, NULL)) goto error; ------------------ | Branch (277:17): [True: 0, False: 37.6k] ------------------ 278| 37.6k| if (pthread_cond_init(&t->task_thread.td.cond, NULL)) { ------------------ | Branch (278:17): [True: 0, False: 37.6k] ------------------ 279| 0| pthread_mutex_destroy(&t->task_thread.td.lock); 280| 0| goto error; 281| 0| } 282| 37.6k| if (pthread_create(&t->task_thread.td.thread, &thread_attr, dav1d_worker_task, t)) { ------------------ | Branch (282:17): [True: 0, False: 37.6k] ------------------ 283| 0| pthread_cond_destroy(&t->task_thread.td.cond); 284| 0| pthread_mutex_destroy(&t->task_thread.td.lock); 285| 0| goto error; 286| 0| } 287| 37.6k| t->task_thread.td.inited = 1; 288| 37.6k| } 289| 37.6k| } 290| 9.41k| dav1d_pal_dsp_init(&c->pal_dsp); 291| 9.41k| dav1d_refmvs_dsp_init(&c->refmvs_dsp); 292| | 293| 9.41k| pthread_attr_destroy(&thread_attr); 294| | 295| 9.41k| return 0; 296| | 297| 0|error: 298| 0| if (c) close_internal(c_out, 0); ------------------ | Branch (298:9): [True: 0, False: 0] ------------------ 299| 0| pthread_attr_destroy(&thread_attr); 300| 0| return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 301| 9.41k|} dav1d_send_data: 439| 430k|{ 440| 430k| validate_input_or_ret(c != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 430k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 430k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 441| 430k| validate_input_or_ret(in != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 430k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 430k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 442| | 443| 430k| if (in->data) { ------------------ | Branch (443:9): [True: 430k, False: 0] ------------------ 444| 430k| validate_input_or_ret(in->sz > 0 && in->sz <= SIZE_MAX / 2, DAV1D_ERR(EINVAL)); ------------------ | | 52| 860k| if (!(x)) { \ | | ------------------ | | | Branch (52:11): [True: 430k, False: 0] | | | Branch (52:11): [True: 430k, False: 0] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 445| 430k| c->drain = 0; 446| 430k| } 447| 430k| if (c->in.data) ------------------ | Branch (447:9): [True: 19.9k, False: 410k] ------------------ 448| 19.9k| return DAV1D_ERR(EAGAIN); ------------------ | | 58| 19.9k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 449| 410k| dav1d_data_ref(&c->in, in); 450| | 451| 410k| int res = gen_picture(c); 452| 410k| if (!res) ------------------ | Branch (452:9): [True: 370k, False: 39.8k] ------------------ 453| 370k| dav1d_data_unref_internal(in); 454| | 455| 410k| return res; 456| 430k|} dav1d_get_picture: 459| 411k|{ 460| 411k| validate_input_or_ret(c != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 411k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 411k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 461| 411k| validate_input_or_ret(out != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 411k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 411k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 462| | 463| 411k| const int drain = c->drain; 464| 411k| c->drain = 1; 465| | 466| 411k| int res = gen_picture(c); 467| 411k| if (res < 0) ------------------ | Branch (467:9): [True: 141, False: 411k] ------------------ 468| 141| return res; 469| | 470| 411k| if (c->cached_error) { ------------------ | Branch (470:9): [True: 176k, False: 234k] ------------------ 471| 176k| const int res = c->cached_error; 472| 176k| c->cached_error = 0; 473| 176k| return res; 474| 176k| } 475| | 476| 234k| if (output_picture_ready(c, c->n_fc == 1)) ------------------ | Branch (476:9): [True: 168k, False: 65.5k] ------------------ 477| 168k| return output_image(c, out); 478| | 479| 65.5k| if (c->n_fc > 1 && drain) ------------------ | Branch (479:9): [True: 65.5k, False: 0] | Branch (479:24): [True: 17.3k, False: 48.1k] ------------------ 480| 17.3k| return drain_picture(c, out); 481| | 482| 48.1k| return DAV1D_ERR(EAGAIN); ------------------ | | 58| 48.1k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 483| 65.5k|} dav1d_apply_grain: 487| 10.1k|{ 488| 10.1k| validate_input_or_ret(c != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 10.1k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 10.1k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 489| 10.1k| validate_input_or_ret(out != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 10.1k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 10.1k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 490| 10.1k| validate_input_or_ret(in != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 10.1k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 10.1k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 491| | 492| 10.1k| if (!has_grain(in)) { ------------------ | Branch (492:9): [True: 0, False: 10.1k] ------------------ 493| 0| dav1d_picture_ref(out, in); 494| 0| return 0; 495| 0| } 496| | 497| 10.1k| int res = dav1d_picture_alloc_copy(c, out, in->p.w, in); 498| 10.1k| if (res < 0) goto error; ------------------ | Branch (498:9): [True: 0, False: 10.1k] ------------------ 499| | 500| 10.1k| if (c->n_tc > 1) { ------------------ | Branch (500:9): [True: 10.1k, False: 0] ------------------ 501| 10.1k| dav1d_task_delayed_fg(c, out, in); 502| 10.1k| } else { 503| 0| switch (out->p.bpc) { 504| 0|#if CONFIG_8BPC 505| 0| case 8: ------------------ | Branch (505:9): [True: 0, False: 0] ------------------ 506| 0| dav1d_apply_grain_8bpc(&c->dsp[0].fg, out, in); 507| 0| break; 508| 0|#endif 509| 0|#if CONFIG_16BPC 510| 0| case 10: ------------------ | Branch (510:9): [True: 0, False: 0] ------------------ 511| 0| case 12: ------------------ | Branch (511:9): [True: 0, False: 0] ------------------ 512| 0| dav1d_apply_grain_16bpc(&c->dsp[(out->p.bpc >> 1) - 4].fg, out, in); 513| 0| break; 514| 0|#endif 515| 0| default: abort(); ------------------ | Branch (515:9): [True: 0, False: 0] ------------------ 516| 0| } 517| 0| } 518| | 519| 10.1k| return 0; 520| | 521| 0|error: 522| 0| dav1d_picture_unref_internal(out); 523| 0| return res; 524| 10.1k|} dav1d_flush: 526| 9.41k|void dav1d_flush(Dav1dContext *const c) { 527| 9.41k| dav1d_data_unref_internal(&c->in); 528| 9.41k| if (c->out.p.frame_hdr) ------------------ | Branch (528:9): [True: 0, False: 9.41k] ------------------ 529| 0| dav1d_thread_picture_unref(&c->out); 530| 9.41k| if (c->cache.p.frame_hdr) ------------------ | Branch (530:9): [True: 0, False: 9.41k] ------------------ 531| 0| dav1d_thread_picture_unref(&c->cache); 532| | 533| 9.41k| c->drain = 0; 534| 9.41k| c->cached_error = 0; 535| | 536| 84.7k| for (int i = 0; i < 8; i++) { ------------------ | Branch (536:21): [True: 75.3k, False: 9.41k] ------------------ 537| 75.3k| if (c->refs[i].p.p.frame_hdr) ------------------ | Branch (537:13): [True: 65.8k, False: 9.48k] ------------------ 538| 65.8k| dav1d_thread_picture_unref(&c->refs[i].p); 539| 75.3k| dav1d_ref_dec(&c->refs[i].segmap); 540| 75.3k| dav1d_ref_dec(&c->refs[i].refmvs); 541| 75.3k| dav1d_cdf_thread_unref(&c->cdf[i]); 542| 75.3k| } 543| 9.41k| c->frame_hdr = NULL; 544| 9.41k| c->seq_hdr = NULL; 545| 9.41k| dav1d_ref_dec(&c->seq_hdr_ref); 546| | 547| 9.41k| c->mastering_display = NULL; 548| 9.41k| c->content_light = NULL; 549| 9.41k| c->itut_t35 = NULL; 550| 9.41k| c->n_itut_t35 = 0; 551| 9.41k| dav1d_ref_dec(&c->mastering_display_ref); 552| 9.41k| dav1d_ref_dec(&c->content_light_ref); 553| 9.41k| dav1d_ref_dec(&c->itut_t35_ref); 554| | 555| 9.41k| dav1d_data_props_unref_internal(&c->cached_error_props); 556| | 557| 9.41k| if (c->n_fc == 1 && c->n_tc == 1) return; ------------------ | Branch (557:9): [True: 0, False: 9.41k] | Branch (557:25): [True: 0, False: 0] ------------------ 558| 9.41k| atomic_store(c->flush, 1); 559| | 560| 9.41k| if (c->n_tc > 1) { ------------------ | Branch (560:9): [True: 9.41k, False: 0] ------------------ 561| 9.41k| pthread_mutex_lock(&c->task_thread.lock); 562| | // stop running tasks in worker threads 563| 47.0k| for (unsigned i = 0; i < c->n_tc; i++) { ------------------ | Branch (563:30): [True: 37.6k, False: 9.41k] ------------------ 564| 37.6k| Dav1dTaskContext *const tc = &c->tc[i]; 565| 40.0k| while (!tc->task_thread.flushed) { ------------------ | Branch (565:20): [True: 2.33k, False: 37.6k] ------------------ 566| 2.33k| pthread_cond_wait(&tc->task_thread.td.cond, &c->task_thread.lock); 567| 2.33k| } 568| 37.6k| } 569| 47.0k| for (unsigned i = 0; i < c->n_fc; i++) { ------------------ | Branch (569:30): [True: 37.6k, False: 9.41k] ------------------ 570| 37.6k| c->fc[i].task_thread.task_head = NULL; 571| 37.6k| c->fc[i].task_thread.task_tail = NULL; 572| 37.6k| c->fc[i].task_thread.task_cur_prev = NULL; 573| 37.6k| c->fc[i].task_thread.pending_tasks.head = NULL; 574| 37.6k| c->fc[i].task_thread.pending_tasks.tail = NULL; 575| 37.6k| atomic_init(&c->fc[i].task_thread.pending_tasks.merge, 0); 576| 37.6k| } 577| 9.41k| atomic_init(&c->task_thread.first, 0); 578| 9.41k| c->task_thread.cur = c->n_fc; 579| 9.41k| atomic_store(&c->task_thread.reset_task_cur, UINT_MAX); 580| 9.41k| atomic_store(&c->task_thread.cond_signaled, 0); 581| 9.41k| pthread_mutex_unlock(&c->task_thread.lock); 582| 9.41k| } 583| | 584| 9.41k| if (c->n_fc > 1) { ------------------ | Branch (584:9): [True: 9.41k, False: 0] ------------------ 585| 47.0k| for (unsigned n = 0, next = c->frame_thread.next; n < c->n_fc; n++, next++) { ------------------ | Branch (585:59): [True: 37.6k, False: 9.41k] ------------------ 586| 37.6k| if (next == c->n_fc) next = 0; ------------------ | Branch (586:17): [True: 7.81k, False: 29.8k] ------------------ 587| 37.6k| Dav1dFrameContext *const f = &c->fc[next]; 588| 37.6k| dav1d_decode_frame_exit(f, -1); 589| 37.6k| f->n_tile_data = 0; 590| 37.6k| f->task_thread.retval = 0; 591| 37.6k| f->task_thread.error = 0; 592| 37.6k| Dav1dThreadPicture *out_delayed = &c->frame_thread.out_delayed[next]; 593| 37.6k| if (out_delayed->p.frame_hdr) { ------------------ | Branch (593:17): [True: 4.31k, False: 33.3k] ------------------ 594| 4.31k| dav1d_thread_picture_unref(out_delayed); 595| 4.31k| } 596| 37.6k| } 597| 9.41k| c->frame_thread.next = 0; 598| 9.41k| } 599| | atomic_store(c->flush, 0); 600| 9.41k|} dav1d_close: 602| 9.41k|COLD void dav1d_close(Dav1dContext **const c_out) { 603| 9.41k| validate_input(c_out != NULL); ------------------ | | 59| 9.41k|#define validate_input(x) validate_input_or_ret(x, ) | | ------------------ | | | | 52| 9.41k| if (!(x)) { \ | | | | ------------------ | | | | | Branch (52:9): [True: 0, False: 9.41k] | | | | ------------------ | | | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | | | ------------------ | | | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | | | ------------------ | | | | 54| 0| #x, __func__); \ | | | | 55| 0| debug_abort(); \ | | | | ------------------ | | | | | | 39| 0|#define debug_abort abort | | | | ------------------ | | | | 56| 0| return r; \ | | | | 57| 0| } | | ------------------ ------------------ 604| |#if TRACK_HEAP_ALLOCATIONS 605| | dav1d_log_alloc_stats(*c_out); 606| |#endif 607| 9.41k| close_internal(c_out, 1); 608| 9.41k|} dav1d_picture_unref: 727| 176k|void dav1d_picture_unref(Dav1dPicture *const p) { 728| 176k| dav1d_picture_unref_internal(p); 729| 176k|} dav1d_data_create: 731| 411k|uint8_t *dav1d_data_create(Dav1dData *const buf, const size_t sz) { 732| 411k| return dav1d_data_create_internal(buf, sz); 733| 411k|} dav1d_data_unref: 756| 41.2k|void dav1d_data_unref(Dav1dData *const buf) { 757| 41.2k| dav1d_data_unref_internal(buf); 758| 41.2k|} lib.c:get_num_threads: 111| 9.41k|{ 112| | /* ceil(sqrt(n)) */ 113| 9.41k| static const uint8_t fc_lut[49] = { 114| 9.41k| 1, /* 1 */ 115| 9.41k| 2, 2, 2, /* 2- 4 */ 116| 9.41k| 3, 3, 3, 3, 3, /* 5- 9 */ 117| 9.41k| 4, 4, 4, 4, 4, 4, 4, /* 10-16 */ 118| 9.41k| 5, 5, 5, 5, 5, 5, 5, 5, 5, /* 17-25 */ 119| 9.41k| 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, /* 26-36 */ 120| 9.41k| 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, /* 37-49 */ 121| 9.41k| }; 122| 9.41k| *n_tc = s->n_threads ? s->n_threads : ------------------ | Branch (122:13): [True: 9.41k, False: 0] ------------------ 123| 9.41k| iclip(dav1d_num_logical_processors(c), 1, DAV1D_MAX_THREADS); ------------------ | | 46| 0|#define DAV1D_MAX_THREADS 256 ------------------ 124| 9.41k| *n_fc = s->max_frame_delay ? umin(s->max_frame_delay, *n_tc) : ------------------ | Branch (124:13): [True: 9.41k, False: 0] ------------------ 125| 9.41k| *n_tc < 50 ? fc_lut[*n_tc - 1] : 8; // min(8, ceil(sqrt(n))) ------------------ | Branch (125:13): [True: 0, False: 0] ------------------ 126| 9.41k|} lib.c:init_internal: 53| 1|static COLD void init_internal(void) { 54| 1| dav1d_init_cpu(); 55| 1| dav1d_init_ii_wedge_masks(); 56| 1| dav1d_init_intra_edge_tree(); 57| 1| dav1d_init_qm_tables(); 58| 1| dav1d_init_thread(); ------------------ | | 144| 1|#define dav1d_init_thread() do {} while (0) | | ------------------ | | | Branch (144:42): [Folded, False: 1] | | ------------------ ------------------ 59| 1|} lib.c:get_stack_size_internal: 93| 9.41k|static COLD size_t get_stack_size_internal(const pthread_attr_t *const thread_attr) { 94| | /* glibc has an issue where the size of the TLS is subtracted from the stack 95| | * size instead of allocated separately. As a result the specified stack 96| | * size may be insufficient when used in an application with large amounts 97| | * of TLS data. The following is a workaround to compensate for that. 98| | * See https://sourceware.org/bugzilla/show_bug.cgi?id=11787 */ 99| 9.41k| size_t (*const get_minstack)(const pthread_attr_t*) = 100| 9.41k| dlsym(RTLD_DEFAULT, "__pthread_get_minstack"); 101| 9.41k| if (get_minstack) ------------------ | Branch (101:9): [True: 9.41k, False: 0] ------------------ 102| 9.41k| return get_minstack(thread_attr) - PTHREAD_STACK_MIN; 103| 0| return 0; 104| 9.41k|} lib.c:gen_picture: 413| 821k|{ 414| 821k| Dav1dData *const in = &c->in; 415| | 416| 821k| if (output_picture_ready(c, 0)) ------------------ | Branch (416:9): [True: 337k, False: 484k] ------------------ 417| 337k| return 0; 418| | 419| 633k| while (in->sz > 0) { ------------------ | Branch (419:12): [True: 534k, False: 98.8k] ------------------ 420| 534k| const ptrdiff_t res = dav1d_parse_obus(c, in); 421| 534k| if (res < 0) { ------------------ | Branch (421:13): [True: 41.0k, False: 493k] ------------------ 422| 41.0k| dav1d_data_unref_internal(in); 423| 493k| } else { 424| 493k| assert((size_t)res <= in->sz); ------------------ | Branch (424:13): [True: 493k, False: 0] ------------------ 425| 493k| in->sz -= res; 426| 493k| in->data += res; 427| 493k| if (!in->sz) dav1d_data_unref_internal(in); ------------------ | Branch (427:17): [True: 369k, False: 124k] ------------------ 428| 493k| } 429| 534k| if (output_picture_ready(c, 0)) ------------------ | Branch (429:13): [True: 345k, False: 188k] ------------------ 430| 345k| break; 431| 188k| if (res < 0) ------------------ | Branch (431:13): [True: 40.0k, False: 148k] ------------------ 432| 40.0k| return (int)res; 433| 188k| } 434| | 435| 444k| return 0; 436| 484k|} lib.c:output_picture_ready: 332| 1.60M|static int output_picture_ready(Dav1dContext *const c, const int drain) { 333| 1.60M| if (c->cached_error) return 1; ------------------ | Branch (333:9): [True: 351k, False: 1.25M] ------------------ 334| 1.25M| if (!c->all_layers && c->max_spatial_id) { ------------------ | Branch (334:9): [True: 0, False: 1.25M] | Branch (334:27): [True: 0, False: 0] ------------------ 335| 0| if (c->out.p.data[0] && c->cache.p.data[0]) { ------------------ | Branch (335:13): [True: 0, False: 0] | Branch (335:33): [True: 0, False: 0] ------------------ 336| 0| if (c->max_spatial_id == c->cache.p.frame_hdr->spatial_id || ------------------ | Branch (336:17): [True: 0, False: 0] ------------------ 337| 0| c->out.flags & PICTURE_FLAG_NEW_TEMPORAL_UNIT) ------------------ | Branch (337:17): [True: 0, False: 0] ------------------ 338| 0| return 1; 339| 0| dav1d_thread_picture_unref(&c->cache); 340| 0| dav1d_thread_picture_move_ref(&c->cache, &c->out); 341| 0| return 0; 342| 0| } else if (c->cache.p.data[0] && drain) { ------------------ | Branch (342:20): [True: 0, False: 0] | Branch (342:42): [True: 0, False: 0] ------------------ 343| 0| return 1; 344| 0| } else if (c->out.p.data[0]) { ------------------ | Branch (344:20): [True: 0, False: 0] ------------------ 345| 0| dav1d_thread_picture_move_ref(&c->cache, &c->out); 346| 0| return 0; 347| 0| } 348| 0| } 349| | 350| 1.25M| return !!c->out.p.data[0]; 351| 1.25M|} lib.c:output_image: 312| 176k|{ 313| 176k| int res = 0; 314| | 315| 176k| Dav1dThreadPicture *const in = (c->all_layers || !c->max_spatial_id) ------------------ | Branch (315:37): [True: 176k, False: 0] | Branch (315:54): [True: 0, False: 0] ------------------ 316| 176k| ? &c->out : &c->cache; 317| 176k| if (!c->apply_grain || !has_grain(&in->p)) { ------------------ | Branch (317:9): [True: 0, False: 176k] | Branch (317:28): [True: 166k, False: 10.1k] ------------------ 318| 166k| dav1d_picture_move_ref(out, &in->p); 319| 166k| dav1d_thread_picture_unref(in); 320| 166k| goto end; 321| 166k| } 322| | 323| 10.1k| res = dav1d_apply_grain(c, out, &in->p); 324| 10.1k| dav1d_thread_picture_unref(in); 325| 176k|end: 326| 176k| if (!c->all_layers && c->max_spatial_id && c->out.p.data[0]) { ------------------ | Branch (326:9): [True: 0, False: 176k] | Branch (326:27): [True: 0, False: 0] | Branch (326:48): [True: 0, False: 0] ------------------ 327| 0| dav1d_thread_picture_move_ref(in, &c->out); 328| 0| } 329| 176k| return res; 330| 10.1k|} lib.c:drain_picture: 353| 17.3k|static int drain_picture(Dav1dContext *const c, Dav1dPicture *const out) { 354| 17.3k| unsigned drain_count = 0; 355| 17.3k| int drained = 0; 356| 48.9k| do { 357| 48.9k| const unsigned next = c->frame_thread.next; 358| 48.9k| Dav1dFrameContext *const f = &c->fc[next]; 359| 48.9k| pthread_mutex_lock(&c->task_thread.lock); 360| 59.0k| while (f->n_tile_data > 0) ------------------ | Branch (360:16): [True: 10.0k, False: 48.9k] ------------------ 361| 10.0k| pthread_cond_wait(&f->task_thread.cond, 362| 10.0k| &f->task_thread.ttd->lock); 363| 48.9k| Dav1dThreadPicture *const out_delayed = 364| 48.9k| &c->frame_thread.out_delayed[next]; 365| 48.9k| if (out_delayed->p.data[0] || atomic_load(&f->task_thread.error)) { ------------------ | Branch (365:13): [True: 13.0k, False: 35.9k] | Branch (365:39): [True: 53, False: 35.8k] ------------------ 366| 13.0k| unsigned first = atomic_load(&c->task_thread.first); 367| 13.0k| if (first + 1U < c->n_fc) ------------------ | Branch (367:17): [True: 12.0k, False: 1.03k] ------------------ 368| 13.0k| atomic_fetch_add(&c->task_thread.first, 1U); 369| 1.03k| else 370| 13.0k| atomic_store(&c->task_thread.first, 0); 371| 13.0k| atomic_compare_exchange_strong(&c->task_thread.reset_task_cur, 372| 13.0k| &first, UINT_MAX); 373| 13.0k| if (c->task_thread.cur && c->task_thread.cur < c->n_fc) ------------------ | Branch (373:17): [True: 13.0k, False: 14] | Branch (373:39): [True: 1.92k, False: 11.1k] ------------------ 374| 1.92k| c->task_thread.cur--; 375| 13.0k| drained = 1; 376| 35.8k| } else if (drained) { ------------------ | Branch (376:20): [True: 350, False: 35.5k] ------------------ 377| 350| pthread_mutex_unlock(&c->task_thread.lock); 378| 350| break; 379| 350| } 380| 48.5k| if (++c->frame_thread.next == c->n_fc) ------------------ | Branch (380:13): [True: 12.1k, False: 36.4k] ------------------ 381| 12.1k| c->frame_thread.next = 0; 382| 48.5k| pthread_mutex_unlock(&c->task_thread.lock); 383| 48.5k| const int error = f->task_thread.retval; 384| 48.5k| if (error) { ------------------ | Branch (384:13): [True: 4.11k, False: 44.4k] ------------------ 385| 4.11k| f->task_thread.retval = 0; 386| 4.11k| dav1d_data_props_copy(&c->cached_error_props, &out_delayed->p.m); 387| 4.11k| dav1d_thread_picture_unref(out_delayed); 388| 4.11k| return error; 389| 4.11k| } 390| 44.4k| if (out_delayed->p.data[0]) { ------------------ | Branch (390:13): [True: 8.92k, False: 35.5k] ------------------ 391| 8.92k| const unsigned progress = 392| 8.92k| atomic_load_explicit(&out_delayed->progress[1], 393| 8.92k| memory_order_relaxed); 394| 8.92k| if ((out_delayed->visible || c->output_invisible_frames) && ------------------ | Branch (394:18): [True: 7.83k, False: 1.08k] | Branch (394:42): [True: 0, False: 1.08k] ------------------ 395| 7.83k| progress != FRAME_ERROR) ------------------ | | 35| 7.83k|#define FRAME_ERROR (UINT_MAX - 1) ------------------ | Branch (395:17): [True: 7.68k, False: 151] ------------------ 396| 7.68k| { 397| 7.68k| dav1d_thread_picture_ref(&c->out, out_delayed); 398| 7.68k| c->event_flags |= dav1d_picture_get_event_flags(out_delayed); 399| 7.68k| } 400| 8.92k| dav1d_thread_picture_unref(out_delayed); 401| 8.92k| if (output_picture_ready(c, 0)) ------------------ | Branch (401:17): [True: 7.68k, False: 1.23k] ------------------ 402| 7.68k| return output_image(c, out); 403| 8.92k| } 404| 44.4k| } while (++drain_count < c->n_fc); ------------------ | Branch (404:14): [True: 31.6k, False: 5.19k] ------------------ 405| | 406| 5.54k| if (output_picture_ready(c, 1)) ------------------ | Branch (406:9): [True: 0, False: 5.54k] ------------------ 407| 0| return output_image(c, out); 408| | 409| 5.54k| return DAV1D_ERR(EAGAIN); ------------------ | | 58| 5.54k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 410| 5.54k|} lib.c:has_grain: 304| 186k|{ 305| 186k| const Dav1dFilmGrainData *fgdata = &pic->frame_hdr->film_grain.data; 306| 186k| return fgdata->num_y_points || fgdata->num_uv_points[0] || ------------------ | Branch (306:12): [True: 9.13k, False: 177k] | Branch (306:36): [True: 1.96k, False: 175k] ------------------ 307| 175k| fgdata->num_uv_points[1] || (fgdata->clip_to_restricted_range && ------------------ | Branch (307:12): [True: 5.67k, False: 169k] | Branch (307:41): [True: 4.05k, False: 165k] ------------------ 308| 4.05k| fgdata->chroma_scaling_from_luma); ------------------ | Branch (308:41): [True: 3.45k, False: 593] ------------------ 309| 186k|} lib.c:close_internal: 610| 9.41k|static COLD void close_internal(Dav1dContext **const c_out, int flush) { 611| 9.41k| Dav1dContext *const c = *c_out; 612| 9.41k| if (!c) return; ------------------ | Branch (612:9): [True: 0, False: 9.41k] ------------------ 613| | 614| 9.41k| if (flush) dav1d_flush(c); ------------------ | Branch (614:9): [True: 9.41k, False: 0] ------------------ 615| | 616| 9.41k| if (c->tc) { ------------------ | Branch (616:9): [True: 9.41k, False: 0] ------------------ 617| 9.41k| struct TaskThreadData *ttd = &c->task_thread; 618| 9.41k| if (ttd->inited) { ------------------ | Branch (618:13): [True: 9.41k, False: 0] ------------------ 619| 9.41k| pthread_mutex_lock(&ttd->lock); 620| 47.0k| for (unsigned n = 0; n < c->n_tc && c->tc[n].task_thread.td.inited; n++) ------------------ | Branch (620:34): [True: 37.6k, False: 9.41k] | Branch (620:49): [True: 37.6k, False: 0] ------------------ 621| 37.6k| c->tc[n].task_thread.die = 1; 622| 9.41k| pthread_cond_broadcast(&ttd->cond); 623| 9.41k| pthread_mutex_unlock(&ttd->lock); 624| 47.0k| for (unsigned n = 0; n < c->n_tc; n++) { ------------------ | Branch (624:34): [True: 37.6k, False: 9.41k] ------------------ 625| 37.6k| Dav1dTaskContext *const pf = &c->tc[n]; 626| 37.6k| if (!pf->task_thread.td.inited) break; ------------------ | Branch (626:21): [True: 0, False: 37.6k] ------------------ 627| 37.6k| pthread_join(pf->task_thread.td.thread, NULL); 628| 37.6k| pthread_cond_destroy(&pf->task_thread.td.cond); 629| 37.6k| pthread_mutex_destroy(&pf->task_thread.td.lock); 630| 37.6k| } 631| 9.41k| pthread_cond_destroy(&ttd->delayed_fg.cond); 632| 9.41k| pthread_cond_destroy(&ttd->cond); 633| 9.41k| pthread_mutex_destroy(&ttd->lock); 634| 9.41k| } 635| 9.41k| dav1d_free_aligned(c->tc); ------------------ | | 136| 9.41k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 636| 9.41k| } 637| | 638| 47.0k| for (unsigned n = 0; c->fc && n < c->n_fc; n++) { ------------------ | Branch (638:26): [True: 47.0k, False: 0] | Branch (638:35): [True: 37.6k, False: 9.41k] ------------------ 639| 37.6k| Dav1dFrameContext *const f = &c->fc[n]; 640| | 641| | // clean-up threading stuff 642| 37.6k| if (c->n_fc > 1) { ------------------ | Branch (642:13): [True: 37.6k, False: 0] ------------------ 643| 37.6k| dav1d_free(f->tile_thread.lowest_pixel_mem); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 644| 37.6k| dav1d_free(f->frame_thread.b); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 645| 37.6k| dav1d_free_aligned(f->frame_thread.cbi); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 646| 37.6k| dav1d_free_aligned(f->frame_thread.pal_idx); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 647| 37.6k| dav1d_free_aligned(f->frame_thread.cf); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 648| 37.6k| dav1d_free(f->frame_thread.tile_start_off); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 649| 37.6k| dav1d_free_aligned(f->frame_thread.pal); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 650| 37.6k| } 651| 37.6k| if (c->n_tc > 1) { ------------------ | Branch (651:13): [True: 37.6k, False: 0] ------------------ 652| 37.6k| pthread_mutex_destroy(&f->task_thread.pending_tasks.lock); 653| 37.6k| pthread_cond_destroy(&f->task_thread.cond); 654| 37.6k| pthread_mutex_destroy(&f->task_thread.lock); 655| 37.6k| } 656| 37.6k| dav1d_free(f->frame_thread.frame_progress); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 657| 37.6k| dav1d_free(f->task_thread.tasks); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 658| 37.6k| dav1d_free(f->task_thread.tile_tasks[0]); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 659| 37.6k| dav1d_free_aligned(f->ts); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 660| 37.6k| dav1d_free_aligned(f->ipred_edge[0]); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 661| 37.6k| dav1d_free(f->a); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 662| 37.6k| dav1d_free(f->tile); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 663| 37.6k| dav1d_free(f->lf.mask); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 664| 37.6k| dav1d_free(f->lf.level); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 665| 37.6k| dav1d_free(f->lf.lr_mask); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 666| 37.6k| dav1d_free(f->lf.tx_lpf_right_edge[0]); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 667| 37.6k| dav1d_free(f->lf.start_of_tile_row); ------------------ | | 135| 37.6k|#define dav1d_free(ptr) free(ptr) ------------------ 668| 37.6k| dav1d_free_aligned(f->rf.r); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 669| 37.6k| dav1d_free_aligned(f->lf.cdef_line_buf); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 670| 37.6k| dav1d_free_aligned(f->lf.lr_line_buf); ------------------ | | 136| 37.6k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 671| 37.6k| } 672| 9.41k| dav1d_free_aligned(c->fc); ------------------ | | 136| 9.41k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 673| 9.41k| if (c->n_fc > 1 && c->frame_thread.out_delayed) { ------------------ | Branch (673:9): [True: 9.41k, False: 0] | Branch (673:24): [True: 9.41k, False: 0] ------------------ 674| 47.0k| for (unsigned n = 0; n < c->n_fc; n++) ------------------ | Branch (674:30): [True: 37.6k, False: 9.41k] ------------------ 675| 37.6k| if (c->frame_thread.out_delayed[n].p.frame_hdr) ------------------ | Branch (675:17): [True: 0, False: 37.6k] ------------------ 676| 0| dav1d_thread_picture_unref(&c->frame_thread.out_delayed[n]); 677| 9.41k| dav1d_free(c->frame_thread.out_delayed); ------------------ | | 135| 9.41k|#define dav1d_free(ptr) free(ptr) ------------------ 678| 9.41k| } 679| 9.52k| for (int n = 0; n < c->n_tile_data; n++) ------------------ | Branch (679:21): [True: 104, False: 9.41k] ------------------ 680| 104| dav1d_data_unref_internal(&c->tile[n].data); 681| 9.41k| dav1d_free(c->tile); ------------------ | | 135| 9.41k|#define dav1d_free(ptr) free(ptr) ------------------ 682| 84.7k| for (int n = 0; n < 8; n++) { ------------------ | Branch (682:21): [True: 75.3k, False: 9.41k] ------------------ 683| 75.3k| dav1d_cdf_thread_unref(&c->cdf[n]); 684| 75.3k| if (c->refs[n].p.p.frame_hdr) ------------------ | Branch (684:13): [True: 0, False: 75.3k] ------------------ 685| 0| dav1d_thread_picture_unref(&c->refs[n].p); 686| 75.3k| dav1d_ref_dec(&c->refs[n].refmvs); 687| 75.3k| dav1d_ref_dec(&c->refs[n].segmap); 688| 75.3k| } 689| 9.41k| dav1d_ref_dec(&c->seq_hdr_ref); 690| 9.41k| dav1d_ref_dec(&c->frame_hdr_ref); 691| | 692| 9.41k| dav1d_ref_dec(&c->mastering_display_ref); 693| 9.41k| dav1d_ref_dec(&c->content_light_ref); 694| 9.41k| dav1d_ref_dec(&c->itut_t35_ref); 695| | 696| 9.41k| dav1d_mem_pool_end(c->seq_hdr_pool); 697| 9.41k| dav1d_mem_pool_end(c->frame_hdr_pool); 698| 9.41k| dav1d_mem_pool_end(c->segmap_pool); 699| 9.41k| dav1d_mem_pool_end(c->refmvs_pool); 700| 9.41k| dav1d_mem_pool_end(c->cdf_pool); 701| 9.41k| dav1d_mem_pool_end(c->picture_pool); 702| 9.41k| dav1d_mem_pool_end(c->pic_ctx_pool); 703| | 704| 9.41k| dav1d_freep_aligned(c_out); 705| 9.41k|} dav1d_loop_filter_dsp_init_8bpc: 259| 3.46k|COLD void bitfn(dav1d_loop_filter_dsp_init)(Dav1dLoopFilterDSPContext *const c) { 260| 3.46k| c->loop_filter_sb[0][0] = loop_filter_h_sb128y_c; 261| 3.46k| c->loop_filter_sb[0][1] = loop_filter_v_sb128y_c; 262| 3.46k| c->loop_filter_sb[1][0] = loop_filter_h_sb128uv_c; 263| 3.46k| c->loop_filter_sb[1][1] = loop_filter_v_sb128uv_c; 264| | 265| 3.46k|#if HAVE_ASM 266| |#if ARCH_AARCH64 || ARCH_ARM 267| | loop_filter_dsp_init_arm(c); 268| |#elif ARCH_LOONGARCH64 269| | loop_filter_dsp_init_loongarch(c); 270| |#elif ARCH_PPC64LE 271| | loop_filter_dsp_init_ppc(c); 272| |#elif ARCH_X86 273| | loop_filter_dsp_init_x86(c); 274| 3.46k|#endif 275| 3.46k|#endif 276| 3.46k|} dav1d_loop_filter_dsp_init_16bpc: 259| 5.11k|COLD void bitfn(dav1d_loop_filter_dsp_init)(Dav1dLoopFilterDSPContext *const c) { 260| 5.11k| c->loop_filter_sb[0][0] = loop_filter_h_sb128y_c; 261| 5.11k| c->loop_filter_sb[0][1] = loop_filter_v_sb128y_c; 262| 5.11k| c->loop_filter_sb[1][0] = loop_filter_h_sb128uv_c; 263| 5.11k| c->loop_filter_sb[1][1] = loop_filter_v_sb128uv_c; 264| | 265| 5.11k|#if HAVE_ASM 266| |#if ARCH_AARCH64 || ARCH_ARM 267| | loop_filter_dsp_init_arm(c); 268| |#elif ARCH_LOONGARCH64 269| | loop_filter_dsp_init_loongarch(c); 270| |#elif ARCH_PPC64LE 271| | loop_filter_dsp_init_ppc(c); 272| |#elif ARCH_X86 273| | loop_filter_dsp_init_x86(c); 274| 5.11k|#endif 275| 5.11k|#endif 276| 5.11k|} dav1d_loop_restoration_dsp_init_8bpc: 1367| 3.46k|{ 1368| 3.46k| c->wiener[0] = c->wiener[1] = wiener_c; 1369| 3.46k| c->sgr[0] = sgr_5x5_c; 1370| 3.46k| c->sgr[1] = sgr_3x3_c; 1371| 3.46k| c->sgr[2] = sgr_mix_c; 1372| | 1373| 3.46k|#if HAVE_ASM 1374| |#if ARCH_AARCH64 || ARCH_ARM 1375| | loop_restoration_dsp_init_arm(c, bpc); 1376| |#elif ARCH_LOONGARCH64 1377| | loop_restoration_dsp_init_loongarch(c, bpc); 1378| |#elif ARCH_PPC64LE 1379| | loop_restoration_dsp_init_ppc(c, bpc); 1380| |#elif ARCH_X86 1381| | loop_restoration_dsp_init_x86(c, bpc); 1382| 3.46k|#endif 1383| 3.46k|#endif 1384| 3.46k|} looprestoration_tmpl.c:sgr_5x5_c: 830| 7.58k|{ 831| 7.58k| ALIGN_STK_16(int32_t, sumsq_buf, BUF_STRIDE * 5 + 16,); ------------------ | | 100| 7.58k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.58k| line __attribute__((aligned(align))) | | ------------------ ------------------ 832| 7.58k| ALIGN_STK_16(coef, sum_buf, BUF_STRIDE * 5 + 16,); ------------------ | | 100| 7.58k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.58k| line __attribute__((aligned(align))) | | ------------------ ------------------ 833| 7.58k| int32_t *sumsq_ptrs[5], *sumsq_rows[5]; 834| 7.58k| coef *sum_ptrs[5], *sum_rows[5]; 835| 45.5k| for (int i = 0; i < 5; i++) { ------------------ | Branch (835:21): [True: 37.9k, False: 7.58k] ------------------ 836| 37.9k| sumsq_rows[i] = &sumsq_buf[i * BUF_STRIDE]; ------------------ | | 685| 37.9k|#define BUF_STRIDE (384 + 16) ------------------ 837| 37.9k| sum_rows[i] = &sum_buf[i * BUF_STRIDE]; ------------------ | | 685| 37.9k|#define BUF_STRIDE (384 + 16) ------------------ 838| 37.9k| } 839| | 840| 7.58k| ALIGN_STK_16(int32_t, A_buf, BUF_STRIDE * 2 + 16,); ------------------ | | 100| 7.58k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.58k| line __attribute__((aligned(align))) | | ------------------ ------------------ 841| 7.58k| ALIGN_STK_16(coef, B_buf, BUF_STRIDE * 2 + 16,); ------------------ | | 100| 7.58k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.58k| line __attribute__((aligned(align))) | | ------------------ ------------------ 842| 7.58k| int32_t *A_ptrs[2]; 843| 7.58k| coef *B_ptrs[2]; 844| 22.7k| for (int i = 0; i < 2; i++) { ------------------ | Branch (844:21): [True: 15.1k, False: 7.58k] ------------------ 845| 15.1k| A_ptrs[i] = &A_buf[i * BUF_STRIDE]; ------------------ | | 685| 15.1k|#define BUF_STRIDE (384 + 16) ------------------ 846| 15.1k| B_ptrs[i] = &B_buf[i * BUF_STRIDE]; ------------------ | | 685| 15.1k|#define BUF_STRIDE (384 + 16) ------------------ 847| 15.1k| } 848| 7.58k| const pixel *src = dst; 849| 7.58k| const pixel *lpf_bottom = lpf + 6*PXSTRIDE(stride); ------------------ | | 53| 7.58k|#define PXSTRIDE(x) (x) ------------------ 850| | 851| 7.58k| if (edges & LR_HAVE_TOP) { ------------------ | Branch (851:9): [True: 4.90k, False: 2.68k] ------------------ 852| 4.90k| sumsq_ptrs[0] = sumsq_rows[0]; 853| 4.90k| sumsq_ptrs[1] = sumsq_rows[0]; 854| 4.90k| sumsq_ptrs[2] = sumsq_rows[1]; 855| 4.90k| sumsq_ptrs[3] = sumsq_rows[2]; 856| 4.90k| sumsq_ptrs[4] = sumsq_rows[3]; 857| 4.90k| sum_ptrs[0] = sum_rows[0]; 858| 4.90k| sum_ptrs[1] = sum_rows[0]; 859| 4.90k| sum_ptrs[2] = sum_rows[1]; 860| 4.90k| sum_ptrs[3] = sum_rows[2]; 861| 4.90k| sum_ptrs[4] = sum_rows[3]; 862| | 863| 4.90k| sgr_box5_row_h(sumsq_rows[0], sum_rows[0], NULL, lpf, w, edges); 864| 4.90k| lpf += PXSTRIDE(stride); ------------------ | | 53| 4.90k|#define PXSTRIDE(x) (x) ------------------ 865| 4.90k| sgr_box5_row_h(sumsq_rows[1], sum_rows[1], NULL, lpf, w, edges); 866| | 867| 4.90k| sgr_box5_row_h(sumsq_rows[2], sum_rows[2], left, src, w, edges); 868| 4.90k| left++; 869| 4.90k| src += PXSTRIDE(stride); ------------------ | | 53| 4.90k|#define PXSTRIDE(x) (x) ------------------ 870| | 871| 4.90k| if (--h <= 0) ------------------ | Branch (871:13): [True: 605, False: 4.30k] ------------------ 872| 605| goto vert_1; 873| | 874| 4.30k| sgr_box5_row_h(sumsq_rows[3], sum_rows[3], left, src, w, edges); 875| 4.30k| left++; 876| 4.30k| src += PXSTRIDE(stride); ------------------ | | 53| 4.30k|#define PXSTRIDE(x) (x) ------------------ 877| 4.30k| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 878| 4.30k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 4.30k|#define BITDEPTH_MAX 0xff ------------------ 879| 4.30k| rotate(A_ptrs, B_ptrs, 2); 880| | 881| 4.30k| if (--h <= 0) ------------------ | Branch (881:13): [True: 417, False: 3.88k] ------------------ 882| 417| goto vert_2; 883| | 884| | // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set 885| | // one of them to point at the previously unused rows[4]. 886| 3.88k| sumsq_ptrs[3] = sumsq_rows[4]; 887| 3.88k| sum_ptrs[3] = sum_rows[4]; 888| 3.88k| } else { 889| 2.68k| sumsq_ptrs[0] = sumsq_rows[0]; 890| 2.68k| sumsq_ptrs[1] = sumsq_rows[0]; 891| 2.68k| sumsq_ptrs[2] = sumsq_rows[0]; 892| 2.68k| sumsq_ptrs[3] = sumsq_rows[0]; 893| 2.68k| sumsq_ptrs[4] = sumsq_rows[0]; 894| 2.68k| sum_ptrs[0] = sum_rows[0]; 895| 2.68k| sum_ptrs[1] = sum_rows[0]; 896| 2.68k| sum_ptrs[2] = sum_rows[0]; 897| 2.68k| sum_ptrs[3] = sum_rows[0]; 898| 2.68k| sum_ptrs[4] = sum_rows[0]; 899| | 900| 2.68k| sgr_box5_row_h(sumsq_rows[0], sum_rows[0], left, src, w, edges); 901| 2.68k| left++; 902| 2.68k| src += PXSTRIDE(stride); ------------------ | | 53| 2.68k|#define PXSTRIDE(x) (x) ------------------ 903| | 904| 2.68k| if (--h <= 0) ------------------ | Branch (904:13): [True: 169, False: 2.51k] ------------------ 905| 169| goto vert_1; 906| | 907| 2.51k| sumsq_ptrs[4] = sumsq_rows[1]; 908| 2.51k| sum_ptrs[4] = sum_rows[1]; 909| | 910| 2.51k| sgr_box5_row_h(sumsq_rows[1], sum_rows[1], left, src, w, edges); 911| 2.51k| left++; 912| 2.51k| src += PXSTRIDE(stride); ------------------ | | 53| 2.51k|#define PXSTRIDE(x) (x) ------------------ 913| | 914| 2.51k| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 915| 2.51k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 2.51k|#define BITDEPTH_MAX 0xff ------------------ 916| 2.51k| rotate(A_ptrs, B_ptrs, 2); 917| | 918| 2.51k| if (--h <= 0) ------------------ | Branch (918:13): [True: 180, False: 2.33k] ------------------ 919| 180| goto vert_2; 920| | 921| 2.33k| sumsq_ptrs[3] = sumsq_rows[2]; 922| 2.33k| sumsq_ptrs[4] = sumsq_rows[3]; 923| 2.33k| sum_ptrs[3] = sum_rows[2]; 924| 2.33k| sum_ptrs[4] = sum_rows[3]; 925| | 926| 2.33k| sgr_box5_row_h(sumsq_rows[2], sum_rows[2], left, src, w, edges); 927| 2.33k| left++; 928| 2.33k| src += PXSTRIDE(stride); ------------------ | | 53| 2.33k|#define PXSTRIDE(x) (x) ------------------ 929| | 930| 2.33k| if (--h <= 0) ------------------ | Branch (930:13): [True: 172, False: 2.15k] ------------------ 931| 172| goto odd; 932| | 933| 2.15k| sgr_box5_row_h(sumsq_rows[3], sum_rows[3], left, src, w, edges); 934| 2.15k| left++; 935| 2.15k| src += PXSTRIDE(stride); ------------------ | | 53| 2.15k|#define PXSTRIDE(x) (x) ------------------ 936| | 937| 2.15k| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 938| 2.15k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 2.15k|#define BITDEPTH_MAX 0xff ------------------ 939| 2.15k| sgr_finish2(&dst, stride, A_ptrs, B_ptrs, 940| 2.15k| w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX); 941| | 942| 2.15k| if (--h <= 0) ------------------ | Branch (942:13): [True: 93, False: 2.06k] ------------------ 943| 93| goto vert_2; 944| | 945| | // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set 946| | // one of them to point at the previously unused rows[4]. 947| 2.06k| sumsq_ptrs[3] = sumsq_rows[4]; 948| 2.06k| sum_ptrs[3] = sum_rows[4]; 949| 2.06k| } 950| | 951| 151k| do { 952| 151k| sgr_box5_row_h(sumsq_ptrs[3], sum_ptrs[3], left, src, w, edges); 953| 151k| left++; 954| 151k| src += PXSTRIDE(stride); ------------------ | | 53| 151k|#define PXSTRIDE(x) (x) ------------------ 955| | 956| 151k| if (--h <= 0) ------------------ | Branch (956:13): [True: 705, False: 150k] ------------------ 957| 705| goto odd; 958| | 959| 150k| sgr_box5_row_h(sumsq_ptrs[4], sum_ptrs[4], left, src, w, edges); 960| 150k| left++; 961| 150k| src += PXSTRIDE(stride); ------------------ | | 53| 150k|#define PXSTRIDE(x) (x) ------------------ 962| | 963| 150k| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 964| 150k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 150k|#define BITDEPTH_MAX 0xff ------------------ 965| 150k| sgr_finish2(&dst, stride, A_ptrs, B_ptrs, 966| 150k| w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX); 967| 150k| } while (--h > 0); ------------------ | Branch (967:14): [True: 145k, False: 5.24k] ------------------ 968| | 969| 5.24k| if (!(edges & LR_HAVE_BOTTOM)) ------------------ | Branch (969:9): [True: 306, False: 4.93k] ------------------ 970| 306| goto vert_2; 971| | 972| 4.93k| sgr_box5_row_h(sumsq_ptrs[3], sum_ptrs[3], NULL, lpf_bottom, w, edges); 973| 4.93k| lpf_bottom += PXSTRIDE(stride); ------------------ | | 53| 4.93k|#define PXSTRIDE(x) (x) ------------------ 974| 4.93k| sgr_box5_row_h(sumsq_ptrs[4], sum_ptrs[4], NULL, lpf_bottom, w, edges); 975| | 976| 5.93k|output_2: 977| 5.93k| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 978| 5.93k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 5.93k|#define BITDEPTH_MAX 0xff ------------------ 979| 5.93k| sgr_finish2(&dst, stride, A_ptrs, B_ptrs, 980| 5.93k| w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX); 981| 5.93k| return; 982| | 983| 996|vert_2: 984| | // Duplicate the last row twice more 985| 996| sumsq_ptrs[3] = sumsq_ptrs[2]; 986| 996| sumsq_ptrs[4] = sumsq_ptrs[2]; 987| 996| sum_ptrs[3] = sum_ptrs[2]; 988| 996| sum_ptrs[4] = sum_ptrs[2]; 989| 996| goto output_2; 990| | 991| 877|odd: 992| | // Copy the last row as padding once 993| 877| sumsq_ptrs[4] = sumsq_ptrs[3]; 994| 877| sum_ptrs[4] = sum_ptrs[3]; 995| | 996| 877| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 997| 877| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 877|#define BITDEPTH_MAX 0xff ------------------ 998| 877| sgr_finish2(&dst, stride, A_ptrs, B_ptrs, 999| 877| w, 2, params->sgr.w0 HIGHBD_TAIL_SUFFIX); 1000| | 1001| 1.65k|output_1: 1002| | // Duplicate the last row twice more 1003| 1.65k| sumsq_ptrs[3] = sumsq_ptrs[2]; 1004| 1.65k| sumsq_ptrs[4] = sumsq_ptrs[2]; 1005| 1.65k| sum_ptrs[3] = sum_ptrs[2]; 1006| 1.65k| sum_ptrs[4] = sum_ptrs[2]; 1007| | 1008| 1.65k| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 1009| 1.65k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 1.65k|#define BITDEPTH_MAX 0xff ------------------ 1010| | // Output only one row 1011| 1.65k| sgr_finish2(&dst, stride, A_ptrs, B_ptrs, 1012| 1.65k| w, 1, params->sgr.w0 HIGHBD_TAIL_SUFFIX); 1013| 1.65k| return; 1014| | 1015| 774|vert_1: 1016| | // Copy the last row as padding once 1017| 774| sumsq_ptrs[4] = sumsq_ptrs[3]; 1018| 774| sum_ptrs[4] = sum_ptrs[3]; 1019| | 1020| 774| sgr_box5_vert(sumsq_ptrs, sum_ptrs, A_ptrs[1], B_ptrs[1], 1021| 774| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 774|#define BITDEPTH_MAX 0xff ------------------ 1022| 774| rotate(A_ptrs, B_ptrs, 2); 1023| | 1024| 774| goto output_1; 1025| 877|} looprestoration_tmpl.c:sgr_box5_row_h: 441| 1.48M|{ 442| 1.48M| sumsq++; 443| 1.48M| sum++; 444| 1.48M| int a = edges & LR_HAVE_LEFT ? (left ? left[0][1] : src[-3]) : src[0]; ------------------ | Branch (444:13): [True: 818k, False: 663k] | Branch (444:37): [True: 769k, False: 48.6k] ------------------ 445| 1.48M| int b = edges & LR_HAVE_LEFT ? (left ? left[0][2] : src[-2]) : src[0]; ------------------ | Branch (445:13): [True: 818k, False: 663k] | Branch (445:37): [True: 769k, False: 48.6k] ------------------ 446| 1.48M| int c = edges & LR_HAVE_LEFT ? (left ? left[0][3] : src[-1]) : src[0]; ------------------ | Branch (446:13): [True: 818k, False: 663k] | Branch (446:37): [True: 769k, False: 48.6k] ------------------ 447| 1.48M| int d = src[0]; 448| 139M| for (int x = -1; x < w + 1; x++) { ------------------ | Branch (448:22): [True: 137M, False: 1.48M] ------------------ 449| 137M| int e = (x + 2 < w || (edges & LR_HAVE_RIGHT)) ? src[x + 2] : src[w - 1]; ------------------ | Branch (449:18): [True: 133M, False: 4.42M] | Branch (449:31): [True: 2.47M, False: 1.95M] ------------------ 450| 137M| sum[x] = a + b + c + d + e; 451| 137M| sumsq[x] = a * a + b * b + c * c + d * d + e * e; 452| 137M| a = b; 453| 137M| b = c; 454| 137M| c = d; 455| 137M| d = e; 456| 137M| } 457| 1.48M|} looprestoration_tmpl.c:sgr_box5_vert: 537| 735k|{ 538| 735k| sgr_box5_row_v(sumsq, sum, sumsq_out, sum_out, w); 539| 735k| sgr_calc_row_ab(sumsq_out, sum_out, w, s, bitdepth_max, 25, 164); 540| 735k| rotate5_x2(sumsq, sum); 541| 735k|} looprestoration_tmpl.c:sgr_box5_row_v: 488| 735k|{ 489| 69.0M| for (int x = 0; x < w + 2; x++) { ------------------ | Branch (489:21): [True: 68.3M, False: 735k] ------------------ 490| 68.3M| int sq_a = sumsq[0][x]; 491| 68.3M| int sq_b = sumsq[1][x]; 492| 68.3M| int sq_c = sumsq[2][x]; 493| 68.3M| int sq_d = sumsq[3][x]; 494| 68.3M| int sq_e = sumsq[4][x]; 495| 68.3M| int s_a = sum[0][x]; 496| 68.3M| int s_b = sum[1][x]; 497| 68.3M| int s_c = sum[2][x]; 498| 68.3M| int s_d = sum[3][x]; 499| 68.3M| int s_e = sum[4][x]; 500| 68.3M| sumsq_out[x] = sq_a + sq_b + sq_c + sq_d + sq_e; 501| 68.3M| sum_out[x] = s_a + s_b + s_c + s_d + s_e; 502| 68.3M| } 503| 735k|} looprestoration_tmpl.c:sgr_calc_row_ab: 507| 2.19M|{ 508| 2.19M| const int bitdepth_min_8 = bitdepth_from_max(bitdepth_max) - 8; ------------------ | | 58| 2.19M|#define bitdepth_from_max(x) 8 ------------------ 509| 188M| for (int i = 0; i < w + 2; i++) { ------------------ | Branch (509:21): [True: 186M, False: 2.19M] ------------------ 510| 186M| const int a = 511| 186M| (AA[i] + ((1 << (2 * bitdepth_min_8)) >> 1)) >> (2 * bitdepth_min_8); 512| 186M| const int b = 513| 186M| (BB[i] + ((1 << bitdepth_min_8) >> 1)) >> bitdepth_min_8; 514| | 515| 186M| const unsigned p = imax(a * n - b * b, 0); 516| 186M| const unsigned z = (p * s + (1 << 19)) >> 20; 517| 186M| const unsigned x = dav1d_sgr_x_by_x[umin(z, 255)]; 518| | 519| | // This is where we invert A and B, so that B is of size coef. 520| 186M| AA[i] = (x * BB[i] * sgr_one_by_x + (1 << 11)) >> 12; 521| 186M| BB[i] = x; 522| 186M| } 523| 2.19M|} looprestoration_tmpl.c:rotate5_x2: 402| 732k|{ 403| 732k| int32_t *tmp32[2]; 404| 732k| coef *tmpc[2]; 405| 2.19M| for (int i = 0; i < 2; i++) { ------------------ | Branch (405:21): [True: 1.46M, False: 732k] ------------------ 406| 1.46M| tmp32[i] = sumsq_ptrs[i]; 407| 1.46M| tmpc[i] = sum_ptrs[i]; 408| 1.46M| } 409| 2.92M| for (int i = 0; i < 3; i++) { ------------------ | Branch (409:21): [True: 2.19M, False: 732k] ------------------ 410| 2.19M| sumsq_ptrs[i] = sumsq_ptrs[i + 2]; 411| 2.19M| sum_ptrs[i] = sum_ptrs[i + 2]; 412| 2.19M| } 413| 2.19M| for (int i = 0; i < 2; i++) { ------------------ | Branch (413:21): [True: 1.46M, False: 732k] ------------------ 414| 1.46M| sumsq_ptrs[3 + i] = tmp32[i]; 415| 1.46M| sum_ptrs[3 + i] = tmpc[i]; 416| 1.46M| } 417| 732k|} looprestoration_tmpl.c:rotate: 390| 3.63M|{ 391| 3.63M| int32_t *tmp32 = sumsq_ptrs[0]; 392| 3.63M| coef *tmpc = sum_ptrs[0]; 393| 11.3M| for (int i = 0; i < n - 1; i++) { ------------------ | Branch (393:21): [True: 7.66M, False: 3.63M] ------------------ 394| 7.66M| sumsq_ptrs[i] = sumsq_ptrs[i + 1]; 395| 7.66M| sum_ptrs[i] = sum_ptrs[i + 1]; 396| 7.66M| } 397| 3.63M| sumsq_ptrs[n - 1] = tmp32; 398| 3.63M| sum_ptrs[n - 1] = tmpc; 399| 3.63M|} looprestoration_tmpl.c:sgr_finish2: 645| 161k|{ 646| 161k| ALIGN_STK_16(coef, tmp, 2*FILTER_OUT_STRIDE,); ------------------ | | 100| 161k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 161k| line __attribute__((aligned(align))) | | ------------------ ------------------ 647| | 648| 161k| sgr_finish_filter2(tmp, *dst, stride, A_ptrs, B_ptrs, w, h); 649| 161k| sgr_weighted_row1(*dst, tmp, w, w1 HIGHBD_TAIL_SUFFIX); 650| 161k| *dst += PXSTRIDE(stride); ------------------ | | 53| 161k|#define PXSTRIDE(x) (x) ------------------ 651| 161k| if (h > 1) { ------------------ | Branch (651:9): [True: 159k, False: 1.75k] ------------------ 652| 159k| sgr_weighted_row1(*dst, tmp + FILTER_OUT_STRIDE, w, w1 HIGHBD_TAIL_SUFFIX); ------------------ | | 572| 159k|#define FILTER_OUT_STRIDE (384) ------------------ 653| 159k| *dst += PXSTRIDE(stride); ------------------ | | 53| 159k|#define PXSTRIDE(x) (x) ------------------ 654| 159k| } 655| 161k| rotate(A_ptrs, B_ptrs, 2); 656| 161k|} looprestoration_tmpl.c:sgr_finish_filter2: 579| 701k|{ 580| 701k|#define SIX_NEIGHBORS(P, i)\ 581| 701k| ((P[0][i] + P[1][i]) * 6 + \ 582| 701k| (P[0][i - 1] + P[1][i - 1] + \ 583| 701k| P[0][i + 1] + P[1][i + 1]) * 5) 584| 64.4M| for (int i = 0; i < w; i++) { ------------------ | Branch (584:21): [True: 63.7M, False: 701k] ------------------ 585| 63.7M| const int a = SIX_NEIGHBORS(B_ptrs, i + 1); ------------------ | | 581| 63.7M| ((P[0][i] + P[1][i]) * 6 + \ | | 582| 63.7M| (P[0][i - 1] + P[1][i - 1] + \ | | 583| 63.7M| P[0][i + 1] + P[1][i + 1]) * 5) ------------------ 586| 63.7M| const int b = SIX_NEIGHBORS(A_ptrs, i + 1); ------------------ | | 581| 63.7M| ((P[0][i] + P[1][i]) * 6 + \ | | 582| 63.7M| (P[0][i - 1] + P[1][i - 1] + \ | | 583| 63.7M| P[0][i + 1] + P[1][i + 1]) * 5) ------------------ 587| 63.7M| tmp[i] = (b - a * src[i] + (1 << 8)) >> 9; 588| 63.7M| } 589| 701k| if (h <= 1) ------------------ | Branch (589:9): [True: 7.02k, False: 694k] ------------------ 590| 7.02k| return; 591| 694k| tmp += FILTER_OUT_STRIDE; ------------------ | | 572| 694k|#define FILTER_OUT_STRIDE (384) ------------------ 592| 694k| src += PXSTRIDE(src_stride); ------------------ | | 53| 694k|#define PXSTRIDE(x) (x) ------------------ 593| 694k| const int32_t *A = &A_ptrs[1][1]; 594| 694k| const coef *B = &B_ptrs[1][1]; 595| 64.8M| for (int i = 0; i < w; i++) { ------------------ | Branch (595:21): [True: 64.1M, False: 694k] ------------------ 596| 64.1M| const int a = B[i] * 6 + (B[i - 1] + B[i + 1]) * 5; 597| 64.1M| const int b = A[i] * 6 + (A[i - 1] + A[i + 1]) * 5; 598| 64.1M| tmp[i] = (b - a * src[i] + (1 << 7)) >> 8; 599| 64.1M| } 600| 694k|#undef SIX_NEIGHBORS 601| 694k|} looprestoration_tmpl.c:sgr_weighted_row1: 605| 639k|{ 606| 72.8M| for (int i = 0; i < w; i++) { ------------------ | Branch (606:21): [True: 72.2M, False: 639k] ------------------ 607| 72.2M| const int v = w1 * t1[i]; 608| 72.2M| dst[i] = iclip_pixel(dst[i] + ((v + (1 << 10)) >> 11)); ------------------ | | 49| 72.2M|#define iclip_pixel iclip_u8 ------------------ 609| 72.2M| } 610| 639k|} looprestoration_tmpl.c:sgr_3x3_c: 684| 7.69k|{ 685| 7.69k|#define BUF_STRIDE (384 + 16) 686| 7.69k| ALIGN_STK_16(int32_t, sumsq_buf, BUF_STRIDE * 3 + 16,); ------------------ | | 100| 7.69k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.69k| line __attribute__((aligned(align))) | | ------------------ ------------------ 687| 7.69k| ALIGN_STK_16(coef, sum_buf, BUF_STRIDE * 3 + 16,); ------------------ | | 100| 7.69k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.69k| line __attribute__((aligned(align))) | | ------------------ ------------------ 688| 7.69k| int32_t *sumsq_ptrs[3], *sumsq_rows[3]; 689| 7.69k| coef *sum_ptrs[3], *sum_rows[3]; 690| 30.7k| for (int i = 0; i < 3; i++) { ------------------ | Branch (690:21): [True: 23.0k, False: 7.69k] ------------------ 691| 23.0k| sumsq_rows[i] = &sumsq_buf[i * BUF_STRIDE]; ------------------ | | 685| 23.0k|#define BUF_STRIDE (384 + 16) ------------------ 692| 23.0k| sum_rows[i] = &sum_buf[i * BUF_STRIDE]; ------------------ | | 685| 23.0k|#define BUF_STRIDE (384 + 16) ------------------ 693| 23.0k| } 694| | 695| 7.69k| ALIGN_STK_16(int32_t, A_buf, BUF_STRIDE * 3 + 16,); ------------------ | | 100| 7.69k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.69k| line __attribute__((aligned(align))) | | ------------------ ------------------ 696| 7.69k| ALIGN_STK_16(coef, B_buf, BUF_STRIDE * 3 + 16,); ------------------ | | 100| 7.69k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 7.69k| line __attribute__((aligned(align))) | | ------------------ ------------------ 697| 7.69k| int32_t *A_ptrs[3]; 698| 7.69k| coef *B_ptrs[3]; 699| 30.7k| for (int i = 0; i < 3; i++) { ------------------ | Branch (699:21): [True: 23.0k, False: 7.69k] ------------------ 700| 23.0k| A_ptrs[i] = &A_buf[i * BUF_STRIDE]; ------------------ | | 685| 23.0k|#define BUF_STRIDE (384 + 16) ------------------ 701| 23.0k| B_ptrs[i] = &B_buf[i * BUF_STRIDE]; ------------------ | | 685| 23.0k|#define BUF_STRIDE (384 + 16) ------------------ 702| 23.0k| } 703| 7.69k| const pixel *src = dst; 704| 7.69k| const pixel *lpf_bottom = lpf + 6*PXSTRIDE(stride); ------------------ | | 53| 7.69k|#define PXSTRIDE(x) (x) ------------------ 705| | 706| 7.69k| if (edges & LR_HAVE_TOP) { ------------------ | Branch (706:9): [True: 5.13k, False: 2.55k] ------------------ 707| 5.13k| sumsq_ptrs[0] = sumsq_rows[0]; 708| 5.13k| sumsq_ptrs[1] = sumsq_rows[1]; 709| 5.13k| sumsq_ptrs[2] = sumsq_rows[2]; 710| 5.13k| sum_ptrs[0] = sum_rows[0]; 711| 5.13k| sum_ptrs[1] = sum_rows[1]; 712| 5.13k| sum_ptrs[2] = sum_rows[2]; 713| | 714| 5.13k| sgr_box3_row_h(sumsq_rows[0], sum_rows[0], NULL, lpf, w, edges); 715| 5.13k| lpf += PXSTRIDE(stride); ------------------ | | 53| 5.13k|#define PXSTRIDE(x) (x) ------------------ 716| 5.13k| sgr_box3_row_h(sumsq_rows[1], sum_rows[1], NULL, lpf, w, edges); 717| | 718| 5.13k| sgr_box3_hv(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 719| 5.13k| left, src, w, params->sgr.s1, edges, BITDEPTH_MAX); ------------------ | | 59| 5.13k|#define BITDEPTH_MAX 0xff ------------------ 720| 5.13k| left++; 721| 5.13k| src += PXSTRIDE(stride); ------------------ | | 53| 5.13k|#define PXSTRIDE(x) (x) ------------------ 722| 5.13k| rotate(A_ptrs, B_ptrs, 3); 723| | 724| 5.13k| if (--h <= 0) ------------------ | Branch (724:13): [True: 757, False: 4.37k] ------------------ 725| 757| goto vert_1; 726| | 727| 4.37k| sgr_box3_hv(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 728| 4.37k| left, src, w, params->sgr.s1, edges, BITDEPTH_MAX); ------------------ | | 59| 4.37k|#define BITDEPTH_MAX 0xff ------------------ 729| 4.37k| left++; 730| 4.37k| src += PXSTRIDE(stride); ------------------ | | 53| 4.37k|#define PXSTRIDE(x) (x) ------------------ 731| 4.37k| rotate(A_ptrs, B_ptrs, 3); 732| | 733| 4.37k| if (--h <= 0) ------------------ | Branch (733:13): [True: 947, False: 3.43k] ------------------ 734| 947| goto vert_2; 735| 4.37k| } else { 736| 2.55k| sumsq_ptrs[0] = sumsq_rows[0]; 737| 2.55k| sumsq_ptrs[1] = sumsq_rows[0]; 738| 2.55k| sumsq_ptrs[2] = sumsq_rows[0]; 739| 2.55k| sum_ptrs[0] = sum_rows[0]; 740| 2.55k| sum_ptrs[1] = sum_rows[0]; 741| 2.55k| sum_ptrs[2] = sum_rows[0]; 742| | 743| 2.55k| sgr_box3_row_h(sumsq_rows[0], sum_rows[0], left, src, w, edges); 744| 2.55k| left++; 745| 2.55k| src += PXSTRIDE(stride); ------------------ | | 53| 2.55k|#define PXSTRIDE(x) (x) ------------------ 746| | 747| 2.55k| sgr_box3_vert(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 748| 2.55k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 2.55k|#define BITDEPTH_MAX 0xff ------------------ 749| 2.55k| rotate(A_ptrs, B_ptrs, 3); 750| | 751| 2.55k| if (--h <= 0) ------------------ | Branch (751:13): [True: 128, False: 2.43k] ------------------ 752| 128| goto vert_1; 753| | 754| 2.43k| sumsq_ptrs[2] = sumsq_rows[1]; 755| 2.43k| sum_ptrs[2] = sum_rows[1]; 756| | 757| 2.43k| sgr_box3_hv(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 758| 2.43k| left, src, w, params->sgr.s1, edges, BITDEPTH_MAX); ------------------ | | 59| 2.43k|#define BITDEPTH_MAX 0xff ------------------ 759| 2.43k| left++; 760| 2.43k| src += PXSTRIDE(stride); ------------------ | | 53| 2.43k|#define PXSTRIDE(x) (x) ------------------ 761| 2.43k| rotate(A_ptrs, B_ptrs, 3); 762| | 763| 2.43k| if (--h <= 0) ------------------ | Branch (763:13): [True: 97, False: 2.33k] ------------------ 764| 97| goto vert_2; 765| | 766| 2.33k| sumsq_ptrs[2] = sumsq_rows[2]; 767| 2.33k| sum_ptrs[2] = sum_rows[2]; 768| 2.33k| } 769| | 770| 305k| do { 771| 305k| sgr_box3_hv(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 772| 305k| left, src, w, params->sgr.s1, edges, BITDEPTH_MAX); ------------------ | | 59| 305k|#define BITDEPTH_MAX 0xff ------------------ 773| 305k| left++; 774| 305k| src += PXSTRIDE(stride); ------------------ | | 53| 305k|#define PXSTRIDE(x) (x) ------------------ 775| | 776| 305k| sgr_finish1(&dst, stride, A_ptrs, B_ptrs, 777| 305k| w, params->sgr.w1 HIGHBD_TAIL_SUFFIX); 778| 305k| } while (--h > 0); ------------------ | Branch (778:14): [True: 299k, False: 5.76k] ------------------ 779| | 780| 5.76k| if (!(edges & LR_HAVE_BOTTOM)) ------------------ | Branch (780:9): [True: 606, False: 5.15k] ------------------ 781| 606| goto vert_2; 782| | 783| 5.15k| sgr_box3_hv(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 784| 5.15k| NULL, lpf_bottom, w, params->sgr.s1, edges, BITDEPTH_MAX); ------------------ | | 59| 5.15k|#define BITDEPTH_MAX 0xff ------------------ 785| 5.15k| lpf_bottom += PXSTRIDE(stride); ------------------ | | 53| 5.15k|#define PXSTRIDE(x) (x) ------------------ 786| | 787| 5.15k| sgr_finish1(&dst, stride, A_ptrs, B_ptrs, 788| 5.15k| w, params->sgr.w1 HIGHBD_TAIL_SUFFIX); 789| | 790| 5.15k| sgr_box3_hv(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 791| 5.15k| NULL, lpf_bottom, w, params->sgr.s1, edges, BITDEPTH_MAX); ------------------ | | 59| 5.15k|#define BITDEPTH_MAX 0xff ------------------ 792| | 793| 5.15k| sgr_finish1(&dst, stride, A_ptrs, B_ptrs, 794| 5.15k| w, params->sgr.w1 HIGHBD_TAIL_SUFFIX); 795| 5.15k| return; 796| | 797| 1.65k|vert_2: 798| 1.65k| sumsq_ptrs[2] = sumsq_ptrs[1]; 799| 1.65k| sum_ptrs[2] = sum_ptrs[1]; 800| 1.65k| sgr_box3_vert(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 801| 1.65k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 1.65k|#define BITDEPTH_MAX 0xff ------------------ 802| | 803| 1.65k| sgr_finish1(&dst, stride, A_ptrs, B_ptrs, 804| 1.65k| w, params->sgr.w1 HIGHBD_TAIL_SUFFIX); 805| | 806| 2.53k|output_1: 807| 2.53k| sumsq_ptrs[2] = sumsq_ptrs[1]; 808| 2.53k| sum_ptrs[2] = sum_ptrs[1]; 809| 2.53k| sgr_box3_vert(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 810| 2.53k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 2.53k|#define BITDEPTH_MAX 0xff ------------------ 811| | 812| 2.53k| sgr_finish1(&dst, stride, A_ptrs, B_ptrs, 813| 2.53k| w, params->sgr.w1 HIGHBD_TAIL_SUFFIX); 814| 2.53k| return; 815| | 816| 885|vert_1: 817| 885| sumsq_ptrs[2] = sumsq_ptrs[1]; 818| 885| sum_ptrs[2] = sum_ptrs[1]; 819| 885| sgr_box3_vert(sumsq_ptrs, sum_ptrs, A_ptrs[2], B_ptrs[2], 820| 885| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 885|#define BITDEPTH_MAX 0xff ------------------ 821| 885| rotate(A_ptrs, B_ptrs, 3); 822| 885| goto output_1; 823| 1.65k|} looprestoration_tmpl.c:sgr_box3_row_h: 423| 1.47M|{ 424| 1.47M| sumsq++; 425| 1.47M| sum++; 426| 1.47M| int a = edges & LR_HAVE_LEFT ? (left ? left[0][2] : src[-2]) : src[0]; ------------------ | Branch (426:13): [True: 914k, False: 562k] | Branch (426:37): [True: 859k, False: 54.7k] ------------------ 427| 1.47M| int b = edges & LR_HAVE_LEFT ? (left ? left[0][3] : src[-1]) : src[0]; ------------------ | Branch (427:13): [True: 914k, False: 562k] | Branch (427:37): [True: 859k, False: 54.8k] ------------------ 428| 147M| for (int x = -1; x < w + 1; x++) { ------------------ | Branch (428:22): [True: 146M, False: 1.47M] ------------------ 429| 146M| int c = (x + 1 < w || (edges & LR_HAVE_RIGHT)) ? src[x + 1] : src[w - 1]; ------------------ | Branch (429:18): [True: 143M, False: 2.95M] | Branch (429:31): [True: 1.82M, False: 1.13M] ------------------ 430| 146M| sum[x] = a + b + c; 431| 146M| sumsq[x] = a * a + b * b + c * c; 432| 146M| a = b; 433| 146M| b = c; 434| 146M| } 435| 1.47M|} looprestoration_tmpl.c:sgr_box3_hv: 550| 327k|{ 551| 327k| sgr_box3_row_h(sumsq[2], sum[2], left, src, w, edges); 552| 327k| sgr_box3_vert(sumsq, sum, AA, BB, w, s, bitdepth_max); 553| 327k|} looprestoration_tmpl.c:sgr_box3_vert: 528| 1.45M|{ 529| 1.45M| sgr_box3_row_v(sumsq, sum, sumsq_out, sum_out, w); 530| 1.45M| sgr_calc_row_ab(sumsq_out, sum_out, w, s, bitdepth_max, 9, 455); 531| 1.45M| rotate(sumsq, sum, 3); 532| 1.45M|} looprestoration_tmpl.c:sgr_box3_row_v: 472| 1.45M|{ 473| 144M| for (int x = 0; x < w + 2; x++) { ------------------ | Branch (473:21): [True: 142M, False: 1.45M] ------------------ 474| 142M| int sq_a = sumsq[0][x]; 475| 142M| int sq_b = sumsq[1][x]; 476| 142M| int sq_c = sumsq[2][x]; 477| 142M| int s_a = sum[0][x]; 478| 142M| int s_b = sum[1][x]; 479| 142M| int s_c = sum[2][x]; 480| 142M| sumsq_out[x] = sq_a + sq_b + sq_c; 481| 142M| sum_out[x] = s_a + s_b + s_c; 482| 142M| } 483| 1.45M|} looprestoration_tmpl.c:sgr_finish1: 631| 319k|{ 632| | // Only one single row, no stride needed 633| 319k| ALIGN_STK_16(coef, tmp, 384,); ------------------ | | 100| 319k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 319k| line __attribute__((aligned(align))) | | ------------------ ------------------ 634| | 635| 319k| sgr_finish_filter_row1(tmp, *dst, A_ptrs, B_ptrs, w); 636| 319k| sgr_weighted_row1(*dst, tmp, w, w1 HIGHBD_TAIL_SUFFIX); 637| 319k| *dst += PXSTRIDE(stride); ------------------ | | 53| 319k|#define PXSTRIDE(x) (x) ------------------ 638| 319k| rotate(A_ptrs, B_ptrs, 3); 639| 319k|} looprestoration_tmpl.c:sgr_finish_filter_row1: 559| 1.39M|{ 560| 1.39M|#define EIGHT_NEIGHBORS(P, i)\ 561| 1.39M| ((P[1][i] + P[1][i - 1] + P[1][i + 1] + P[0][i] + P[2][i]) * 4 + \ 562| 1.39M| (P[0][i - 1] + P[2][i - 1] + \ 563| 1.39M| P[0][i + 1] + P[2][i + 1]) * 3) 564| 132M| for (int i = 0; i < w; i++) { ------------------ | Branch (564:21): [True: 130M, False: 1.39M] ------------------ 565| 130M| const int a = EIGHT_NEIGHBORS(B_ptrs, i + 1); ------------------ | | 561| 130M| ((P[1][i] + P[1][i - 1] + P[1][i + 1] + P[0][i] + P[2][i]) * 4 + \ | | 562| 130M| (P[0][i - 1] + P[2][i - 1] + \ | | 563| 130M| P[0][i + 1] + P[2][i + 1]) * 3) ------------------ 566| 130M| const int b = EIGHT_NEIGHBORS(A_ptrs, i + 1); ------------------ | | 561| 130M| ((P[1][i] + P[1][i - 1] + P[1][i + 1] + P[0][i] + P[2][i]) * 4 + \ | | 562| 130M| (P[0][i - 1] + P[2][i - 1] + \ | | 563| 130M| P[0][i + 1] + P[2][i + 1]) * 3) ------------------ 567| 130M| tmp[i] = (b - a * src[i] + (1 << 8)) >> 9; 568| 130M| } 569| 1.39M|#undef EIGHT_NEIGHBORS 570| 1.39M|} looprestoration_tmpl.c:sgr_mix_c: 1032| 24.1k|{ 1033| 24.1k| ALIGN_STK_16(int32_t, sumsq5_buf, BUF_STRIDE * 5 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1034| 24.1k| ALIGN_STK_16(coef, sum5_buf, BUF_STRIDE * 5 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1035| 24.1k| int32_t *sumsq5_ptrs[5], *sumsq5_rows[5]; 1036| 24.1k| coef *sum5_ptrs[5], *sum5_rows[5]; 1037| 145k| for (int i = 0; i < 5; i++) { ------------------ | Branch (1037:21): [True: 120k, False: 24.1k] ------------------ 1038| 120k| sumsq5_rows[i] = &sumsq5_buf[i * BUF_STRIDE]; ------------------ | | 685| 120k|#define BUF_STRIDE (384 + 16) ------------------ 1039| 120k| sum5_rows[i] = &sum5_buf[i * BUF_STRIDE]; ------------------ | | 685| 120k|#define BUF_STRIDE (384 + 16) ------------------ 1040| 120k| } 1041| 24.1k| ALIGN_STK_16(int32_t, sumsq3_buf, BUF_STRIDE * 3 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1042| 24.1k| ALIGN_STK_16(coef, sum3_buf, BUF_STRIDE * 3 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1043| 24.1k| int32_t *sumsq3_ptrs[3], *sumsq3_rows[3]; 1044| 24.1k| coef *sum3_ptrs[3], *sum3_rows[3]; 1045| 96.7k| for (int i = 0; i < 3; i++) { ------------------ | Branch (1045:21): [True: 72.5k, False: 24.1k] ------------------ 1046| 72.5k| sumsq3_rows[i] = &sumsq3_buf[i * BUF_STRIDE]; ------------------ | | 685| 72.5k|#define BUF_STRIDE (384 + 16) ------------------ 1047| 72.5k| sum3_rows[i] = &sum3_buf[i * BUF_STRIDE]; ------------------ | | 685| 72.5k|#define BUF_STRIDE (384 + 16) ------------------ 1048| 72.5k| } 1049| | 1050| 24.1k| ALIGN_STK_16(int32_t, A5_buf, BUF_STRIDE * 2 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1051| 24.1k| ALIGN_STK_16(coef, B5_buf, BUF_STRIDE * 2 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1052| 24.1k| int32_t *A5_ptrs[2]; 1053| 24.1k| coef *B5_ptrs[2]; 1054| 72.5k| for (int i = 0; i < 2; i++) { ------------------ | Branch (1054:21): [True: 48.3k, False: 24.1k] ------------------ 1055| 48.3k| A5_ptrs[i] = &A5_buf[i * BUF_STRIDE]; ------------------ | | 685| 48.3k|#define BUF_STRIDE (384 + 16) ------------------ 1056| 48.3k| B5_ptrs[i] = &B5_buf[i * BUF_STRIDE]; ------------------ | | 685| 48.3k|#define BUF_STRIDE (384 + 16) ------------------ 1057| 48.3k| } 1058| 24.1k| ALIGN_STK_16(int32_t, A3_buf, BUF_STRIDE * 4 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1059| 24.1k| ALIGN_STK_16(coef, B3_buf, BUF_STRIDE * 4 + 16,); ------------------ | | 100| 24.1k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 24.1k| line __attribute__((aligned(align))) | | ------------------ ------------------ 1060| 24.1k| int32_t *A3_ptrs[4]; 1061| 24.1k| coef *B3_ptrs[4]; 1062| 120k| for (int i = 0; i < 4; i++) { ------------------ | Branch (1062:21): [True: 96.7k, False: 24.1k] ------------------ 1063| 96.7k| A3_ptrs[i] = &A3_buf[i * BUF_STRIDE]; ------------------ | | 685| 96.7k|#define BUF_STRIDE (384 + 16) ------------------ 1064| 96.7k| B3_ptrs[i] = &B3_buf[i * BUF_STRIDE]; ------------------ | | 685| 96.7k|#define BUF_STRIDE (384 + 16) ------------------ 1065| 96.7k| } 1066| 24.1k| const pixel *src = dst; 1067| 24.1k| const pixel *lpf_bottom = lpf + 6*PXSTRIDE(stride); ------------------ | | 53| 24.1k|#define PXSTRIDE(x) (x) ------------------ 1068| | 1069| 24.1k| if (edges & LR_HAVE_TOP) { ------------------ | Branch (1069:9): [True: 16.1k, False: 8.03k] ------------------ 1070| 16.1k| sumsq5_ptrs[0] = sumsq5_rows[0]; 1071| 16.1k| sumsq5_ptrs[1] = sumsq5_rows[0]; 1072| 16.1k| sumsq5_ptrs[2] = sumsq5_rows[1]; 1073| 16.1k| sumsq5_ptrs[3] = sumsq5_rows[2]; 1074| 16.1k| sumsq5_ptrs[4] = sumsq5_rows[3]; 1075| 16.1k| sum5_ptrs[0] = sum5_rows[0]; 1076| 16.1k| sum5_ptrs[1] = sum5_rows[0]; 1077| 16.1k| sum5_ptrs[2] = sum5_rows[1]; 1078| 16.1k| sum5_ptrs[3] = sum5_rows[2]; 1079| 16.1k| sum5_ptrs[4] = sum5_rows[3]; 1080| | 1081| 16.1k| sumsq3_ptrs[0] = sumsq3_rows[0]; 1082| 16.1k| sumsq3_ptrs[1] = sumsq3_rows[1]; 1083| 16.1k| sumsq3_ptrs[2] = sumsq3_rows[2]; 1084| 16.1k| sum3_ptrs[0] = sum3_rows[0]; 1085| 16.1k| sum3_ptrs[1] = sum3_rows[1]; 1086| 16.1k| sum3_ptrs[2] = sum3_rows[2]; 1087| | 1088| 16.1k| sgr_box35_row_h(sumsq3_rows[0], sum3_rows[0], 1089| 16.1k| sumsq5_rows[0], sum5_rows[0], 1090| 16.1k| NULL, lpf, w, edges); 1091| 16.1k| lpf += PXSTRIDE(stride); ------------------ | | 53| 16.1k|#define PXSTRIDE(x) (x) ------------------ 1092| 16.1k| sgr_box35_row_h(sumsq3_rows[1], sum3_rows[1], 1093| 16.1k| sumsq5_rows[1], sum5_rows[1], 1094| 16.1k| NULL, lpf, w, edges); 1095| | 1096| 16.1k| sgr_box35_row_h(sumsq3_rows[2], sum3_rows[2], 1097| 16.1k| sumsq5_rows[2], sum5_rows[2], 1098| 16.1k| left, src, w, edges); 1099| 16.1k| left++; 1100| 16.1k| src += PXSTRIDE(stride); ------------------ | | 53| 16.1k|#define PXSTRIDE(x) (x) ------------------ 1101| | 1102| 16.1k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1103| 16.1k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 16.1k|#define BITDEPTH_MAX 0xff ------------------ 1104| 16.1k| rotate(A3_ptrs, B3_ptrs, 4); 1105| | 1106| 16.1k| if (--h <= 0) ------------------ | Branch (1106:13): [True: 2.09k, False: 14.0k] ------------------ 1107| 2.09k| goto vert_1; 1108| | 1109| 14.0k| sgr_box35_row_h(sumsq3_ptrs[2], sum3_ptrs[2], 1110| 14.0k| sumsq5_rows[3], sum5_rows[3], 1111| 14.0k| left, src, w, edges); 1112| 14.0k| left++; 1113| 14.0k| src += PXSTRIDE(stride); ------------------ | | 53| 14.0k|#define PXSTRIDE(x) (x) ------------------ 1114| 14.0k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1115| 14.0k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 14.0k|#define BITDEPTH_MAX 0xff ------------------ 1116| 14.0k| rotate(A5_ptrs, B5_ptrs, 2); 1117| 14.0k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1118| 14.0k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 14.0k|#define BITDEPTH_MAX 0xff ------------------ 1119| 14.0k| rotate(A3_ptrs, B3_ptrs, 4); 1120| | 1121| 14.0k| if (--h <= 0) ------------------ | Branch (1121:13): [True: 1.68k, False: 12.3k] ------------------ 1122| 1.68k| goto vert_2; 1123| | 1124| | // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set 1125| | // one of them to point at the previously unused rows[4]. 1126| 12.3k| sumsq5_ptrs[3] = sumsq5_rows[4]; 1127| 12.3k| sum5_ptrs[3] = sum5_rows[4]; 1128| 12.3k| } else { 1129| 8.03k| sumsq5_ptrs[0] = sumsq5_rows[0]; 1130| 8.03k| sumsq5_ptrs[1] = sumsq5_rows[0]; 1131| 8.03k| sumsq5_ptrs[2] = sumsq5_rows[0]; 1132| 8.03k| sumsq5_ptrs[3] = sumsq5_rows[0]; 1133| 8.03k| sumsq5_ptrs[4] = sumsq5_rows[0]; 1134| 8.03k| sum5_ptrs[0] = sum5_rows[0]; 1135| 8.03k| sum5_ptrs[1] = sum5_rows[0]; 1136| 8.03k| sum5_ptrs[2] = sum5_rows[0]; 1137| 8.03k| sum5_ptrs[3] = sum5_rows[0]; 1138| 8.03k| sum5_ptrs[4] = sum5_rows[0]; 1139| | 1140| 8.03k| sumsq3_ptrs[0] = sumsq3_rows[0]; 1141| 8.03k| sumsq3_ptrs[1] = sumsq3_rows[0]; 1142| 8.03k| sumsq3_ptrs[2] = sumsq3_rows[0]; 1143| 8.03k| sum3_ptrs[0] = sum3_rows[0]; 1144| 8.03k| sum3_ptrs[1] = sum3_rows[0]; 1145| 8.03k| sum3_ptrs[2] = sum3_rows[0]; 1146| | 1147| 8.03k| sgr_box35_row_h(sumsq3_rows[0], sum3_rows[0], 1148| 8.03k| sumsq5_rows[0], sum5_rows[0], 1149| 8.03k| left, src, w, edges); 1150| 8.03k| left++; 1151| 8.03k| src += PXSTRIDE(stride); ------------------ | | 53| 8.03k|#define PXSTRIDE(x) (x) ------------------ 1152| | 1153| 8.03k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1154| 8.03k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 8.03k|#define BITDEPTH_MAX 0xff ------------------ 1155| 8.03k| rotate(A3_ptrs, B3_ptrs, 4); 1156| | 1157| 8.03k| if (--h <= 0) ------------------ | Branch (1157:13): [True: 1.36k, False: 6.67k] ------------------ 1158| 1.36k| goto vert_1; 1159| | 1160| 6.67k| sumsq5_ptrs[4] = sumsq5_rows[1]; 1161| 6.67k| sum5_ptrs[4] = sum5_rows[1]; 1162| | 1163| 6.67k| sumsq3_ptrs[2] = sumsq3_rows[1]; 1164| 6.67k| sum3_ptrs[2] = sum3_rows[1]; 1165| | 1166| 6.67k| sgr_box35_row_h(sumsq3_rows[1], sum3_rows[1], 1167| 6.67k| sumsq5_rows[1], sum5_rows[1], 1168| 6.67k| left, src, w, edges); 1169| 6.67k| left++; 1170| 6.67k| src += PXSTRIDE(stride); ------------------ | | 53| 6.67k|#define PXSTRIDE(x) (x) ------------------ 1171| | 1172| 6.67k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1173| 6.67k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 6.67k|#define BITDEPTH_MAX 0xff ------------------ 1174| 6.67k| rotate(A5_ptrs, B5_ptrs, 2); 1175| 6.67k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1176| 6.67k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 6.67k|#define BITDEPTH_MAX 0xff ------------------ 1177| 6.67k| rotate(A3_ptrs, B3_ptrs, 4); 1178| | 1179| 6.67k| if (--h <= 0) ------------------ | Branch (1179:13): [True: 193, False: 6.48k] ------------------ 1180| 193| goto vert_2; 1181| | 1182| 6.48k| sumsq5_ptrs[3] = sumsq5_rows[2]; 1183| 6.48k| sumsq5_ptrs[4] = sumsq5_rows[3]; 1184| 6.48k| sum5_ptrs[3] = sum5_rows[2]; 1185| 6.48k| sum5_ptrs[4] = sum5_rows[3]; 1186| | 1187| 6.48k| sumsq3_ptrs[2] = sumsq3_rows[2]; 1188| 6.48k| sum3_ptrs[2] = sum3_rows[2]; 1189| | 1190| 6.48k| sgr_box35_row_h(sumsq3_rows[2], sum3_rows[2], 1191| 6.48k| sumsq5_rows[2], sum5_rows[2], 1192| 6.48k| left, src, w, edges); 1193| 6.48k| left++; 1194| 6.48k| src += PXSTRIDE(stride); ------------------ | | 53| 6.48k|#define PXSTRIDE(x) (x) ------------------ 1195| | 1196| 6.48k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1197| 6.48k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 6.48k|#define BITDEPTH_MAX 0xff ------------------ 1198| 6.48k| rotate(A3_ptrs, B3_ptrs, 4); 1199| | 1200| 6.48k| if (--h <= 0) ------------------ | Branch (1200:13): [True: 242, False: 6.23k] ------------------ 1201| 242| goto odd; 1202| | 1203| 6.23k| sgr_box35_row_h(sumsq3_ptrs[2], sum3_ptrs[2], 1204| 6.23k| sumsq5_rows[3], sum5_rows[3], 1205| 6.23k| left, src, w, edges); 1206| 6.23k| left++; 1207| 6.23k| src += PXSTRIDE(stride); ------------------ | | 53| 6.23k|#define PXSTRIDE(x) (x) ------------------ 1208| | 1209| 6.23k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1210| 6.23k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 6.23k|#define BITDEPTH_MAX 0xff ------------------ 1211| 6.23k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1212| 6.23k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 6.23k|#define BITDEPTH_MAX 0xff ------------------ 1213| 6.23k| sgr_finish_mix(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs, 1214| 6.23k| w, 2, params->sgr.w0, params->sgr.w1 1215| 6.23k| HIGHBD_TAIL_SUFFIX); 1216| | 1217| 6.23k| if (--h <= 0) ------------------ | Branch (1217:13): [True: 89, False: 6.14k] ------------------ 1218| 89| goto vert_2; 1219| | 1220| | // ptrs are rotated by 2; both [3] and [4] now point at rows[0]; set 1221| | // one of them to point at the previously unused rows[4]. 1222| 6.14k| sumsq5_ptrs[3] = sumsq5_rows[4]; 1223| 6.14k| sum5_ptrs[3] = sum5_rows[4]; 1224| 6.14k| } 1225| | 1226| 511k| do { 1227| 511k| sgr_box35_row_h(sumsq3_ptrs[2], sum3_ptrs[2], 1228| 511k| sumsq5_ptrs[3], sum5_ptrs[3], 1229| 511k| left, src, w, edges); 1230| 511k| left++; 1231| 511k| src += PXSTRIDE(stride); ------------------ | | 53| 511k|#define PXSTRIDE(x) (x) ------------------ 1232| | 1233| 511k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1234| 511k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 511k|#define BITDEPTH_MAX 0xff ------------------ 1235| 511k| rotate(A3_ptrs, B3_ptrs, 4); 1236| | 1237| 511k| if (--h <= 0) ------------------ | Branch (1237:13): [True: 1.68k, False: 509k] ------------------ 1238| 1.68k| goto odd; 1239| | 1240| 509k| sgr_box35_row_h(sumsq3_ptrs[2], sum3_ptrs[2], 1241| 509k| sumsq5_ptrs[4], sum5_ptrs[4], 1242| 509k| left, src, w, edges); 1243| 509k| left++; 1244| 509k| src += PXSTRIDE(stride); ------------------ | | 53| 509k|#define PXSTRIDE(x) (x) ------------------ 1245| | 1246| 509k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1247| 509k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 509k|#define BITDEPTH_MAX 0xff ------------------ 1248| 509k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1249| 509k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 509k|#define BITDEPTH_MAX 0xff ------------------ 1250| 509k| sgr_finish_mix(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs, 1251| 509k| w, 2, params->sgr.w0, params->sgr.w1 1252| 509k| HIGHBD_TAIL_SUFFIX); 1253| 509k| } while (--h > 0); ------------------ | Branch (1253:14): [True: 492k, False: 16.8k] ------------------ 1254| | 1255| 16.8k| if (!(edges & LR_HAVE_BOTTOM)) ------------------ | Branch (1255:9): [True: 690, False: 16.1k] ------------------ 1256| 690| goto vert_2; 1257| | 1258| 16.1k| sgr_box35_row_h(sumsq3_ptrs[2], sum3_ptrs[2], 1259| 16.1k| sumsq5_ptrs[3], sum5_ptrs[3], 1260| 16.1k| NULL, lpf_bottom, w, edges); 1261| 16.1k| lpf_bottom += PXSTRIDE(stride); ------------------ | | 53| 16.1k|#define PXSTRIDE(x) (x) ------------------ 1262| 16.1k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1263| 16.1k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 16.1k|#define BITDEPTH_MAX 0xff ------------------ 1264| 16.1k| rotate(A3_ptrs, B3_ptrs, 4); 1265| | 1266| 16.1k| sgr_box35_row_h(sumsq3_ptrs[2], sum3_ptrs[2], 1267| 16.1k| sumsq5_ptrs[4], sum5_ptrs[4], 1268| 16.1k| NULL, lpf_bottom, w, edges); 1269| | 1270| 18.8k|output_2: 1271| 18.8k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1272| 18.8k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 18.8k|#define BITDEPTH_MAX 0xff ------------------ 1273| 18.8k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1274| 18.8k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 18.8k|#define BITDEPTH_MAX 0xff ------------------ 1275| 18.8k| sgr_finish_mix(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs, 1276| 18.8k| w, 2, params->sgr.w0, params->sgr.w1 1277| 18.8k| HIGHBD_TAIL_SUFFIX); 1278| 18.8k| return; 1279| | 1280| 2.65k|vert_2: 1281| | // Duplicate the last row twice more 1282| 2.65k| sumsq5_ptrs[3] = sumsq5_ptrs[2]; 1283| 2.65k| sumsq5_ptrs[4] = sumsq5_ptrs[2]; 1284| 2.65k| sum5_ptrs[3] = sum5_ptrs[2]; 1285| 2.65k| sum5_ptrs[4] = sum5_ptrs[2]; 1286| | 1287| 2.65k| sumsq3_ptrs[2] = sumsq3_ptrs[1]; 1288| 2.65k| sum3_ptrs[2] = sum3_ptrs[1]; 1289| 2.65k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1290| 2.65k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 2.65k|#define BITDEPTH_MAX 0xff ------------------ 1291| 2.65k| rotate(A3_ptrs, B3_ptrs, 4); 1292| | 1293| 2.65k| sumsq3_ptrs[2] = sumsq3_ptrs[1]; 1294| 2.65k| sum3_ptrs[2] = sum3_ptrs[1]; 1295| | 1296| 2.65k| goto output_2; 1297| | 1298| 1.92k|odd: 1299| | // Copy the last row as padding once 1300| 1.92k| sumsq5_ptrs[4] = sumsq5_ptrs[3]; 1301| 1.92k| sum5_ptrs[4] = sum5_ptrs[3]; 1302| | 1303| 1.92k| sumsq3_ptrs[2] = sumsq3_ptrs[1]; 1304| 1.92k| sum3_ptrs[2] = sum3_ptrs[1]; 1305| | 1306| 1.92k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1307| 1.92k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 1.92k|#define BITDEPTH_MAX 0xff ------------------ 1308| 1.92k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1309| 1.92k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 1.92k|#define BITDEPTH_MAX 0xff ------------------ 1310| 1.92k| sgr_finish_mix(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs, 1311| 1.92k| w, 2, params->sgr.w0, params->sgr.w1 1312| 1.92k| HIGHBD_TAIL_SUFFIX); 1313| | 1314| 5.37k|output_1: 1315| | // Duplicate the last row twice more 1316| 5.37k| sumsq5_ptrs[3] = sumsq5_ptrs[2]; 1317| 5.37k| sumsq5_ptrs[4] = sumsq5_ptrs[2]; 1318| 5.37k| sum5_ptrs[3] = sum5_ptrs[2]; 1319| 5.37k| sum5_ptrs[4] = sum5_ptrs[2]; 1320| | 1321| 5.37k| sumsq3_ptrs[2] = sumsq3_ptrs[1]; 1322| 5.37k| sum3_ptrs[2] = sum3_ptrs[1]; 1323| | 1324| 5.37k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1325| 5.37k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 5.37k|#define BITDEPTH_MAX 0xff ------------------ 1326| 5.37k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1327| 5.37k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 5.37k|#define BITDEPTH_MAX 0xff ------------------ 1328| 5.37k| rotate(A3_ptrs, B3_ptrs, 4); 1329| | // Output only one row 1330| 5.37k| sgr_finish_mix(&dst, stride, A5_ptrs, B5_ptrs, A3_ptrs, B3_ptrs, 1331| 5.37k| w, 1, params->sgr.w0, params->sgr.w1 1332| 5.37k| HIGHBD_TAIL_SUFFIX); 1333| 5.37k| return; 1334| | 1335| 3.45k|vert_1: 1336| | // Copy the last row as padding once 1337| 3.45k| sumsq5_ptrs[4] = sumsq5_ptrs[3]; 1338| 3.45k| sum5_ptrs[4] = sum5_ptrs[3]; 1339| | 1340| 3.45k| sumsq3_ptrs[2] = sumsq3_ptrs[1]; 1341| 3.45k| sum3_ptrs[2] = sum3_ptrs[1]; 1342| | 1343| 3.45k| sgr_box5_vert(sumsq5_ptrs, sum5_ptrs, A5_ptrs[1], B5_ptrs[1], 1344| 3.45k| w, params->sgr.s0, BITDEPTH_MAX); ------------------ | | 59| 3.45k|#define BITDEPTH_MAX 0xff ------------------ 1345| 3.45k| rotate(A5_ptrs, B5_ptrs, 2); 1346| 3.45k| sgr_box3_vert(sumsq3_ptrs, sum3_ptrs, A3_ptrs[3], B3_ptrs[3], 1347| 3.45k| w, params->sgr.s1, BITDEPTH_MAX); ------------------ | | 59| 3.45k|#define BITDEPTH_MAX 0xff ------------------ 1348| 3.45k| rotate(A3_ptrs, B3_ptrs, 4); 1349| | 1350| 3.45k| goto output_1; 1351| 1.92k|} looprestoration_tmpl.c:sgr_box35_row_h: 464| 1.13M|{ 465| 1.13M| sgr_box3_row_h(sumsq3, sum3, left, src, w, edges); 466| 1.13M| sgr_box5_row_h(sumsq5, sum5, left, src, w, edges); 467| 1.13M|} looprestoration_tmpl.c:sgr_finish_mix: 663| 540k|{ 664| 540k| ALIGN_STK_16(coef, tmp5, 2*FILTER_OUT_STRIDE,); ------------------ | | 100| 540k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 540k| line __attribute__((aligned(align))) | | ------------------ ------------------ 665| 540k| ALIGN_STK_16(coef, tmp3, 2*FILTER_OUT_STRIDE,); ------------------ | | 100| 540k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 540k| line __attribute__((aligned(align))) | | ------------------ ------------------ 666| | 667| 540k| sgr_finish_filter2(tmp5, *dst, stride, A5_ptrs, B5_ptrs, w, h); 668| 540k| sgr_finish_filter_row1(tmp3, *dst, A3_ptrs, B3_ptrs, w); 669| 540k| if (h > 1) ------------------ | Branch (669:9): [True: 536k, False: 4.69k] ------------------ 670| 536k| sgr_finish_filter_row1(tmp3 + FILTER_OUT_STRIDE, *dst + PXSTRIDE(stride), ------------------ | | 572| 536k|#define FILTER_OUT_STRIDE (384) ------------------ sgr_finish_filter_row1(tmp3 + FILTER_OUT_STRIDE, *dst + PXSTRIDE(stride), ------------------ | | 53| 536k|#define PXSTRIDE(x) (x) ------------------ 671| 536k| &A3_ptrs[1], &B3_ptrs[1], w); 672| 540k| sgr_weighted2(*dst, stride, tmp5, tmp3, w, h, w0, w1 HIGHBD_TAIL_SUFFIX); 673| 540k| *dst += h*PXSTRIDE(stride); ------------------ | | 53| 540k|#define PXSTRIDE(x) (x) ------------------ 674| 540k| rotate(A5_ptrs, B5_ptrs, 2); 675| 540k| rotate(A3_ptrs, B3_ptrs, 4); 676| 540k|} looprestoration_tmpl.c:sgr_weighted2: 616| 540k|{ 617| 1.61M| for (int j = 0; j < h; j++) { ------------------ | Branch (617:21): [True: 1.07M, False: 540k] ------------------ 618| 96.6M| for (int i = 0; i < w; i++) { ------------------ | Branch (618:25): [True: 95.6M, False: 1.07M] ------------------ 619| 95.6M| const int v = w0 * t1[i] + w1 * t2[i]; 620| 95.6M| dst[i] = iclip_pixel(dst[i] + ((v + (1 << 10)) >> 11)); ------------------ | | 49| 95.6M|#define iclip_pixel iclip_u8 ------------------ 621| 95.6M| } 622| 1.07M| dst += PXSTRIDE(dst_stride); ------------------ | | 53| 1.07M|#define PXSTRIDE(x) (x) ------------------ 623| 1.07M| t1 += FILTER_OUT_STRIDE; ------------------ | | 572| 1.07M|#define FILTER_OUT_STRIDE (384) ------------------ 624| 1.07M| t2 += FILTER_OUT_STRIDE; ------------------ | | 572| 1.07M|#define FILTER_OUT_STRIDE (384) ------------------ 625| 1.07M| } 626| 540k|} dav1d_loop_restoration_dsp_init_16bpc: 1367| 5.11k|{ 1368| 5.11k| c->wiener[0] = c->wiener[1] = wiener_c; 1369| 5.11k| c->sgr[0] = sgr_5x5_c; 1370| 5.11k| c->sgr[1] = sgr_3x3_c; 1371| 5.11k| c->sgr[2] = sgr_mix_c; 1372| | 1373| 5.11k|#if HAVE_ASM 1374| |#if ARCH_AARCH64 || ARCH_ARM 1375| | loop_restoration_dsp_init_arm(c, bpc); 1376| |#elif ARCH_LOONGARCH64 1377| | loop_restoration_dsp_init_loongarch(c, bpc); 1378| |#elif ARCH_PPC64LE 1379| | loop_restoration_dsp_init_ppc(c, bpc); 1380| |#elif ARCH_X86 1381| | loop_restoration_dsp_init_x86(c, bpc); 1382| 5.11k|#endif 1383| 5.11k|#endif 1384| 5.11k|} dav1d_lr_sbrow_8bpc: 171| 175k|{ 172| 175k| const int offset_y = 8 * !!sby; 173| 175k| const ptrdiff_t *const dst_stride = f->sr_cur.p.stride; 174| 175k| const int restore_planes = f->lf.restore_planes; 175| 175k| const int sb128 = f->seq_hdr->sb128; 176| 175k| const int not_last = sby + 1 < f->sbh; 177| | 178| 175k| if (restore_planes & LR_RESTORE_Y) { ------------------ | Branch (178:9): [True: 171k, False: 3.71k] ------------------ 179| 171k| const int h = f->sr_cur.p.p.h; 180| 171k| const int w = f->sr_cur.p.p.w; 181| 171k| const int next_row_y = (sby + 1) << (6 + sb128); 182| 171k| const int row_h = imin(next_row_y - 8 * not_last, h); 183| 171k| const int y_stripe = (sby << (6 + sb128)) - offset_y; 184| 171k| lr_sbrow(f, dst[0] - offset_y * PXSTRIDE(dst_stride[0]), y_stripe, w, ------------------ | | 53| 171k|#define PXSTRIDE(x) (x) ------------------ 185| 171k| h, row_h, 0); 186| 171k| } 187| 175k| if (restore_planes & (LR_RESTORE_U | LR_RESTORE_V)) { ------------------ | Branch (187:9): [True: 8.27k, False: 167k] ------------------ 188| 8.27k| const int ss_ver = f->sr_cur.p.p.layout == DAV1D_PIXEL_LAYOUT_I420; 189| 8.27k| const int ss_hor = f->sr_cur.p.p.layout != DAV1D_PIXEL_LAYOUT_I444; 190| 8.27k| const int h = (f->sr_cur.p.p.h + ss_ver) >> ss_ver; 191| 8.27k| const int w = (f->sr_cur.p.p.w + ss_hor) >> ss_hor; 192| 8.27k| const int next_row_y = (sby + 1) << ((6 - ss_ver) + sb128); 193| 8.27k| const int row_h = imin(next_row_y - (8 >> ss_ver) * not_last, h); 194| 8.27k| const int offset_uv = offset_y >> ss_ver; 195| 8.27k| const int y_stripe = (sby << ((6 - ss_ver) + sb128)) - offset_uv; 196| 8.27k| if (restore_planes & LR_RESTORE_U) ------------------ | Branch (196:13): [True: 6.17k, False: 2.10k] ------------------ 197| 6.17k| lr_sbrow(f, dst[1] - offset_uv * PXSTRIDE(dst_stride[1]), y_stripe, ------------------ | | 53| 6.17k|#define PXSTRIDE(x) (x) ------------------ 198| 6.17k| w, h, row_h, 1); 199| | 200| 8.27k| if (restore_planes & LR_RESTORE_V) ------------------ | Branch (200:13): [True: 6.17k, False: 2.10k] ------------------ 201| 6.17k| lr_sbrow(f, dst[2] - offset_uv * PXSTRIDE(dst_stride[1]), y_stripe, ------------------ | | 53| 6.17k|#define PXSTRIDE(x) (x) ------------------ 202| 6.17k| w, h, row_h, 2); 203| 8.27k| } 204| 175k|} lr_apply_tmpl.c:lr_sbrow: 109| 239k|{ 110| 239k| const int chroma = !!plane; 111| 239k| const int ss_ver = chroma & (f->sr_cur.p.p.layout == DAV1D_PIXEL_LAYOUT_I420); 112| 239k| const int ss_hor = chroma & (f->sr_cur.p.p.layout != DAV1D_PIXEL_LAYOUT_I444); 113| 239k| const ptrdiff_t p_stride = f->sr_cur.p.stride[chroma]; 114| | 115| 239k| const int unit_size_log2 = f->frame_hdr->restoration.unit_size[!!plane]; 116| 239k| const int unit_size = 1 << unit_size_log2; 117| 239k| const int half_unit_size = unit_size >> 1; 118| 239k| const int max_unit_size = unit_size + half_unit_size; 119| | 120| | // Y coordinate of the sbrow (y is 8 luma pixel rows above row_y) 121| 239k| const int row_y = y + ((8 >> ss_ver) * !!y); 122| | 123| | // FIXME This is an ugly hack to lookup the proper AV1Filter unit for 124| | // chroma planes. Question: For Multithreaded decoding, is it better 125| | // to store the chroma LR information with collocated Luma information? 126| | // In other words. For a chroma restoration unit locate at 128,128 and 127| | // with a 4:2:0 chroma subsampling, do we store the filter information at 128| | // the AV1Filter unit located at (128,128) or (256,256) 129| | // TODO Support chroma subsampling. 130| 239k| const int shift_hor = 7 - ss_hor; 131| | 132| | /* maximum sbrow height is 128 + 8 rows offset */ 133| 239k| ALIGN_STK_16(pixel, pre_lr_border, 2, [128 + 8][4]); ------------------ | | 100| 239k| ALIGN(type var[sz1d]sznd, ALIGN_16_VAL) | | ------------------ | | | | 86| 239k| line __attribute__((aligned(align))) | | ------------------ ------------------ 134| 239k| const Av1RestorationUnit *lr[2]; 135| | 136| 239k| enum LrEdgeFlags edges = (y > 0 ? LR_HAVE_TOP : 0) | LR_HAVE_RIGHT; ------------------ | Branch (136:31): [True: 220k, False: 18.9k] ------------------ 137| | 138| 239k| int aligned_unit_pos = row_y & ~(unit_size - 1); 139| 239k| if (aligned_unit_pos && aligned_unit_pos + half_unit_size > h) ------------------ | Branch (139:9): [True: 214k, False: 25.6k] | Branch (139:29): [True: 4.76k, False: 209k] ------------------ 140| 4.76k| aligned_unit_pos -= unit_size; 141| 239k| aligned_unit_pos <<= ss_ver; 142| 239k| const int sb_idx = (aligned_unit_pos >> 7) * f->sr_sb128w; 143| 239k| const int unit_idx = ((aligned_unit_pos >> 6) & 1) << 1; 144| 239k| lr[0] = &f->lf.lr_mask[sb_idx].lr[plane][unit_idx]; 145| 239k| int restore = lr[0]->type != DAV1D_RESTORATION_NONE; 146| 239k| int x = 0, bit = 0; 147| 239k| const int backup_h = row_h - y; 148| 289k| for (; x + max_unit_size <= w; p += unit_size, edges |= LR_HAVE_LEFT, bit ^= 1) { ------------------ | Branch (148:12): [True: 49.9k, False: 239k] ------------------ 149| 49.9k| const int next_x = x + unit_size; 150| 49.9k| const int next_u_idx = unit_idx + ((next_x >> (shift_hor - 1)) & 1); 151| 49.9k| lr[!bit] = 152| 49.9k| &f->lf.lr_mask[sb_idx + (next_x >> shift_hor)].lr[plane][next_u_idx]; 153| 49.9k| const int restore_next = lr[!bit]->type != DAV1D_RESTORATION_NONE; 154| 49.9k| if (restore_next) ------------------ | Branch (154:13): [True: 26.5k, False: 23.4k] ------------------ 155| 26.5k| backup4xU(pre_lr_border[bit], p + unit_size - 4, p_stride, backup_h); 156| 49.9k| if (restore) ------------------ | Branch (156:13): [True: 26.6k, False: 23.2k] ------------------ 157| 26.6k| lr_stripe(f, p, pre_lr_border[!bit], x, y, plane, unit_size, row_h, 158| 26.6k| lr[bit], edges); 159| 49.9k| x = next_x; 160| 49.9k| restore = restore_next; 161| 49.9k| } 162| 239k| if (restore) { ------------------ | Branch (162:9): [True: 25.2k, False: 214k] ------------------ 163| 25.2k| edges &= ~LR_HAVE_RIGHT; 164| 25.2k| const int unit_w = w - x; 165| 25.2k| lr_stripe(f, p, pre_lr_border[!bit], x, y, plane, unit_w, row_h, lr[bit], edges); 166| 25.2k| } 167| 239k|} lr_apply_tmpl.c:backup4xU: 102| 26.5k|{ 103| 2.49M| for (; u > 0; u--, dst++, src += PXSTRIDE(src_stride)) ------------------ | | 53| 2.47M|#define PXSTRIDE(x) (x) ------------------ | Branch (103:12): [True: 2.47M, False: 26.5k] ------------------ 104| 2.47M| pixel_copy(dst, src, 4); ------------------ | | 47| 2.47M|#define pixel_copy memcpy ------------------ 105| 26.5k|} lr_apply_tmpl.c:lr_stripe: 40| 51.9k|{ 41| 51.9k| const Dav1dDSPContext *const dsp = f->dsp; 42| 51.9k| const int chroma = !!plane; 43| 51.9k| const int ss_ver = chroma & (f->sr_cur.p.p.layout == DAV1D_PIXEL_LAYOUT_I420); 44| 51.9k| const ptrdiff_t stride = f->sr_cur.p.stride[chroma]; 45| 51.9k| const int sby = (y + (y ? 8 << ss_ver : 0)) >> (6 - ss_ver + f->seq_hdr->sb128); ------------------ | Branch (45:27): [True: 33.6k, False: 18.2k] ------------------ 46| 51.9k| const int have_tt = f->c->n_tc > 1; 47| 51.9k| const pixel *lpf = f->lf.lr_lpf_line[plane] + 48| 51.9k| have_tt * (sby * (4 << f->seq_hdr->sb128) - 4) * PXSTRIDE(stride) + x; ------------------ | | 53| 51.9k|#define PXSTRIDE(x) (x) ------------------ 49| | 50| | // The first stripe of the frame is shorter by 8 luma pixel rows. 51| 51.9k| int stripe_h = imin((64 - 8 * !y) >> ss_ver, row_h - y); 52| | 53| 51.9k| looprestorationfilter_fn lr_fn; 54| 51.9k| LooprestorationParams params; 55| 51.9k| if (lr->type == DAV1D_RESTORATION_WIENER) { ------------------ | Branch (55:9): [True: 21.7k, False: 30.1k] ------------------ 56| 21.7k| int16_t (*const filter)[8] = params.filter; 57| 21.7k| filter[0][0] = filter[0][6] = lr->filter_h[0]; 58| 21.7k| filter[0][1] = filter[0][5] = lr->filter_h[1]; 59| 21.7k| filter[0][2] = filter[0][4] = lr->filter_h[2]; 60| 21.7k| filter[0][3] = -(filter[0][0] + filter[0][1] + filter[0][2]) * 2; 61| |#if BITDEPTH != 8 62| | /* For 8-bit SIMD it's beneficial to handle the +128 separately 63| | * in order to avoid overflows. */ 64| | filter[0][3] += 128; 65| |#endif 66| | 67| 21.7k| filter[1][0] = filter[1][6] = lr->filter_v[0]; 68| 21.7k| filter[1][1] = filter[1][5] = lr->filter_v[1]; 69| 21.7k| filter[1][2] = filter[1][4] = lr->filter_v[2]; 70| 21.7k| filter[1][3] = 128 - (filter[1][0] + filter[1][1] + filter[1][2]) * 2; 71| | 72| 21.7k| lr_fn = dsp->lr.wiener[!(filter[0][0] | filter[1][0])]; 73| 30.1k| } else { 74| 30.1k| assert(lr->type >= DAV1D_RESTORATION_SGRPROJ); ------------------ | Branch (74:9): [True: 30.1k, False: 18.4E] ------------------ 75| 30.1k| const int sgr_idx = lr->type - DAV1D_RESTORATION_SGRPROJ; 76| 30.1k| const uint16_t *const sgr_params = dav1d_sgr_params[sgr_idx]; 77| 30.1k| params.sgr.s0 = sgr_params[0]; 78| 30.1k| params.sgr.s1 = sgr_params[1]; 79| 30.1k| params.sgr.w0 = lr->sgr_weights[0]; 80| 30.1k| params.sgr.w1 = 128 - (lr->sgr_weights[0] + lr->sgr_weights[1]); 81| | 82| 30.1k| lr_fn = dsp->lr.sgr[!!sgr_params[0] + !!sgr_params[1] * 2 - 1]; 83| 30.1k| } 84| | 85| 92.8k| while (y + stripe_h <= row_h) { ------------------ | Branch (85:12): [True: 92.8k, False: 0] ------------------ 86| | // Change the HAVE_BOTTOM bit in edges to (sby + 1 != f->sbh || y + stripe_h != row_h) 87| 92.8k| edges ^= (-(sby + 1 != f->sbh || y + stripe_h != row_h) ^ edges) & LR_HAVE_BOTTOM; ------------------ | Branch (87:21): [True: 55.5k, False: 37.3k] | Branch (87:42): [True: 19.1k, False: 18.1k] ------------------ 88| 92.8k| lr_fn(p, stride, left, lpf, unit_w, stripe_h, ¶ms, edges HIGHBD_CALL_SUFFIX); 89| | 90| 92.8k| left += stripe_h; 91| 92.8k| y += stripe_h; 92| 92.8k| p += stripe_h * PXSTRIDE(stride); ------------------ | | 53| 92.8k|#define PXSTRIDE(x) (x) ------------------ 93| 92.8k| edges |= LR_HAVE_TOP; 94| 92.8k| stripe_h = imin(64 >> ss_ver, row_h - y); 95| 92.8k| if (stripe_h == 0) break; ------------------ | Branch (95:13): [True: 51.9k, False: 40.8k] ------------------ 96| 40.8k| lpf += 4 * PXSTRIDE(stride); ------------------ | | 53| 40.8k|#define PXSTRIDE(x) (x) ------------------ 97| 40.8k| } 98| 51.9k|} dav1d_lr_sbrow_16bpc: 171| 42.4k|{ 172| 42.4k| const int offset_y = 8 * !!sby; 173| 42.4k| const ptrdiff_t *const dst_stride = f->sr_cur.p.stride; 174| 42.4k| const int restore_planes = f->lf.restore_planes; 175| 42.4k| const int sb128 = f->seq_hdr->sb128; 176| 42.4k| const int not_last = sby + 1 < f->sbh; 177| | 178| 42.4k| if (restore_planes & LR_RESTORE_Y) { ------------------ | Branch (178:9): [True: 41.4k, False: 991] ------------------ 179| 41.4k| const int h = f->sr_cur.p.p.h; 180| 41.4k| const int w = f->sr_cur.p.p.w; 181| 41.4k| const int next_row_y = (sby + 1) << (6 + sb128); 182| 41.4k| const int row_h = imin(next_row_y - 8 * not_last, h); 183| 41.4k| const int y_stripe = (sby << (6 + sb128)) - offset_y; 184| 41.4k| lr_sbrow(f, dst[0] - offset_y * PXSTRIDE(dst_stride[0]), y_stripe, w, 185| 41.4k| h, row_h, 0); 186| 41.4k| } 187| 42.4k| if (restore_planes & (LR_RESTORE_U | LR_RESTORE_V)) { ------------------ | Branch (187:9): [True: 8.85k, False: 33.5k] ------------------ 188| 8.85k| const int ss_ver = f->sr_cur.p.p.layout == DAV1D_PIXEL_LAYOUT_I420; 189| 8.85k| const int ss_hor = f->sr_cur.p.p.layout != DAV1D_PIXEL_LAYOUT_I444; 190| 8.85k| const int h = (f->sr_cur.p.p.h + ss_ver) >> ss_ver; 191| 8.85k| const int w = (f->sr_cur.p.p.w + ss_hor) >> ss_hor; 192| 8.85k| const int next_row_y = (sby + 1) << ((6 - ss_ver) + sb128); 193| 8.85k| const int row_h = imin(next_row_y - (8 >> ss_ver) * not_last, h); 194| 8.85k| const int offset_uv = offset_y >> ss_ver; 195| 8.85k| const int y_stripe = (sby << ((6 - ss_ver) + sb128)) - offset_uv; 196| 8.85k| if (restore_planes & LR_RESTORE_U) ------------------ | Branch (196:13): [True: 6.56k, False: 2.28k] ------------------ 197| 6.56k| lr_sbrow(f, dst[1] - offset_uv * PXSTRIDE(dst_stride[1]), y_stripe, 198| 6.56k| w, h, row_h, 1); 199| | 200| 8.85k| if (restore_planes & LR_RESTORE_V) ------------------ | Branch (200:13): [True: 7.81k, False: 1.03k] ------------------ 201| 7.81k| lr_sbrow(f, dst[2] - offset_uv * PXSTRIDE(dst_stride[1]), y_stripe, 202| 7.81k| w, h, row_h, 2); 203| 8.85k| } 204| 42.4k|} dav1d_mc_dsp_init_8bpc: 960| 3.46k|COLD void bitfn(dav1d_mc_dsp_init)(Dav1dMCDSPContext *const c) { 961| 3.46k|#define init_mc_fns(type, name) do { \ 962| 3.46k| c->mc [type] = put_##name##_c; \ 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ 964| 3.46k| c->mct [type] = prep_##name##_c; \ 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ 966| 3.46k|} while (0) 967| | 968| 3.46k| init_mc_fns(FILTER_2D_8TAP_REGULAR, 8tap_regular); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 969| 3.46k| init_mc_fns(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 970| 3.46k| init_mc_fns(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 971| 3.46k| init_mc_fns(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 972| 3.46k| init_mc_fns(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 973| 3.46k| init_mc_fns(FILTER_2D_8TAP_SHARP, 8tap_sharp); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 974| 3.46k| init_mc_fns(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 975| 3.46k| init_mc_fns(FILTER_2D_8TAP_SMOOTH, 8tap_smooth); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 976| 3.46k| init_mc_fns(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 977| 3.46k| init_mc_fns(FILTER_2D_BILINEAR, bilin); ------------------ | | 961| 3.46k|#define init_mc_fns(type, name) do { \ | | 962| 3.46k| c->mc [type] = put_##name##_c; \ | | 963| 3.46k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 3.46k| c->mct [type] = prep_##name##_c; \ | | 965| 3.46k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 3.46k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 3.46k] | | ------------------ ------------------ 978| | 979| 3.46k| c->avg = avg_c; 980| 3.46k| c->w_avg = w_avg_c; 981| 3.46k| c->mask = mask_c; 982| 3.46k| c->blend = blend_c; 983| 3.46k| c->blend_v = blend_v_c; 984| 3.46k| c->blend_h = blend_h_c; 985| 3.46k| c->w_mask[0] = w_mask_444_c; 986| 3.46k| c->w_mask[1] = w_mask_422_c; 987| 3.46k| c->w_mask[2] = w_mask_420_c; 988| 3.46k| c->warp8x8 = warp_affine_8x8_c; 989| 3.46k| c->warp8x8t = warp_affine_8x8t_c; 990| 3.46k| c->emu_edge = emu_edge_c; 991| 3.46k| c->resize = resize_c; 992| | 993| 3.46k|#if HAVE_ASM 994| |#if ARCH_AARCH64 || ARCH_ARM 995| | mc_dsp_init_arm(c); 996| |#elif ARCH_LOONGARCH64 997| | mc_dsp_init_loongarch(c); 998| |#elif ARCH_PPC64LE 999| | mc_dsp_init_ppc(c); 1000| |#elif ARCH_RISCV 1001| | mc_dsp_init_riscv(c); 1002| |#elif ARCH_X86 1003| | mc_dsp_init_x86(c); 1004| 3.46k|#endif 1005| 3.46k|#endif 1006| 3.46k|} dav1d_mc_dsp_init_16bpc: 960| 5.11k|COLD void bitfn(dav1d_mc_dsp_init)(Dav1dMCDSPContext *const c) { 961| 5.11k|#define init_mc_fns(type, name) do { \ 962| 5.11k| c->mc [type] = put_##name##_c; \ 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ 964| 5.11k| c->mct [type] = prep_##name##_c; \ 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ 966| 5.11k|} while (0) 967| | 968| 5.11k| init_mc_fns(FILTER_2D_8TAP_REGULAR, 8tap_regular); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 969| 5.11k| init_mc_fns(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 970| 5.11k| init_mc_fns(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 971| 5.11k| init_mc_fns(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 972| 5.11k| init_mc_fns(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 973| 5.11k| init_mc_fns(FILTER_2D_8TAP_SHARP, 8tap_sharp); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 974| 5.11k| init_mc_fns(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 975| 5.11k| init_mc_fns(FILTER_2D_8TAP_SMOOTH, 8tap_smooth); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 976| 5.11k| init_mc_fns(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 977| 5.11k| init_mc_fns(FILTER_2D_BILINEAR, bilin); ------------------ | | 961| 5.11k|#define init_mc_fns(type, name) do { \ | | 962| 5.11k| c->mc [type] = put_##name##_c; \ | | 963| 5.11k| c->mc_scaled [type] = put_##name##_scaled_c; \ | | 964| 5.11k| c->mct [type] = prep_##name##_c; \ | | 965| 5.11k| c->mct_scaled[type] = prep_##name##_scaled_c; \ | | 966| 5.11k|} while (0) | | ------------------ | | | Branch (966:10): [Folded, False: 5.11k] | | ------------------ ------------------ 978| | 979| 5.11k| c->avg = avg_c; 980| 5.11k| c->w_avg = w_avg_c; 981| 5.11k| c->mask = mask_c; 982| 5.11k| c->blend = blend_c; 983| 5.11k| c->blend_v = blend_v_c; 984| 5.11k| c->blend_h = blend_h_c; 985| 5.11k| c->w_mask[0] = w_mask_444_c; 986| 5.11k| c->w_mask[1] = w_mask_422_c; 987| 5.11k| c->w_mask[2] = w_mask_420_c; 988| 5.11k| c->warp8x8 = warp_affine_8x8_c; 989| 5.11k| c->warp8x8t = warp_affine_8x8t_c; 990| 5.11k| c->emu_edge = emu_edge_c; 991| 5.11k| c->resize = resize_c; 992| | 993| 5.11k|#if HAVE_ASM 994| |#if ARCH_AARCH64 || ARCH_ARM 995| | mc_dsp_init_arm(c); 996| |#elif ARCH_LOONGARCH64 997| | mc_dsp_init_loongarch(c); 998| |#elif ARCH_PPC64LE 999| | mc_dsp_init_ppc(c); 1000| |#elif ARCH_RISCV 1001| | mc_dsp_init_riscv(c); 1002| |#elif ARCH_X86 1003| | mc_dsp_init_x86(c); 1004| 5.11k|#endif 1005| 5.11k|#endif 1006| 5.11k|} dav1d_mem_pool_push: 224| 1.60M|void dav1d_mem_pool_push(Dav1dMemPool *const pool, void *const ptr) { 225| 1.60M| pthread_mutex_lock(&pool->lock); 226| 1.60M| Dav1dMemPoolBuffer *const buf = (Dav1dMemPoolBuffer*)((uintptr_t)ptr - 64); 227| 1.60M| const int ref_cnt = --pool->ref_cnt; 228| 1.60M| if (!pool->end) { ------------------ | Branch (228:9): [True: 1.57M, False: 20.5k] ------------------ 229| 1.57M| buf->next = pool->buf; 230| 1.57M| pool->buf = buf; 231| 1.57M| pthread_mutex_unlock(&pool->lock); 232| 1.57M| assert(ref_cnt > 0); ------------------ | Branch (232:9): [True: 1.57M, False: 18.4E] ------------------ 233| 1.57M| } else { 234| 20.5k| pthread_mutex_unlock(&pool->lock); 235| 20.5k| dav1d_free_aligned(buf); ------------------ | | 136| 20.5k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 236| 20.6k| if (!ref_cnt) mem_pool_destroy(pool); ------------------ | Branch (236:13): [True: 20.6k, False: 18.4E] ------------------ 237| 20.5k| } 238| 1.60M|} dav1d_mem_pool_pop: 240| 1.60M|void *dav1d_mem_pool_pop(Dav1dMemPool *const pool, const size_t size) { 241| 1.60M| pthread_mutex_lock(&pool->lock); 242| 1.60M| Dav1dMemPoolBuffer *buf = pool->buf; 243| 1.60M| pool->ref_cnt++; 244| | 245| 1.60M| if (buf) { ------------------ | Branch (245:9): [True: 1.50M, False: 97.6k] ------------------ 246| 1.50M| pool->buf = buf->next; 247| 1.50M| pthread_mutex_unlock(&pool->lock); 248| 1.50M| if (buf->size != size) { ------------------ | Branch (248:13): [True: 20.7k, False: 1.48M] ------------------ 249| | /* Reallocate if the size has changed */ 250| 20.7k| dav1d_free_aligned(buf); ------------------ | | 136| 20.7k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 251| 20.7k| goto alloc; 252| 20.7k| } 253| |#if TRACK_HEAP_ALLOCATIONS 254| | dav1d_track_reuse(pool->type); 255| |#endif 256| 1.50M| } else { 257| 97.6k| pthread_mutex_unlock(&pool->lock); 258| 118k|alloc: 259| 118k| buf = dav1d_alloc_aligned(pool->type, size + 64, 64); ------------------ | | 134| 118k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 260| 118k| if (!buf) { ------------------ | Branch (260:13): [True: 0, False: 118k] ------------------ 261| 0| pthread_mutex_lock(&pool->lock); 262| 0| const int ref_cnt = --pool->ref_cnt; 263| 0| pthread_mutex_unlock(&pool->lock); 264| 0| if (!ref_cnt) mem_pool_destroy(pool); ------------------ | Branch (264:17): [True: 0, False: 0] ------------------ 265| 0| return NULL; 266| 0| } 267| 118k| buf->size = size; 268| 118k| } 269| | 270| 1.60M| return (void*)((uintptr_t)buf + 64); 271| 1.60M|} dav1d_mem_pool_init: 275| 65.9k|{ 276| 65.9k| Dav1dMemPool *const pool = dav1d_malloc(ALLOC_COMMON_CTX, ------------------ | | 132| 65.9k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 277| 65.9k| sizeof(Dav1dMemPool)); 278| 65.9k| if (pool) { ------------------ | Branch (278:9): [True: 65.9k, False: 0] ------------------ 279| 65.9k| if (!pthread_mutex_init(&pool->lock, NULL)) { ------------------ | Branch (279:13): [True: 65.9k, False: 0] ------------------ 280| 65.9k| pool->buf = NULL; 281| 65.9k| pool->ref_cnt = 1; 282| 65.9k| pool->end = 0; 283| |#if TRACK_HEAP_ALLOCATIONS 284| | pool->type = type; 285| |#endif 286| 65.9k| *ppool = pool; 287| 65.9k| return 0; 288| 65.9k| } 289| 0| dav1d_free(pool); ------------------ | | 135| 0|#define dav1d_free(ptr) free(ptr) ------------------ 290| 0| } 291| 0| *ppool = NULL; 292| 0| return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 293| 65.9k|} dav1d_mem_pool_end: 295| 65.9k|COLD void dav1d_mem_pool_end(Dav1dMemPool *const pool) { 296| 65.9k| if (pool) { ------------------ | Branch (296:9): [True: 65.9k, False: 0] ------------------ 297| 65.9k| pthread_mutex_lock(&pool->lock); 298| 65.9k| Dav1dMemPoolBuffer *buf = pool->buf; 299| 65.9k| const int ref_cnt = --pool->ref_cnt; 300| 65.9k| pool->buf = NULL; 301| 65.9k| pool->end = 1; 302| 65.9k| pthread_mutex_unlock(&pool->lock); 303| | 304| 142k| while (buf) { ------------------ | Branch (304:16): [True: 76.9k, False: 65.9k] ------------------ 305| 76.9k| void *const ptr = buf; 306| 76.9k| buf = buf->next; 307| 76.9k| dav1d_free_aligned(ptr); ------------------ | | 136| 76.9k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 308| 76.9k| } 309| 65.9k| if (!ref_cnt) mem_pool_destroy(pool); ------------------ | Branch (309:13): [True: 45.2k, False: 20.6k] ------------------ 310| 65.9k| } 311| 65.9k|} mem.c:mem_pool_destroy: 219| 65.9k|static COLD void mem_pool_destroy(Dav1dMemPool *const pool) { 220| 65.9k| pthread_mutex_destroy(&pool->lock); 221| 65.9k| dav1d_free(pool); ------------------ | | 135| 65.9k|#define dav1d_free(ptr) free(ptr) ------------------ 222| 65.9k|} lib.c:dav1d_alloc_aligned_internal: 89| 28.2k|static inline void *dav1d_alloc_aligned_internal(const size_t sz, const size_t align) { 90| 28.2k| assert(!(align & (align - 1))); ------------------ | Branch (90:5): [True: 28.2k, False: 0] ------------------ 91| |#ifdef _WIN32 92| | return _aligned_malloc(sz, align); 93| |#elif HAVE_POSIX_MEMALIGN 94| 28.2k| void *ptr; 95| 28.2k| if (posix_memalign(&ptr, align, sz)) return NULL; ------------------ | Branch (95:9): [True: 0, False: 28.2k] ------------------ 96| 28.2k| return ptr; 97| |#elif HAVE_MEMALIGN 98| | return memalign(align, sz); 99| |#elif HAVE_ALIGNED_ALLOC 100| | // The C11 standard specifies that the size parameter 101| | // must be an integral multiple of alignment. 102| | return aligned_alloc(align, ROUND_UP(sz, align)); 103| |#else 104| | void *const buf = malloc(sz + align + sizeof(void *)); 105| | if (!buf) return NULL; 106| | 107| | void *const ptr = (void *)(((uintptr_t)buf + sizeof(void *) + align - 1) & ~(align - 1)); 108| | ((void **)ptr)[-1] = buf; 109| | return ptr; 110| |#endif 111| 28.2k|} lib.c:dav1d_free_aligned_internal: 113| 367k|static inline void dav1d_free_aligned_internal(void *ptr) { 114| |#ifdef _WIN32 115| | _aligned_free(ptr); 116| |#elif HAVE_POSIX_MEMALIGN || HAVE_MEMALIGN || HAVE_ALIGNED_ALLOC 117| | free(ptr); 118| |#else 119| | if (ptr) free(((void **)ptr)[-1]); 120| |#endif 121| 367k|} lib.c:dav1d_freep_aligned: 144| 9.41k|static inline void dav1d_freep_aligned(void *ptr) { 145| 9.41k| void **mem = (void **) ptr; 146| 9.41k| if (*mem) { ------------------ | Branch (146:9): [True: 9.41k, False: 0] ------------------ 147| 9.41k| dav1d_free_aligned(*mem); ------------------ | | 136| 9.41k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 148| | *mem = NULL; 149| 9.41k| } 150| 9.41k|} mem.c:dav1d_free_aligned_internal: 113| 118k|static inline void dav1d_free_aligned_internal(void *ptr) { 114| |#ifdef _WIN32 115| | _aligned_free(ptr); 116| |#elif HAVE_POSIX_MEMALIGN || HAVE_MEMALIGN || HAVE_ALIGNED_ALLOC 117| | free(ptr); 118| |#else 119| | if (ptr) free(((void **)ptr)[-1]); 120| |#endif 121| 118k|} mem.c:dav1d_alloc_aligned_internal: 89| 118k|static inline void *dav1d_alloc_aligned_internal(const size_t sz, const size_t align) { 90| 118k| assert(!(align & (align - 1))); ------------------ | Branch (90:5): [True: 118k, False: 0] ------------------ 91| |#ifdef _WIN32 92| | return _aligned_malloc(sz, align); 93| |#elif HAVE_POSIX_MEMALIGN 94| 118k| void *ptr; 95| 118k| if (posix_memalign(&ptr, align, sz)) return NULL; ------------------ | Branch (95:9): [True: 0, False: 118k] ------------------ 96| 118k| return ptr; 97| |#elif HAVE_MEMALIGN 98| | return memalign(align, sz); 99| |#elif HAVE_ALIGNED_ALLOC 100| | // The C11 standard specifies that the size parameter 101| | // must be an integral multiple of alignment. 102| | return aligned_alloc(align, ROUND_UP(sz, align)); 103| |#else 104| | void *const buf = malloc(sz + align + sizeof(void *)); 105| | if (!buf) return NULL; 106| | 107| | void *const ptr = (void *)(((uintptr_t)buf + sizeof(void *) + align - 1) & ~(align - 1)); 108| | ((void **)ptr)[-1] = buf; 109| | return ptr; 110| |#endif 111| 118k|} ref.c:dav1d_alloc_aligned_internal: 89| 412k|static inline void *dav1d_alloc_aligned_internal(const size_t sz, const size_t align) { 90| 412k| assert(!(align & (align - 1))); ------------------ | Branch (90:5): [True: 412k, False: 0] ------------------ 91| |#ifdef _WIN32 92| | return _aligned_malloc(sz, align); 93| |#elif HAVE_POSIX_MEMALIGN 94| 412k| void *ptr; 95| 412k| if (posix_memalign(&ptr, align, sz)) return NULL; ------------------ | Branch (95:9): [True: 0, False: 412k] ------------------ 96| 412k| return ptr; 97| |#elif HAVE_MEMALIGN 98| | return memalign(align, sz); 99| |#elif HAVE_ALIGNED_ALLOC 100| | // The C11 standard specifies that the size parameter 101| | // must be an integral multiple of alignment. 102| | return aligned_alloc(align, ROUND_UP(sz, align)); 103| |#else 104| | void *const buf = malloc(sz + align + sizeof(void *)); 105| | if (!buf) return NULL; 106| | 107| | void *const ptr = (void *)(((uintptr_t)buf + sizeof(void *) + align - 1) & ~(align - 1)); 108| | ((void **)ptr)[-1] = buf; 109| | return ptr; 110| |#endif 111| 412k|} ref.c:dav1d_free_aligned_internal: 113| 412k|static inline void dav1d_free_aligned_internal(void *ptr) { 114| |#ifdef _WIN32 115| | _aligned_free(ptr); 116| |#elif HAVE_POSIX_MEMALIGN || HAVE_MEMALIGN || HAVE_ALIGNED_ALLOC 117| | free(ptr); 118| |#else 119| | if (ptr) free(((void **)ptr)[-1]); 120| |#endif 121| 412k|} refmvs.c:dav1d_free_aligned_internal: 113| 30.5k|static inline void dav1d_free_aligned_internal(void *ptr) { 114| |#ifdef _WIN32 115| | _aligned_free(ptr); 116| |#elif HAVE_POSIX_MEMALIGN || HAVE_MEMALIGN || HAVE_ALIGNED_ALLOC 117| | free(ptr); 118| |#else 119| | if (ptr) free(((void **)ptr)[-1]); 120| |#endif 121| 30.5k|} refmvs.c:dav1d_alloc_aligned_internal: 89| 30.5k|static inline void *dav1d_alloc_aligned_internal(const size_t sz, const size_t align) { 90| 30.5k| assert(!(align & (align - 1))); ------------------ | Branch (90:5): [True: 30.5k, False: 0] ------------------ 91| |#ifdef _WIN32 92| | return _aligned_malloc(sz, align); 93| |#elif HAVE_POSIX_MEMALIGN 94| 30.5k| void *ptr; 95| 30.5k| if (posix_memalign(&ptr, align, sz)) return NULL; ------------------ | Branch (95:9): [True: 0, False: 30.5k] ------------------ 96| 30.5k| return ptr; 97| |#elif HAVE_MEMALIGN 98| | return memalign(align, sz); 99| |#elif HAVE_ALIGNED_ALLOC 100| | // The C11 standard specifies that the size parameter 101| | // must be an integral multiple of alignment. 102| | return aligned_alloc(align, ROUND_UP(sz, align)); 103| |#else 104| | void *const buf = malloc(sz + align + sizeof(void *)); 105| | if (!buf) return NULL; 106| | 107| | void *const ptr = (void *)(((uintptr_t)buf + sizeof(void *) + align - 1) & ~(align - 1)); 108| | ((void **)ptr)[-1] = buf; 109| | return ptr; 110| |#endif 111| 30.5k|} decode.c:dav1d_free_aligned_internal: 113| 198k|static inline void dav1d_free_aligned_internal(void *ptr) { 114| |#ifdef _WIN32 115| | _aligned_free(ptr); 116| |#elif HAVE_POSIX_MEMALIGN || HAVE_MEMALIGN || HAVE_ALIGNED_ALLOC 117| | free(ptr); 118| |#else 119| | if (ptr) free(((void **)ptr)[-1]); 120| |#endif 121| 198k|} decode.c:dav1d_alloc_aligned_internal: 89| 194k|static inline void *dav1d_alloc_aligned_internal(const size_t sz, const size_t align) { 90| 194k| assert(!(align & (align - 1))); ------------------ | Branch (90:5): [True: 194k, False: 18.4E] ------------------ 91| |#ifdef _WIN32 92| | return _aligned_malloc(sz, align); 93| |#elif HAVE_POSIX_MEMALIGN 94| 194k| void *ptr; 95| 194k| if (posix_memalign(&ptr, align, sz)) return NULL; ------------------ | Branch (95:9): [True: 0, False: 194k] ------------------ 96| 194k| return ptr; 97| |#elif HAVE_MEMALIGN 98| | return memalign(align, sz); 99| |#elif HAVE_ALIGNED_ALLOC 100| | // The C11 standard specifies that the size parameter 101| | // must be an integral multiple of alignment. 102| | return aligned_alloc(align, ROUND_UP(sz, align)); 103| |#else 104| | void *const buf = malloc(sz + align + sizeof(void *)); 105| | if (!buf) return NULL; 106| | 107| | void *const ptr = (void *)(((uintptr_t)buf + sizeof(void *) + align - 1) & ~(align - 1)); 108| | ((void **)ptr)[-1] = buf; 109| | return ptr; 110| |#endif 111| 194k|} decode.c:dav1d_freep_aligned: 144| 3.89k|static inline void dav1d_freep_aligned(void *ptr) { 145| 3.89k| void **mem = (void **) ptr; 146| 3.89k| if (*mem) { ------------------ | Branch (146:9): [True: 3.89k, False: 0] ------------------ 147| 3.89k| dav1d_free_aligned(*mem); ------------------ | | 136| 3.89k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 148| | *mem = NULL; 149| 3.89k| } 150| 3.89k|} dav1d_msac_decode_subexp: 62| 231k|{ 63| 231k| assert(n >> k == 8); ------------------ | Branch (63:5): [True: 231k, False: 18.4E] ------------------ 64| | 65| 231k| unsigned a = 0; 66| 231k| if (dav1d_msac_decode_bool_equi(s)) { ------------------ | | 53| 231k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (66:9): [True: 148k, False: 83.2k] ------------------ 67| 148k| if (dav1d_msac_decode_bool_equi(s)) ------------------ | | 53| 148k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (67:13): [True: 104k, False: 43.4k] ------------------ 68| 104k| k += dav1d_msac_decode_bool_equi(s) + 1; ------------------ | | 53| 104k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ 69| 148k| a = 1 << k; 70| 148k| } 71| 231k| const unsigned v = dav1d_msac_decode_bools(s, k) + a; 72| 231k| return ref * 2 <= n ? inv_recenter(ref, v) : ------------------ | Branch (72:12): [True: 134k, False: 96.6k] ------------------ 73| 231k| n - 1 - inv_recenter(n - 1 - ref, v); 74| 231k|} dav1d_msac_init: 206| 315k|{ 207| 315k| s->buf_pos = data; 208| 315k| s->buf_end = data + sz; 209| 315k| s->dif = 0; 210| 315k| s->rng = 0x8000; 211| 315k| s->cnt = -15; 212| 315k| s->allow_update_cdf = !disable_cdf_update_flag; 213| 315k| ctx_refill(s); 214| | 215| 315k|#if ARCH_X86_64 && HAVE_ASM 216| 315k| s->symbol_adapt16 = dav1d_msac_decode_symbol_adapt_c; 217| | 218| 315k| msac_init_x86(s); 219| 315k|#endif 220| 315k|} msac.c:ctx_refill: 41| 315k|static inline void ctx_refill(MsacContext *const s) { 42| 315k| const uint8_t *buf_pos = s->buf_pos; 43| 315k| const uint8_t *buf_end = s->buf_end; 44| 315k| int c = EC_WIN_SIZE - s->cnt - 24; ------------------ | | 39| 315k|#define EC_WIN_SIZE (sizeof(ec_win) << 3) ------------------ 45| 315k| ec_win dif = s->dif; 46| 973k| do { 47| 973k| if (buf_pos >= buf_end) { ------------------ | Branch (47:13): [True: 287k, False: 685k] ------------------ 48| | // set remaining bits to 1; 49| 287k| dif |= ~(~(ec_win)0xff << c); 50| 287k| break; 51| 287k| } 52| 685k| dif |= (ec_win)(*buf_pos++ ^ 0xff) << c; 53| 685k| c -= 8; 54| 685k| } while (c >= 0); ------------------ | Branch (54:14): [True: 657k, False: 27.6k] ------------------ 55| 315k| s->dif = dif; 56| 315k| s->cnt = EC_WIN_SIZE - c - 24; ------------------ | | 39| 315k|#define EC_WIN_SIZE (sizeof(ec_win) << 3) ------------------ 57| 315k| s->buf_pos = buf_pos; 58| 315k|} decode.c:dav1d_msac_decode_bools: 94| 1.45M|static inline unsigned dav1d_msac_decode_bools(MsacContext *const s, unsigned n) { 95| 1.45M| unsigned v = 0; 96| 3.10M| while (n--) ------------------ | Branch (96:12): [True: 1.64M, False: 1.45M] ------------------ 97| 1.64M| v = (v << 1) | dav1d_msac_decode_bool_equi(s); ------------------ | | 53| 1.64M|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ 98| 1.45M| return v; 99| 1.45M|} decode.c:dav1d_msac_decode_uniform: 101| 199k|static inline int dav1d_msac_decode_uniform(MsacContext *const s, const unsigned n) { 102| 199k| assert(n > 0); ------------------ | Branch (102:5): [True: 199k, False: 18.4E] ------------------ 103| 199k| const int l = ulog2(n) + 1; 104| 199k| assert(l > 1); ------------------ | Branch (104:5): [True: 199k, False: 5] ------------------ 105| 199k| const unsigned m = (1 << l) - n; 106| 199k| const unsigned v = dav1d_msac_decode_bools(s, l - 1); 107| 199k| return v < m ? v : (v << 1) - m + dav1d_msac_decode_bool_equi(s); ------------------ | | 53| 56.9k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (107:12): [True: 142k, False: 56.9k] ------------------ 108| 199k|} msac.c:dav1d_msac_decode_bools: 94| 231k|static inline unsigned dav1d_msac_decode_bools(MsacContext *const s, unsigned n) { 95| 231k| unsigned v = 0; 96| 1.06M| while (n--) ------------------ | Branch (96:12): [True: 828k, False: 231k] ------------------ 97| 828k| v = (v << 1) | dav1d_msac_decode_bool_equi(s); ------------------ | | 53| 828k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ 98| 231k| return v; 99| 231k|} recon_tmpl.c:dav1d_msac_decode_bools: 94| 13.7M|static inline unsigned dav1d_msac_decode_bools(MsacContext *const s, unsigned n) { 95| 13.7M| unsigned v = 0; 96| 44.2M| while (n--) ------------------ | Branch (96:12): [True: 30.4M, False: 13.7M] ------------------ 97| 30.4M| v = (v << 1) | dav1d_msac_decode_bool_equi(s); ------------------ | | 53| 30.4M|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ 98| 13.7M| return v; 99| 13.7M|} dav1d_parse_sequence_header: 304| 15.3k|{ 305| 15.3k| validate_input_or_ret(out != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 15.3k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 15.3k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 306| 15.3k| validate_input_or_ret(ptr != NULL, DAV1D_ERR(EINVAL)); ------------------ | | 52| 15.3k| if (!(x)) { \ | | ------------------ | | | Branch (52:9): [True: 0, False: 15.3k] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 307| 15.3k| validate_input_or_ret(sz > 0 && sz <= SIZE_MAX / 2, DAV1D_ERR(EINVAL)); ------------------ | | 52| 30.6k| if (!(x)) { \ | | ------------------ | | | Branch (52:11): [True: 15.3k, False: 0] | | | Branch (52:11): [True: 15.3k, False: 0] | | ------------------ | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | ------------------ | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | ------------------ | | 54| 0| #x, __func__); \ | | 55| 0| debug_abort(); \ | | ------------------ | | | | 39| 0|#define debug_abort abort | | ------------------ | | 56| 0| return r; \ | | 57| 0| } ------------------ 308| | 309| 15.3k| GetBits gb; 310| 15.3k| dav1d_init_get_bits(&gb, ptr, sz); 311| 15.3k| int res = DAV1D_ERR(ENOENT); ------------------ | | 58| 15.3k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 312| | 313| 33.4k| do { 314| 33.4k| dav1d_get_bit(&gb); // obu_forbidden_bit 315| 33.4k| const enum Dav1dObuType type = dav1d_get_bits(&gb, 4); 316| 33.4k| const int has_extension = dav1d_get_bit(&gb); 317| 33.4k| const int has_length_field = dav1d_get_bit(&gb); 318| 33.4k| dav1d_get_bits(&gb, 1 + 8 * has_extension); // ignore 319| | 320| 33.4k| const uint8_t *obu_end = gb.ptr_end; 321| 33.4k| if (has_length_field) { ------------------ | Branch (321:13): [True: 22.0k, False: 11.3k] ------------------ 322| 22.0k| const size_t len = dav1d_get_uleb128(&gb); 323| 22.0k| if (len > (size_t)(obu_end - gb.ptr)) return DAV1D_ERR(EINVAL); ------------------ | | 58| 537|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (323:17): [True: 537, False: 21.5k] ------------------ 324| 21.5k| obu_end = gb.ptr + len; 325| 21.5k| } 326| | 327| 32.9k| if (type == DAV1D_OBU_SEQ_HDR) { ------------------ | Branch (327:13): [True: 12.1k, False: 20.7k] ------------------ 328| 12.1k| if ((res = parse_seq_hdr(out, &gb, 0)) < 0) return res; ------------------ | Branch (328:17): [True: 2.42k, False: 9.74k] ------------------ 329| 9.74k| if (gb.ptr > obu_end) return DAV1D_ERR(EINVAL); ------------------ | | 58| 223|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (329:17): [True: 223, False: 9.52k] ------------------ 330| 9.52k| dav1d_bytealign_get_bits(&gb); 331| 9.52k| } 332| | 333| 30.2k| if (gb.error) return DAV1D_ERR(EINVAL); ------------------ | | 58| 381|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (333:13): [True: 381, False: 29.8k] ------------------ 334| 30.2k| assert(gb.state == 0 && gb.bits_left == 0); ------------------ | Branch (334:9): [True: 29.8k, False: 0] | Branch (334:9): [True: 29.8k, False: 0] ------------------ 335| 29.8k| gb.ptr = obu_end; 336| 29.8k| } while (gb.ptr < gb.ptr_end); ------------------ | Branch (336:14): [True: 18.1k, False: 11.7k] ------------------ 337| | 338| 11.7k| return res; 339| 15.3k|} dav1d_parse_obus: 1169| 534k|ptrdiff_t dav1d_parse_obus(Dav1dContext *const c, Dav1dData *const in) { 1170| 534k| GetBits gb; 1171| 534k| int res; 1172| | 1173| 534k| dav1d_init_get_bits(&gb, in->data, in->sz); 1174| | 1175| | // obu header 1176| 534k| const int obu_forbidden_bit = dav1d_get_bit(&gb); 1177| 534k| if (c->strict_std_compliance && obu_forbidden_bit) goto error; ------------------ | Branch (1177:9): [True: 0, False: 534k] | Branch (1177:37): [True: 0, False: 0] ------------------ 1178| 534k| const enum Dav1dObuType type = dav1d_get_bits(&gb, 4); 1179| 534k| const int has_extension = dav1d_get_bit(&gb); 1180| 534k| const int has_length_field = dav1d_get_bit(&gb); 1181| 534k| dav1d_get_bit(&gb); // reserved 1182| | 1183| 534k| int temporal_id = 0, spatial_id = 0; 1184| 534k| if (has_extension) { ------------------ | Branch (1184:9): [True: 15.1k, False: 519k] ------------------ 1185| 15.1k| temporal_id = dav1d_get_bits(&gb, 3); 1186| 15.1k| spatial_id = dav1d_get_bits(&gb, 2); 1187| 15.1k| dav1d_get_bits(&gb, 3); // reserved 1188| 15.1k| } 1189| | 1190| 534k| if (has_length_field) { ------------------ | Branch (1190:9): [True: 159k, False: 374k] ------------------ 1191| 159k| const size_t len = dav1d_get_uleb128(&gb); 1192| 159k| if (len > (size_t)(gb.ptr_end - gb.ptr)) goto error; ------------------ | Branch (1192:13): [True: 5.83k, False: 154k] ------------------ 1193| 154k| gb.ptr_end = gb.ptr + len; 1194| 154k| } 1195| 528k| if (gb.error) goto error; ------------------ | Branch (1195:9): [True: 483, False: 528k] ------------------ 1196| | 1197| | // We must have read a whole number of bytes at this point (1 byte 1198| | // for the header and whole bytes at a time when reading the 1199| | // leb128 length field). 1200| 528k| assert(gb.bits_left == 0); ------------------ | Branch (1200:5): [True: 528k, False: 0] ------------------ 1201| | 1202| | // skip obu not belonging to the selected temporal/spatial layer 1203| 528k| if (type != DAV1D_OBU_SEQ_HDR && type != DAV1D_OBU_TD && ------------------ | Branch (1203:9): [True: 496k, False: 31.8k] | Branch (1203:38): [True: 434k, False: 61.4k] ------------------ 1204| 434k| has_extension && c->operating_point_idc != 0) ------------------ | Branch (1204:9): [True: 12.4k, False: 422k] | Branch (1204:26): [True: 5.64k, False: 6.85k] ------------------ 1205| 5.64k| { 1206| 5.64k| const int in_temporal_layer = (c->operating_point_idc >> temporal_id) & 1; 1207| 5.64k| const int in_spatial_layer = (c->operating_point_idc >> (spatial_id + 8)) & 1; 1208| 5.64k| if (!in_temporal_layer || !in_spatial_layer) ------------------ | Branch (1208:13): [True: 514, False: 5.12k] | Branch (1208:35): [True: 406, False: 4.72k] ------------------ 1209| 920| return gb.ptr_end - gb.ptr_start; 1210| 5.64k| } 1211| | 1212| 527k| switch (type) { 1213| 31.8k| case DAV1D_OBU_SEQ_HDR: { ------------------ | Branch (1213:5): [True: 31.8k, False: 495k] ------------------ 1214| 31.8k| Dav1dRef *ref = dav1d_ref_create_using_pool(c->seq_hdr_pool, 1215| 31.8k| sizeof(Dav1dSequenceHeader)); 1216| 31.8k| if (!ref) return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (1216:13): [True: 0, False: 31.8k] ------------------ 1217| 31.8k| Dav1dSequenceHeader *seq_hdr = ref->data; 1218| 31.8k| if ((res = parse_seq_hdr(seq_hdr, &gb, c->strict_std_compliance)) < 0) { ------------------ | Branch (1218:13): [True: 4.11k, False: 27.7k] ------------------ 1219| 4.11k| dav1d_log(c, "Error parsing sequence header\n"); ------------------ | | 44| 4.11k|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 4.11k] | | ------------------ ------------------ 1220| 4.11k| dav1d_ref_dec(&ref); 1221| 4.11k| goto error; 1222| 4.11k| } 1223| | 1224| 27.7k| const int op_idx = 1225| 27.7k| c->operating_point < seq_hdr->num_operating_points ? c->operating_point : 0; ------------------ | Branch (1225:13): [True: 27.7k, False: 0] ------------------ 1226| 27.7k| c->operating_point_idc = seq_hdr->operating_points[op_idx].idc; 1227| 27.7k| const unsigned spatial_mask = c->operating_point_idc >> 8; 1228| 27.7k| c->max_spatial_id = spatial_mask ? ulog2(spatial_mask) : 0; ------------------ | Branch (1228:29): [True: 7.45k, False: 20.3k] ------------------ 1229| | 1230| | // If we have read a sequence header which is different from 1231| | // the old one, this is a new video sequence and can't use any 1232| | // previous state. Free that state. 1233| | 1234| 27.7k| if (!c->seq_hdr) { ------------------ | Branch (1234:13): [True: 9.12k, False: 18.6k] ------------------ 1235| 9.12k| c->frame_hdr = NULL; 1236| 9.12k| c->frame_flags |= PICTURE_FLAG_NEW_SEQUENCE; 1237| | // see 7.5, operating_parameter_info is allowed to change in 1238| | // sequence headers of a single sequence 1239| 18.6k| } else if (memcmp(seq_hdr, c->seq_hdr, offsetof(Dav1dSequenceHeader, operating_parameter_info))) { ------------------ | Branch (1239:20): [True: 7.59k, False: 11.0k] ------------------ 1240| 7.59k| c->frame_hdr = NULL; 1241| 7.59k| c->mastering_display = NULL; 1242| 7.59k| c->content_light = NULL; 1243| 7.59k| dav1d_ref_dec(&c->mastering_display_ref); 1244| 7.59k| dav1d_ref_dec(&c->content_light_ref); 1245| 68.3k| for (int i = 0; i < 8; i++) { ------------------ | Branch (1245:29): [True: 60.7k, False: 7.59k] ------------------ 1246| 60.7k| if (c->refs[i].p.p.frame_hdr) ------------------ | Branch (1246:21): [True: 40.7k, False: 20.0k] ------------------ 1247| 40.7k| dav1d_thread_picture_unref(&c->refs[i].p); 1248| 60.7k| dav1d_ref_dec(&c->refs[i].segmap); 1249| 60.7k| dav1d_ref_dec(&c->refs[i].refmvs); 1250| 60.7k| dav1d_cdf_thread_unref(&c->cdf[i]); 1251| 60.7k| } 1252| 7.59k| c->frame_flags |= PICTURE_FLAG_NEW_SEQUENCE; 1253| | // If operating_parameter_info changed, signal it 1254| 11.0k| } else if (memcmp(seq_hdr->operating_parameter_info, c->seq_hdr->operating_parameter_info, ------------------ | Branch (1254:20): [True: 122, False: 10.9k] ------------------ 1255| 11.0k| sizeof(seq_hdr->operating_parameter_info))) 1256| 122| { 1257| 122| c->frame_flags |= PICTURE_FLAG_NEW_OP_PARAMS_INFO; 1258| 122| } 1259| 27.7k| dav1d_ref_dec(&c->seq_hdr_ref); 1260| 27.7k| c->seq_hdr_ref = ref; 1261| 27.7k| c->seq_hdr = seq_hdr; 1262| 27.7k| break; 1263| 31.8k| } 1264| 1.40k| case DAV1D_OBU_REDUNDANT_FRAME_HDR: ------------------ | Branch (1264:5): [True: 1.40k, False: 525k] ------------------ 1265| 1.40k| if (c->frame_hdr) break; ------------------ | Branch (1265:13): [True: 509, False: 896] ------------------ 1266| | // fall-through 1267| 381k| case DAV1D_OBU_FRAME: ------------------ | Branch (1267:5): [True: 380k, False: 146k] ------------------ 1268| 417k| case DAV1D_OBU_FRAME_HDR: ------------------ | Branch (1268:5): [True: 36.3k, False: 490k] ------------------ 1269| 417k| if (!c->seq_hdr) goto error; ------------------ | Branch (1269:13): [True: 207, False: 417k] ------------------ 1270| 417k| if (!c->frame_hdr_ref) { ------------------ | Branch (1270:13): [True: 360k, False: 57.1k] ------------------ 1271| 360k| c->frame_hdr_ref = dav1d_ref_create_using_pool(c->frame_hdr_pool, 1272| 360k| sizeof(Dav1dFrameHeader)); 1273| 360k| if (!c->frame_hdr_ref) return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (1273:17): [True: 0, False: 360k] ------------------ 1274| 360k| } 1275| 417k|#ifndef NDEBUG 1276| | // ensure that the reference is writable 1277| 417k| assert(dav1d_ref_is_writable(c->frame_hdr_ref)); ------------------ | Branch (1277:9): [True: 417k, False: 0] ------------------ 1278| 417k|#endif 1279| 417k| c->frame_hdr = c->frame_hdr_ref->data; 1280| 417k| memset(c->frame_hdr, 0, sizeof(*c->frame_hdr)); 1281| 417k| c->frame_hdr->temporal_id = temporal_id; 1282| 417k| c->frame_hdr->spatial_id = spatial_id; 1283| 417k| if ((res = parse_frame_hdr(c, &gb)) < 0) { ------------------ | Branch (1283:13): [True: 7.38k, False: 410k] ------------------ 1284| 7.38k| c->frame_hdr = NULL; 1285| 7.38k| goto error; 1286| 7.38k| } 1287| 413k| for (int n = 0; n < c->n_tile_data; n++) ------------------ | Branch (1287:25): [True: 3.03k, False: 410k] ------------------ 1288| 3.03k| dav1d_data_unref_internal(&c->tile[n].data); 1289| 410k| c->n_tile_data = 0; 1290| 410k| c->n_tiles = 0; 1291| 410k| if (type != DAV1D_OBU_FRAME) { ------------------ | Branch (1291:13): [True: 35.9k, False: 374k] ------------------ 1292| | // This is actually a frame header OBU so read the 1293| | // trailing bit and check for overrun. 1294| 35.9k| if (check_trailing_bits(&gb, c->strict_std_compliance) < 0) { ------------------ | Branch (1294:17): [True: 2.34k, False: 33.6k] ------------------ 1295| 2.34k| c->frame_hdr = NULL; 1296| 2.34k| goto error; 1297| 2.34k| } 1298| 35.9k| } 1299| | 1300| 407k| if (c->frame_size_limit && (int64_t)c->frame_hdr->width[1] * ------------------ | Branch (1300:13): [True: 407k, False: 0] | Branch (1300:36): [True: 420, False: 407k] ------------------ 1301| 407k| c->frame_hdr->height > c->frame_size_limit) 1302| 420| { 1303| 420| dav1d_log(c, "Frame size %dx%d exceeds limit %u\n", c->frame_hdr->width[1], ------------------ | | 44| 420|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 420] | | ------------------ ------------------ 1304| 420| c->frame_hdr->height, c->frame_size_limit); 1305| 420| c->frame_hdr = NULL; 1306| 420| return DAV1D_ERR(ERANGE); ------------------ | | 58| 420|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 1307| 420| } 1308| | 1309| 407k| if (type != DAV1D_OBU_FRAME) ------------------ | Branch (1309:13): [True: 33.6k, False: 373k] ------------------ 1310| 33.6k| break; 1311| | // OBU_FRAMEs shouldn't be signaled with show_existing_frame 1312| 373k| if (c->frame_hdr->show_existing_frame) { ------------------ | Branch (1312:13): [True: 548, False: 373k] ------------------ 1313| 548| c->frame_hdr = NULL; 1314| 548| goto error; 1315| 548| } 1316| | 1317| | // This is the frame header at the start of a frame OBU. 1318| | // There's no trailing bit at the end to skip, but we do need 1319| | // to align to the next byte. 1320| 373k| dav1d_bytealign_get_bits(&gb); 1321| | // fall-through 1322| 375k| case DAV1D_OBU_TILE_GRP: { ------------------ | Branch (1322:5): [True: 2.35k, False: 524k] ------------------ 1323| 375k| if (!c->frame_hdr) goto error; ------------------ | Branch (1323:13): [True: 1.48k, False: 374k] ------------------ 1324| 374k| if (c->n_tile_data_alloc < c->n_tile_data + 1) { ------------------ | Branch (1324:13): [True: 8.74k, False: 365k] ------------------ 1325| 8.74k| if ((c->n_tile_data + 1) > INT_MAX / (int)sizeof(*c->tile)) goto error; ------------------ | Branch (1325:17): [True: 0, False: 8.74k] ------------------ 1326| 8.74k| struct Dav1dTileGroup *tile = dav1d_realloc(ALLOC_TILE, c->tile, ------------------ | | 133| 8.74k|#define dav1d_realloc(type, ptr, sz) realloc(ptr, sz) ------------------ 1327| 8.74k| (c->n_tile_data + 1) * sizeof(*c->tile)); 1328| 8.74k| if (!tile) goto error; ------------------ | Branch (1328:17): [True: 0, False: 8.74k] ------------------ 1329| 8.74k| c->tile = tile; 1330| 8.74k| memset(c->tile + c->n_tile_data, 0, sizeof(*c->tile)); 1331| 8.74k| c->n_tile_data_alloc = c->n_tile_data + 1; 1332| 8.74k| } 1333| 374k| parse_tile_hdr(c, &gb); 1334| | // Align to the next byte boundary and check for overrun. 1335| 374k| dav1d_bytealign_get_bits(&gb); 1336| 374k| if (gb.error) goto error; ------------------ | Branch (1336:13): [True: 12.7k, False: 361k] ------------------ 1337| | 1338| 361k| dav1d_data_ref(&c->tile[c->n_tile_data].data, in); 1339| 361k| c->tile[c->n_tile_data].data.data = gb.ptr; 1340| 361k| c->tile[c->n_tile_data].data.sz = (size_t)(gb.ptr_end - gb.ptr); 1341| | // ensure tile groups are in order and sane, see 6.10.1 1342| 361k| if (c->tile[c->n_tile_data].start > c->tile[c->n_tile_data].end || ------------------ | Branch (1342:13): [True: 946, False: 360k] ------------------ 1343| 360k| c->tile[c->n_tile_data].start != c->n_tiles) ------------------ | Branch (1343:13): [True: 625, False: 359k] ------------------ 1344| 1.57k| { 1345| 3.30k| for (int i = 0; i <= c->n_tile_data; i++) ------------------ | Branch (1345:29): [True: 1.73k, False: 1.57k] ------------------ 1346| 1.73k| dav1d_data_unref_internal(&c->tile[i].data); 1347| 1.57k| c->n_tile_data = 0; 1348| 1.57k| c->n_tiles = 0; 1349| 1.57k| goto error; 1350| 1.57k| } 1351| 359k| c->n_tiles += 1 + c->tile[c->n_tile_data].end - 1352| 359k| c->tile[c->n_tile_data].start; 1353| 359k| c->n_tile_data++; 1354| 359k| break; 1355| 361k| } 1356| 5.84k| case DAV1D_OBU_METADATA: { ------------------ | Branch (1356:5): [True: 5.84k, False: 521k] ------------------ 1357| 5.84k|#define DEBUG_OBU_METADATA 0 1358| |#if DEBUG_OBU_METADATA 1359| | const uint8_t *const init_ptr = gb.ptr; 1360| |#endif 1361| | // obu metadta type field 1362| 5.84k| const enum ObuMetaType meta_type = dav1d_get_uleb128(&gb); 1363| 5.84k| if (gb.error) goto error; ------------------ | Branch (1363:13): [True: 284, False: 5.55k] ------------------ 1364| | 1365| 5.55k| switch (meta_type) { 1366| 653| case OBU_META_HDR_CLL: { ------------------ | Branch (1366:9): [True: 653, False: 4.90k] ------------------ 1367| 653| Dav1dRef *ref = dav1d_ref_create(ALLOC_OBU_META, ------------------ | | 49| 653|#define dav1d_ref_create(type, size) dav1d_ref_create(size) ------------------ 1368| 653| sizeof(Dav1dContentLightLevel)); 1369| 653| if (!ref) return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (1369:17): [True: 0, False: 653] ------------------ 1370| 653| Dav1dContentLightLevel *const content_light = ref->data; 1371| | 1372| 653| content_light->max_content_light_level = dav1d_get_bits(&gb, 16); 1373| |#if DEBUG_OBU_METADATA 1374| | printf("CLLOBU: max-content-light-level: %d [off=%td]\n", 1375| | content_light->max_content_light_level, 1376| | (gb.ptr - init_ptr) * 8 - gb.bits_left); 1377| |#endif 1378| 653| content_light->max_frame_average_light_level = dav1d_get_bits(&gb, 16); 1379| |#if DEBUG_OBU_METADATA 1380| | printf("CLLOBU: max-frame-average-light-level: %d [off=%td]\n", 1381| | content_light->max_frame_average_light_level, 1382| | (gb.ptr - init_ptr) * 8 - gb.bits_left); 1383| |#endif 1384| | 1385| 653| if (check_trailing_bits(&gb, c->strict_std_compliance) < 0) { ------------------ | Branch (1385:17): [True: 207, False: 446] ------------------ 1386| 207| dav1d_ref_dec(&ref); 1387| 207| goto error; 1388| 207| } 1389| | 1390| 446| dav1d_ref_dec(&c->content_light_ref); 1391| 446| c->content_light = content_light; 1392| 446| c->content_light_ref = ref; 1393| 446| break; 1394| 653| } 1395| 237| case OBU_META_HDR_MDCV: { ------------------ | Branch (1395:9): [True: 237, False: 5.31k] ------------------ 1396| 237| Dav1dRef *ref = dav1d_ref_create(ALLOC_OBU_META, ------------------ | | 49| 237|#define dav1d_ref_create(type, size) dav1d_ref_create(size) ------------------ 1397| 237| sizeof(Dav1dMasteringDisplay)); 1398| 237| if (!ref) return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (1398:17): [True: 0, False: 237] ------------------ 1399| 237| Dav1dMasteringDisplay *const mastering_display = ref->data; 1400| | 1401| 948| for (int i = 0; i < 3; i++) { ------------------ | Branch (1401:29): [True: 711, False: 237] ------------------ 1402| 711| mastering_display->primaries[i][0] = dav1d_get_bits(&gb, 16); 1403| 711| mastering_display->primaries[i][1] = dav1d_get_bits(&gb, 16); 1404| |#if DEBUG_OBU_METADATA 1405| | printf("MDCVOBU: primaries[%d]: (%d, %d) [off=%td]\n", i, 1406| | mastering_display->primaries[i][0], 1407| | mastering_display->primaries[i][1], 1408| | (gb.ptr - init_ptr) * 8 - gb.bits_left); 1409| |#endif 1410| 711| } 1411| 237| mastering_display->white_point[0] = dav1d_get_bits(&gb, 16); 1412| |#if DEBUG_OBU_METADATA 1413| | printf("MDCVOBU: white-point-x: %d [off=%td]\n", 1414| | mastering_display->white_point[0], 1415| | (gb.ptr - init_ptr) * 8 - gb.bits_left); 1416| |#endif 1417| 237| mastering_display->white_point[1] = dav1d_get_bits(&gb, 16); 1418| |#if DEBUG_OBU_METADATA 1419| | printf("MDCVOBU: white-point-y: %d [off=%td]\n", 1420| | mastering_display->white_point[1], 1421| | (gb.ptr - init_ptr) * 8 - gb.bits_left); 1422| |#endif 1423| 237| mastering_display->max_luminance = dav1d_get_bits(&gb, 32); 1424| |#if DEBUG_OBU_METADATA 1425| | printf("MDCVOBU: max-luminance: %d [off=%td]\n", 1426| | mastering_display->max_luminance, 1427| | (gb.ptr - init_ptr) * 8 - gb.bits_left); 1428| |#endif 1429| 237| mastering_display->min_luminance = dav1d_get_bits(&gb, 32); 1430| |#if DEBUG_OBU_METADATA 1431| | printf("MDCVOBU: min-luminance: %d [off=%td]\n", 1432| | mastering_display->min_luminance, 1433| | (gb.ptr - init_ptr) * 8 - gb.bits_left); 1434| |#endif 1435| 237| if (check_trailing_bits(&gb, c->strict_std_compliance) < 0) { ------------------ | Branch (1435:17): [True: 153, False: 84] ------------------ 1436| 153| dav1d_ref_dec(&ref); 1437| 153| goto error; 1438| 153| } 1439| | 1440| 84| dav1d_ref_dec(&c->mastering_display_ref); 1441| 84| c->mastering_display = mastering_display; 1442| 84| c->mastering_display_ref = ref; 1443| 84| break; 1444| 237| } 1445| 4.24k| case OBU_META_ITUT_T35: { ------------------ | Branch (1445:9): [True: 4.24k, False: 1.30k] ------------------ 1446| 4.24k| ptrdiff_t payload_size = gb.ptr_end - gb.ptr; 1447| | // Don't take into account all the trailing bits for payload_size 1448| 4.53k| while (payload_size > 0 && !gb.ptr[payload_size - 1]) ------------------ | Branch (1448:20): [True: 3.87k, False: 662] | Branch (1448:40): [True: 286, False: 3.58k] ------------------ 1449| 286| payload_size--; // trailing_zero_bit x 8 1450| 4.24k| payload_size--; // trailing_one_bit + trailing_zero_bit x 7 1451| | 1452| 4.24k| int country_code_extension_byte = 0; 1453| 4.24k| const int country_code = dav1d_get_bits(&gb, 8); 1454| 4.24k| payload_size--; 1455| 4.24k| if (country_code == 0xFF) { ------------------ | Branch (1455:17): [True: 1.86k, False: 2.38k] ------------------ 1456| 1.86k| country_code_extension_byte = dav1d_get_bits(&gb, 8); 1457| 1.86k| payload_size--; 1458| 1.86k| } 1459| | 1460| 4.24k| if (payload_size <= 0 || gb.ptr[payload_size] != 0x80) { ------------------ | Branch (1460:17): [True: 726, False: 3.52k] | Branch (1460:38): [True: 1.44k, False: 2.08k] ------------------ 1461| 2.16k| dav1d_log(c, "Malformed ITU-T T.35 metadata message format\n"); ------------------ | | 44| 2.16k|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 2.16k] | | ------------------ ------------------ 1462| 2.16k| break; 1463| 2.16k| } 1464| | 1465| 2.08k| if ((c->n_itut_t35 + 1) > INT_MAX / (int)sizeof(*c->itut_t35)) goto error; ------------------ | Branch (1465:17): [True: 0, False: 2.08k] ------------------ 1466| 2.08k| struct Dav1dITUTT35 *itut_t35 = dav1d_realloc(ALLOC_OBU_META, c->itut_t35, ------------------ | | 133| 2.08k|#define dav1d_realloc(type, ptr, sz) realloc(ptr, sz) ------------------ 1467| 2.08k| (c->n_itut_t35 + 1) * sizeof(*c->itut_t35)); 1468| 2.08k| if (!itut_t35) goto error; ------------------ | Branch (1468:17): [True: 0, False: 2.08k] ------------------ 1469| 2.08k| c->itut_t35 = itut_t35; 1470| 2.08k| memset(c->itut_t35 + c->n_itut_t35, 0, sizeof(*c->itut_t35)); 1471| | 1472| 2.08k| struct itut_t35_ctx_context *itut_t35_ctx; 1473| 2.08k| if (!c->n_itut_t35) { ------------------ | Branch (1473:17): [True: 972, False: 1.10k] ------------------ 1474| 972| assert(!c->itut_t35_ref); ------------------ | Branch (1474:17): [True: 972, False: 0] ------------------ 1475| 972| itut_t35_ctx = dav1d_malloc(ALLOC_OBU_META, sizeof(struct itut_t35_ctx_context)); ------------------ | | 132| 972|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 1476| 972| if (!itut_t35_ctx) goto error; ------------------ | Branch (1476:21): [True: 0, False: 972] ------------------ 1477| 972| c->itut_t35_ref = dav1d_ref_init(&itut_t35_ctx->ref, c->itut_t35, 1478| 972| dav1d_picture_free_itut_t35, itut_t35_ctx, 0); 1479| 1.10k| } else { 1480| 1.10k| assert(c->itut_t35_ref && atomic_load(&c->itut_t35_ref->ref_cnt) == 1); ------------------ | Branch (1480:17): [True: 1.10k, False: 0] | Branch (1480:17): [True: 1.10k, False: 0] ------------------ 1481| 1.10k| itut_t35_ctx = c->itut_t35_ref->user_data; 1482| 1.10k| c->itut_t35_ref->const_data = (uint8_t *)c->itut_t35; 1483| 1.10k| } 1484| 2.08k| itut_t35_ctx->itut_t35 = c->itut_t35; 1485| 2.08k| itut_t35_ctx->n_itut_t35 = c->n_itut_t35 + 1; 1486| | 1487| 2.08k| Dav1dITUTT35 *const itut_t35_metadata = &c->itut_t35[c->n_itut_t35]; 1488| 2.08k| itut_t35_metadata->payload = dav1d_malloc(ALLOC_OBU_META, payload_size); ------------------ | | 132| 2.08k|#define dav1d_malloc(type, sz) malloc(sz) ------------------ 1489| 2.08k| if (!itut_t35_metadata->payload) goto error; ------------------ | Branch (1489:17): [True: 0, False: 2.08k] ------------------ 1490| | 1491| 2.08k| itut_t35_metadata->country_code = country_code; 1492| 2.08k| itut_t35_metadata->country_code_extension_byte = country_code_extension_byte; 1493| 2.08k| itut_t35_metadata->payload_size = payload_size; 1494| | 1495| | // We know that we've read a whole number of bytes and that the 1496| | // payload is within the OBU boundaries, so just use memcpy() 1497| 2.08k| assert(gb.bits_left == 0); ------------------ | Branch (1497:13): [True: 2.08k, False: 0] ------------------ 1498| 2.08k| memcpy(itut_t35_metadata->payload, gb.ptr, payload_size); 1499| | 1500| 2.08k| c->n_itut_t35++; 1501| 2.08k| break; 1502| 2.08k| } 1503| 2| case OBU_META_SCALABILITY: ------------------ | Branch (1503:9): [True: 2, False: 5.55k] ------------------ 1504| 13| case OBU_META_TIMECODE: ------------------ | Branch (1504:9): [True: 11, False: 5.54k] ------------------ 1505| | // ignore metadata OBUs we don't care about 1506| 13| break; 1507| 405| default: ------------------ | Branch (1507:9): [True: 405, False: 5.15k] ------------------ 1508| | // print a warning but don't fail for unknown types 1509| 405| if (meta_type > 31) // Types 6 to 31 are "Unregistered user private", so ignore them. ------------------ | Branch (1509:17): [True: 314, False: 91] ------------------ 1510| 314| dav1d_log(c, "Unknown Metadata OBU type %d\n", meta_type); ------------------ | | 44| 314|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 314] | | ------------------ ------------------ 1511| 405| break; 1512| 5.55k| } 1513| | 1514| 5.19k| break; 1515| 5.55k| } 1516| 61.4k| case DAV1D_OBU_TD: ------------------ | Branch (1516:5): [True: 61.4k, False: 465k] ------------------ 1517| 61.4k| c->frame_flags |= PICTURE_FLAG_NEW_TEMPORAL_UNIT; 1518| 61.4k| break; 1519| 549| case DAV1D_OBU_PADDING: ------------------ | Branch (1519:5): [True: 549, False: 526k] ------------------ 1520| | // ignore OBUs we don't care about 1521| 549| break; 1522| 6.67k| default: ------------------ | Branch (1522:5): [True: 6.67k, False: 520k] ------------------ 1523| | // print a warning but don't fail for unknown types 1524| 6.67k| dav1d_log(c, "Unknown OBU type %d of size %td\n", type, gb.ptr_end - gb.ptr); ------------------ | | 44| 6.67k|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 6.67k] | | ------------------ ------------------ 1525| 6.67k| break; 1526| 527k| } 1527| | 1528| 495k| if (c->seq_hdr && c->frame_hdr) { ------------------ | Branch (1528:9): [True: 492k, False: 3.54k] | Branch (1528:23): [True: 402k, False: 89.7k] ------------------ 1529| 402k| if (c->frame_hdr->show_existing_frame) { ------------------ | Branch (1529:13): [True: 31.7k, False: 370k] ------------------ 1530| 31.7k| if (!c->refs[c->frame_hdr->existing_frame_idx].p.p.frame_hdr) goto error; ------------------ | Branch (1530:17): [True: 283, False: 31.4k] ------------------ 1531| 31.4k| switch (c->refs[c->frame_hdr->existing_frame_idx].p.p.frame_hdr->frame_type) { 1532| 17.3k| case DAV1D_FRAME_TYPE_INTER: ------------------ | Branch (1532:13): [True: 17.3k, False: 14.0k] ------------------ 1533| 17.7k| case DAV1D_FRAME_TYPE_SWITCH: ------------------ | Branch (1533:13): [True: 391, False: 31.0k] ------------------ 1534| 17.7k| if (c->decode_frame_type > DAV1D_DECODEFRAMETYPE_REFERENCE) ------------------ | Branch (1534:21): [True: 0, False: 17.7k] ------------------ 1535| 0| goto skip; 1536| 17.7k| break; 1537| 17.7k| case DAV1D_FRAME_TYPE_INTRA: ------------------ | Branch (1537:13): [True: 359, False: 31.0k] ------------------ 1538| 359| if (c->decode_frame_type > DAV1D_DECODEFRAMETYPE_INTRA) ------------------ | Branch (1538:21): [True: 0, False: 359] ------------------ 1539| 0| goto skip; 1540| | // fall-through 1541| 13.6k| default: ------------------ | Branch (1541:13): [True: 13.3k, False: 18.1k] ------------------ 1542| 13.6k| break; 1543| 31.4k| } 1544| 31.4k| if (!c->refs[c->frame_hdr->existing_frame_idx].p.p.data[0]) goto error; ------------------ | Branch (1544:17): [True: 0, False: 31.4k] ------------------ 1545| 31.4k| if (c->strict_std_compliance && ------------------ | Branch (1545:17): [True: 0, False: 31.4k] ------------------ 1546| 0| !c->refs[c->frame_hdr->existing_frame_idx].p.showable) ------------------ | Branch (1546:17): [True: 0, False: 0] ------------------ 1547| 0| { 1548| 0| goto error; 1549| 0| } 1550| 31.4k| if (c->n_fc == 1) { ------------------ | Branch (1550:17): [True: 0, False: 31.4k] ------------------ 1551| 0| dav1d_thread_picture_ref(&c->out, 1552| 0| &c->refs[c->frame_hdr->existing_frame_idx].p); 1553| 0| dav1d_picture_copy_props(&c->out.p, 1554| 0| c->content_light, c->content_light_ref, 1555| 0| c->mastering_display, c->mastering_display_ref, 1556| 0| c->itut_t35, c->itut_t35_ref, c->n_itut_t35, 1557| 0| &in->m); 1558| | // Must be removed from the context after being attached to the frame 1559| 0| dav1d_ref_dec(&c->itut_t35_ref); 1560| 0| c->itut_t35 = NULL; 1561| 0| c->n_itut_t35 = 0; 1562| 0| c->event_flags |= dav1d_picture_get_event_flags(&c->refs[c->frame_hdr->existing_frame_idx].p); 1563| 31.4k| } else { 1564| 31.4k| pthread_mutex_lock(&c->task_thread.lock); 1565| | // need to append this to the frame output queue 1566| 31.4k| const unsigned next = c->frame_thread.next++; 1567| 31.4k| if (c->frame_thread.next == c->n_fc) ------------------ | Branch (1567:21): [True: 7.73k, False: 23.6k] ------------------ 1568| 7.73k| c->frame_thread.next = 0; 1569| | 1570| 31.4k| Dav1dFrameContext *const f = &c->fc[next]; 1571| 34.7k| while (f->n_tile_data > 0) ------------------ | Branch (1571:24): [True: 3.32k, False: 31.4k] ------------------ 1572| 3.32k| pthread_cond_wait(&f->task_thread.cond, 1573| 3.32k| &f->task_thread.ttd->lock); 1574| 31.4k| Dav1dThreadPicture *const out_delayed = 1575| 31.4k| &c->frame_thread.out_delayed[next]; 1576| 31.4k| if (out_delayed->p.data[0] || atomic_load(&f->task_thread.error)) { ------------------ | Branch (1576:21): [True: 30.6k, False: 742] | Branch (1576:47): [True: 120, False: 622] ------------------ 1577| 30.8k| unsigned first = atomic_load(&c->task_thread.first); 1578| 30.8k| if (first + 1U < c->n_fc) ------------------ | Branch (1578:25): [True: 23.2k, False: 7.56k] ------------------ 1579| 30.8k| atomic_fetch_add(&c->task_thread.first, 1U); 1580| 7.56k| else 1581| 30.8k| atomic_store(&c->task_thread.first, 0); 1582| 30.8k| atomic_compare_exchange_strong(&c->task_thread.reset_task_cur, 1583| 30.8k| &first, UINT_MAX); 1584| 30.8k| if (c->task_thread.cur && c->task_thread.cur < c->n_fc) ------------------ | Branch (1584:25): [True: 30.0k, False: 753] | Branch (1584:47): [True: 18.0k, False: 12.0k] ------------------ 1585| 18.0k| c->task_thread.cur--; 1586| 30.8k| } 1587| 31.4k| const int error = f->task_thread.retval; 1588| 31.4k| if (error) { ------------------ | Branch (1588:21): [True: 2.44k, False: 28.9k] ------------------ 1589| 2.44k| c->cached_error = error; 1590| 2.44k| f->task_thread.retval = 0; 1591| 2.44k| dav1d_data_props_copy(&c->cached_error_props, &out_delayed->p.m); 1592| 2.44k| dav1d_thread_picture_unref(out_delayed); 1593| 28.9k| } else if (out_delayed->p.data[0]) { ------------------ | Branch (1593:28): [True: 28.2k, False: 742] ------------------ 1594| 28.2k| const unsigned progress = atomic_load_explicit(&out_delayed->progress[1], 1595| 28.2k| memory_order_relaxed); 1596| 28.2k| if ((out_delayed->visible || c->output_invisible_frames) && ------------------ | Branch (1596:26): [True: 27.9k, False: 242] | Branch (1596:50): [True: 0, False: 242] ------------------ 1597| 27.9k| progress != FRAME_ERROR) ------------------ | | 35| 27.9k|#define FRAME_ERROR (UINT_MAX - 1) ------------------ | Branch (1597:25): [True: 13.7k, False: 14.2k] ------------------ 1598| 13.7k| { 1599| 13.7k| dav1d_thread_picture_ref(&c->out, out_delayed); 1600| 13.7k| c->event_flags |= dav1d_picture_get_event_flags(out_delayed); 1601| 13.7k| } 1602| 28.2k| dav1d_thread_picture_unref(out_delayed); 1603| 28.2k| } 1604| 31.4k| dav1d_thread_picture_ref(out_delayed, 1605| 31.4k| &c->refs[c->frame_hdr->existing_frame_idx].p); 1606| 31.4k| out_delayed->visible = 1; 1607| 31.4k| dav1d_picture_copy_props(&out_delayed->p, 1608| 31.4k| c->content_light, c->content_light_ref, 1609| 31.4k| c->mastering_display, c->mastering_display_ref, 1610| 31.4k| c->itut_t35, c->itut_t35_ref, c->n_itut_t35, 1611| 31.4k| &in->m); 1612| | // Must be removed from the context after being attached to the frame 1613| 31.4k| dav1d_ref_dec(&c->itut_t35_ref); 1614| 31.4k| c->itut_t35 = NULL; 1615| 31.4k| c->n_itut_t35 = 0; 1616| | 1617| 31.4k| pthread_mutex_unlock(&c->task_thread.lock); 1618| 31.4k| } 1619| 31.4k| if (c->refs[c->frame_hdr->existing_frame_idx].p.p.frame_hdr->frame_type == DAV1D_FRAME_TYPE_KEY) { ------------------ | Branch (1619:17): [True: 13.3k, False: 18.1k] ------------------ 1620| 13.3k| const int r = c->frame_hdr->existing_frame_idx; 1621| 13.3k| c->refs[r].p.showable = 0; 1622| 119k| for (int i = 0; i < 8; i++) { ------------------ | Branch (1622:33): [True: 106k, False: 13.3k] ------------------ 1623| 106k| if (i == r) continue; ------------------ | Branch (1623:25): [True: 13.3k, False: 93.2k] ------------------ 1624| | 1625| 93.2k| if (c->refs[i].p.p.frame_hdr) ------------------ | Branch (1625:25): [True: 93.1k, False: 114] ------------------ 1626| 93.1k| dav1d_thread_picture_unref(&c->refs[i].p); 1627| 93.2k| dav1d_thread_picture_ref(&c->refs[i].p, &c->refs[r].p); 1628| | 1629| 93.2k| dav1d_cdf_thread_unref(&c->cdf[i]); 1630| 93.2k| dav1d_cdf_thread_ref(&c->cdf[i], &c->cdf[r]); 1631| | 1632| 93.2k| dav1d_ref_dec(&c->refs[i].segmap); 1633| 93.2k| c->refs[i].segmap = c->refs[r].segmap; 1634| 93.2k| if (c->refs[r].segmap) ------------------ | Branch (1634:25): [True: 9.17k, False: 84.1k] ------------------ 1635| 9.17k| dav1d_ref_inc(c->refs[r].segmap); 1636| 93.2k| dav1d_ref_dec(&c->refs[i].refmvs); 1637| 93.2k| } 1638| 13.3k| } 1639| 31.4k| c->frame_hdr = NULL; 1640| 370k| } else if (c->n_tiles == c->frame_hdr->tiling.cols * c->frame_hdr->tiling.rows) { ------------------ | Branch (1640:20): [True: 359k, False: 11.2k] ------------------ 1641| 359k| switch (c->frame_hdr->frame_type) { 1642| 100k| case DAV1D_FRAME_TYPE_INTER: ------------------ | Branch (1642:13): [True: 100k, False: 258k] ------------------ 1643| 101k| case DAV1D_FRAME_TYPE_SWITCH: ------------------ | Branch (1643:13): [True: 676, False: 358k] ------------------ 1644| 101k| if (c->decode_frame_type > DAV1D_DECODEFRAMETYPE_REFERENCE || ------------------ | Branch (1644:21): [True: 0, False: 101k] ------------------ 1645| 101k| (c->decode_frame_type == DAV1D_DECODEFRAMETYPE_REFERENCE && ------------------ | Branch (1645:22): [True: 0, False: 101k] ------------------ 1646| 0| !c->frame_hdr->refresh_frame_flags)) ------------------ | Branch (1646:22): [True: 0, False: 0] ------------------ 1647| 0| goto skip; 1648| 101k| break; 1649| 101k| case DAV1D_FRAME_TYPE_INTRA: ------------------ | Branch (1649:13): [True: 903, False: 358k] ------------------ 1650| 903| if (c->decode_frame_type > DAV1D_DECODEFRAMETYPE_INTRA || ------------------ | Branch (1650:21): [True: 0, False: 903] ------------------ 1651| 903| (c->decode_frame_type == DAV1D_DECODEFRAMETYPE_REFERENCE && ------------------ | Branch (1651:22): [True: 0, False: 903] ------------------ 1652| 0| !c->frame_hdr->refresh_frame_flags)) ------------------ | Branch (1652:22): [True: 0, False: 0] ------------------ 1653| 0| goto skip; 1654| | // fall-through 1655| 258k| default: ------------------ | Branch (1655:13): [True: 257k, False: 102k] ------------------ 1656| 258k| break; 1657| 359k| } 1658| 359k| if (!c->n_tile_data) ------------------ | Branch (1658:17): [True: 0, False: 359k] ------------------ 1659| 0| goto error; 1660| 359k| if ((res = dav1d_submit_frame(c)) < 0) ------------------ | Branch (1660:17): [True: 3.01k, False: 356k] ------------------ 1661| 3.01k| return res; 1662| 359k| assert(!c->n_tile_data); ------------------ | Branch (1662:13): [True: 356k, False: 0] ------------------ 1663| 356k| c->frame_hdr = NULL; 1664| 356k| c->n_tiles = 0; 1665| 356k| } 1666| 402k| } 1667| | 1668| 492k| return gb.ptr_end - gb.ptr_start; 1669| | 1670| 0|skip: 1671| | // update refs with only the headers in case we skip the frame 1672| 0| for (int i = 0; i < 8; i++) { ------------------ | Branch (1672:21): [True: 0, False: 0] ------------------ 1673| 0| if (c->frame_hdr->refresh_frame_flags & (1 << i)) { ------------------ | Branch (1673:13): [True: 0, False: 0] ------------------ 1674| 0| dav1d_thread_picture_unref(&c->refs[i].p); 1675| 0| c->refs[i].p.p.frame_hdr = c->frame_hdr; 1676| 0| c->refs[i].p.p.seq_hdr = c->seq_hdr; 1677| 0| c->refs[i].p.p.frame_hdr_ref = c->frame_hdr_ref; 1678| 0| c->refs[i].p.p.seq_hdr_ref = c->seq_hdr_ref; 1679| 0| dav1d_ref_inc(c->frame_hdr_ref); 1680| 0| dav1d_ref_inc(c->seq_hdr_ref); 1681| 0| } 1682| 0| } 1683| | 1684| 0| dav1d_ref_dec(&c->frame_hdr_ref); 1685| 0| c->frame_hdr = NULL; 1686| 0| c->n_tiles = 0; 1687| | 1688| 0| return gb.ptr_end - gb.ptr_start; 1689| | 1690| 37.6k|error: 1691| 37.6k| dav1d_data_props_copy(&c->cached_error_props, &in->m); 1692| 37.6k| dav1d_log(c, gb.error ? "Overrun in OBU bit buffer\n" : ------------------ | | 44| 37.6k|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 37.6k] | | ------------------ ------------------ 1693| 37.6k| "Error parsing OBU data\n"); 1694| 37.6k| return DAV1D_ERR(EINVAL); ------------------ | | 58| 37.6k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 1695| 495k|} obu.c:parse_seq_hdr: 75| 44.0k|{ 76| 44.0k|#define DEBUG_SEQ_HDR 0 77| | 78| |#if DEBUG_SEQ_HDR 79| | const unsigned init_bit_pos = dav1d_get_bits_pos(gb); 80| |#endif 81| | 82| 44.0k| memset(hdr, 0, sizeof(*hdr)); 83| 44.0k| hdr->profile = dav1d_get_bits(gb, 3); 84| 44.0k| if (hdr->profile > 2) goto error; ------------------ | Branch (84:9): [True: 642, False: 43.4k] ------------------ 85| |#if DEBUG_SEQ_HDR 86| | printf("SEQHDR: post-profile: off=%u\n", 87| | dav1d_get_bits_pos(gb) - init_bit_pos); 88| |#endif 89| | 90| 43.4k| hdr->still_picture = dav1d_get_bit(gb); 91| 43.4k| hdr->reduced_still_picture_header = dav1d_get_bit(gb); 92| 43.4k| if (hdr->reduced_still_picture_header && !hdr->still_picture) goto error; ------------------ | Branch (92:9): [True: 23.4k, False: 19.9k] | Branch (92:46): [True: 1.17k, False: 22.2k] ------------------ 93| |#if DEBUG_SEQ_HDR 94| | printf("SEQHDR: post-stillpicture_flags: off=%u\n", 95| | dav1d_get_bits_pos(gb) - init_bit_pos); 96| |#endif 97| | 98| 42.2k| if (hdr->reduced_still_picture_header) { ------------------ | Branch (98:9): [True: 22.2k, False: 19.9k] ------------------ 99| 22.2k| hdr->num_operating_points = 1; 100| 22.2k| hdr->operating_points[0].major_level = dav1d_get_bits(gb, 3); 101| 22.2k| hdr->operating_points[0].minor_level = dav1d_get_bits(gb, 2); 102| 22.2k| hdr->operating_points[0].initial_display_delay = 10; 103| 22.2k| } else { 104| 19.9k| hdr->timing_info_present = dav1d_get_bit(gb); 105| 19.9k| if (hdr->timing_info_present) { ------------------ | Branch (105:13): [True: 1.29k, False: 18.6k] ------------------ 106| 1.29k| hdr->num_units_in_tick = dav1d_get_bits(gb, 32); 107| 1.29k| hdr->time_scale = dav1d_get_bits(gb, 32); 108| 1.29k| if (strict_std_compliance && (!hdr->num_units_in_tick || !hdr->time_scale)) ------------------ | Branch (108:17): [True: 0, False: 1.29k] | Branch (108:43): [True: 0, False: 0] | Branch (108:70): [True: 0, False: 0] ------------------ 109| 0| goto error; 110| 1.29k| hdr->equal_picture_interval = dav1d_get_bit(gb); 111| 1.29k| if (hdr->equal_picture_interval) { ------------------ | Branch (111:17): [True: 650, False: 640] ------------------ 112| 650| const unsigned num_ticks_per_picture = dav1d_get_vlc(gb); 113| 650| if (num_ticks_per_picture == UINT32_MAX) ------------------ | Branch (113:21): [True: 73, False: 577] ------------------ 114| 73| goto error; 115| 577| hdr->num_ticks_per_picture = num_ticks_per_picture + 1; 116| 577| } 117| | 118| 1.21k| hdr->decoder_model_info_present = dav1d_get_bit(gb); 119| 1.21k| if (hdr->decoder_model_info_present) { ------------------ | Branch (119:17): [True: 853, False: 364] ------------------ 120| 853| hdr->encoder_decoder_buffer_delay_length = dav1d_get_bits(gb, 5) + 1; 121| 853| hdr->num_units_in_decoding_tick = dav1d_get_bits(gb, 32); 122| 853| if (strict_std_compliance && !hdr->num_units_in_decoding_tick) ------------------ | Branch (122:21): [True: 0, False: 853] | Branch (122:46): [True: 0, False: 0] ------------------ 123| 0| goto error; 124| 853| hdr->buffer_removal_delay_length = dav1d_get_bits(gb, 5) + 1; 125| 853| hdr->frame_presentation_delay_length = dav1d_get_bits(gb, 5) + 1; 126| 853| } 127| 1.21k| } 128| |#if DEBUG_SEQ_HDR 129| | printf("SEQHDR: post-timinginfo: off=%u\n", 130| | dav1d_get_bits_pos(gb) - init_bit_pos); 131| |#endif 132| | 133| 19.8k| hdr->display_model_info_present = dav1d_get_bit(gb); 134| 19.8k| hdr->num_operating_points = dav1d_get_bits(gb, 5) + 1; 135| 54.3k| for (int i = 0; i < hdr->num_operating_points; i++) { ------------------ | Branch (135:25): [True: 37.0k, False: 17.2k] ------------------ 136| 37.0k| struct Dav1dSequenceHeaderOperatingPoint *const op = 137| 37.0k| &hdr->operating_points[i]; 138| 37.0k| op->idc = dav1d_get_bits(gb, 12); 139| 37.0k| if (op->idc && (!(op->idc & 0xff) || !(op->idc & 0xf00))) ------------------ | Branch (139:17): [True: 25.1k, False: 11.8k] | Branch (139:29): [True: 2.38k, False: 22.7k] | Branch (139:50): [True: 236, False: 22.5k] ------------------ 140| 2.61k| goto error; 141| 34.4k| op->major_level = 2 + dav1d_get_bits(gb, 3); 142| 34.4k| op->minor_level = dav1d_get_bits(gb, 2); 143| 34.4k| if (op->major_level > 3) ------------------ | Branch (143:17): [True: 9.81k, False: 24.6k] ------------------ 144| 9.81k| op->tier = dav1d_get_bit(gb); 145| 34.4k| if (hdr->decoder_model_info_present) { ------------------ | Branch (145:17): [True: 3.37k, False: 31.0k] ------------------ 146| 3.37k| op->decoder_model_param_present = dav1d_get_bit(gb); 147| 3.37k| if (op->decoder_model_param_present) { ------------------ | Branch (147:21): [True: 1.34k, False: 2.03k] ------------------ 148| 1.34k| struct Dav1dSequenceHeaderOperatingParameterInfo *const opi = 149| 1.34k| &hdr->operating_parameter_info[i]; 150| 1.34k| opi->decoder_buffer_delay = 151| 1.34k| dav1d_get_bits(gb, hdr->encoder_decoder_buffer_delay_length); 152| 1.34k| opi->encoder_buffer_delay = 153| 1.34k| dav1d_get_bits(gb, hdr->encoder_decoder_buffer_delay_length); 154| 1.34k| opi->low_delay_mode = dav1d_get_bit(gb); 155| 1.34k| } 156| 3.37k| } 157| 34.4k| if (hdr->display_model_info_present) ------------------ | Branch (157:17): [True: 5.64k, False: 28.7k] ------------------ 158| 5.64k| op->display_model_param_present = dav1d_get_bit(gb); 159| 34.4k| op->initial_display_delay = 160| 34.4k| op->display_model_param_present ? dav1d_get_bits(gb, 4) + 1 : 10; ------------------ | Branch (160:17): [True: 1.24k, False: 33.1k] ------------------ 161| 34.4k| } 162| |#if DEBUG_SEQ_HDR 163| | printf("SEQHDR: post-operating-points: off=%u\n", 164| | dav1d_get_bits_pos(gb) - init_bit_pos); 165| |#endif 166| 19.8k| } 167| | 168| 39.5k| hdr->width_n_bits = dav1d_get_bits(gb, 4) + 1; 169| 39.5k| hdr->height_n_bits = dav1d_get_bits(gb, 4) + 1; 170| 39.5k| hdr->max_width = dav1d_get_bits(gb, hdr->width_n_bits) + 1; 171| 39.5k| hdr->max_height = dav1d_get_bits(gb, hdr->height_n_bits) + 1; 172| |#if DEBUG_SEQ_HDR 173| | printf("SEQHDR: post-size: off=%u\n", 174| | dav1d_get_bits_pos(gb) - init_bit_pos); 175| |#endif 176| 39.5k| if (!hdr->reduced_still_picture_header) { ------------------ | Branch (176:9): [True: 17.2k, False: 22.2k] ------------------ 177| 17.2k| hdr->frame_id_numbers_present = dav1d_get_bit(gb); 178| 17.2k| if (hdr->frame_id_numbers_present) { ------------------ | Branch (178:13): [True: 870, False: 16.4k] ------------------ 179| 870| hdr->delta_frame_id_n_bits = dav1d_get_bits(gb, 4) + 2; 180| 870| hdr->frame_id_n_bits = dav1d_get_bits(gb, 3) + hdr->delta_frame_id_n_bits + 1; 181| 870| } 182| 17.2k| } 183| |#if DEBUG_SEQ_HDR 184| | printf("SEQHDR: post-frame-id-numbers-present: off=%u\n", 185| | dav1d_get_bits_pos(gb) - init_bit_pos); 186| |#endif 187| | 188| 39.5k| hdr->sb128 = dav1d_get_bit(gb); 189| 39.5k| hdr->filter_intra = dav1d_get_bit(gb); 190| 39.5k| hdr->intra_edge_filter = dav1d_get_bit(gb); 191| 39.5k| if (hdr->reduced_still_picture_header) { ------------------ | Branch (191:9): [True: 22.2k, False: 17.2k] ------------------ 192| 22.2k| hdr->screen_content_tools = DAV1D_ADAPTIVE; 193| 22.2k| hdr->force_integer_mv = DAV1D_ADAPTIVE; 194| 22.2k| } else { 195| 17.2k| hdr->inter_intra = dav1d_get_bit(gb); 196| 17.2k| hdr->masked_compound = dav1d_get_bit(gb); 197| 17.2k| hdr->warped_motion = dav1d_get_bit(gb); 198| 17.2k| hdr->dual_filter = dav1d_get_bit(gb); 199| 17.2k| hdr->order_hint = dav1d_get_bit(gb); 200| 17.2k| if (hdr->order_hint) { ------------------ | Branch (200:13): [True: 10.3k, False: 6.92k] ------------------ 201| 10.3k| hdr->jnt_comp = dav1d_get_bit(gb); 202| 10.3k| hdr->ref_frame_mvs = dav1d_get_bit(gb); 203| 10.3k| } 204| 17.2k| hdr->screen_content_tools = dav1d_get_bit(gb) ? DAV1D_ADAPTIVE : dav1d_get_bit(gb); ------------------ | Branch (204:37): [True: 13.5k, False: 3.76k] ------------------ 205| | #if DEBUG_SEQ_HDR 206| | printf("SEQHDR: post-screentools: off=%u\n", 207| | dav1d_get_bits_pos(gb) - init_bit_pos); 208| | #endif 209| 17.2k| hdr->force_integer_mv = hdr->screen_content_tools ? ------------------ | Branch (209:33): [True: 14.7k, False: 2.49k] ------------------ 210| 14.7k| dav1d_get_bit(gb) ? DAV1D_ADAPTIVE : dav1d_get_bit(gb) : 2; ------------------ | Branch (210:33): [True: 7.95k, False: 6.82k] ------------------ 211| 17.2k| if (hdr->order_hint) ------------------ | Branch (211:13): [True: 10.3k, False: 6.92k] ------------------ 212| 10.3k| hdr->order_hint_n_bits = dav1d_get_bits(gb, 3) + 1; 213| 17.2k| } 214| 39.5k| hdr->super_res = dav1d_get_bit(gb); 215| 39.5k| hdr->cdef = dav1d_get_bit(gb); 216| 39.5k| hdr->restoration = dav1d_get_bit(gb); 217| |#if DEBUG_SEQ_HDR 218| | printf("SEQHDR: post-featurebits: off=%u\n", 219| | dav1d_get_bits_pos(gb) - init_bit_pos); 220| |#endif 221| | 222| 39.5k| hdr->hbd = dav1d_get_bit(gb); 223| 39.5k| if (hdr->profile == 2 && hdr->hbd) ------------------ | Branch (223:9): [True: 14.4k, False: 25.0k] | Branch (223:30): [True: 10.9k, False: 3.47k] ------------------ 224| 10.9k| hdr->hbd += dav1d_get_bit(gb); 225| 39.5k| if (hdr->profile != 1) ------------------ | Branch (225:9): [True: 29.5k, False: 10.0k] ------------------ 226| 29.5k| hdr->monochrome = dav1d_get_bit(gb); 227| 39.5k| hdr->color_description_present = dav1d_get_bit(gb); 228| 39.5k| if (hdr->color_description_present) { ------------------ | Branch (228:9): [True: 6.02k, False: 33.5k] ------------------ 229| 6.02k| hdr->pri = dav1d_get_bits(gb, 8); 230| 6.02k| hdr->trc = dav1d_get_bits(gb, 8); 231| 6.02k| hdr->mtrx = dav1d_get_bits(gb, 8); 232| 33.5k| } else { 233| 33.5k| hdr->pri = DAV1D_COLOR_PRI_UNKNOWN; 234| 33.5k| hdr->trc = DAV1D_TRC_UNKNOWN; 235| 33.5k| hdr->mtrx = DAV1D_MC_UNKNOWN; 236| 33.5k| } 237| 39.5k| if (hdr->monochrome) { ------------------ | Branch (237:9): [True: 9.26k, False: 30.2k] ------------------ 238| 9.26k| hdr->color_range = dav1d_get_bit(gb); 239| 9.26k| hdr->layout = DAV1D_PIXEL_LAYOUT_I400; 240| 9.26k| hdr->ss_hor = hdr->ss_ver = 1; 241| 9.26k| hdr->chr = DAV1D_CHR_UNKNOWN; 242| 30.2k| } else if (hdr->pri == DAV1D_COLOR_PRI_BT709 && ------------------ | Branch (242:16): [True: 2.12k, False: 28.1k] ------------------ 243| 2.12k| hdr->trc == DAV1D_TRC_SRGB && ------------------ | Branch (243:16): [True: 1.77k, False: 356] ------------------ 244| 1.77k| hdr->mtrx == DAV1D_MC_IDENTITY) ------------------ | Branch (244:16): [True: 1.39k, False: 373] ------------------ 245| 1.39k| { 246| 1.39k| hdr->layout = DAV1D_PIXEL_LAYOUT_I444; 247| 1.39k| hdr->color_range = 1; 248| 1.39k| if (hdr->profile != 1 && !(hdr->profile == 2 && hdr->hbd == 2)) ------------------ | Branch (248:13): [True: 1.08k, False: 315] | Branch (248:36): [True: 198, False: 885] | Branch (248:57): [True: 93, False: 105] ------------------ 249| 990| goto error; 250| 28.8k| } else { 251| 28.8k| hdr->color_range = dav1d_get_bit(gb); 252| 28.8k| switch (hdr->profile) { ------------------ | Branch (252:17): [True: 28.8k, False: 0] ------------------ 253| 11.2k| case 0: hdr->layout = DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (253:9): [True: 11.2k, False: 17.6k] ------------------ 254| 11.2k| hdr->ss_hor = hdr->ss_ver = 1; 255| 11.2k| break; 256| 9.73k| case 1: hdr->layout = DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (256:9): [True: 9.73k, False: 19.1k] ------------------ 257| 9.73k| break; 258| 7.89k| case 2: ------------------ | Branch (258:9): [True: 7.89k, False: 21.0k] ------------------ 259| 7.89k| if (hdr->hbd == 2) { ------------------ | Branch (259:17): [True: 5.20k, False: 2.69k] ------------------ 260| 5.20k| hdr->ss_hor = dav1d_get_bit(gb); 261| 5.20k| if (hdr->ss_hor) ------------------ | Branch (261:21): [True: 1.37k, False: 3.82k] ------------------ 262| 1.37k| hdr->ss_ver = dav1d_get_bit(gb); 263| 5.20k| } else 264| 2.69k| hdr->ss_hor = 1; 265| 7.89k| hdr->layout = hdr->ss_hor ? ------------------ | Branch (265:27): [True: 4.06k, False: 3.82k] ------------------ 266| 4.06k| hdr->ss_ver ? DAV1D_PIXEL_LAYOUT_I420 : ------------------ | Branch (266:27): [True: 1.28k, False: 2.78k] ------------------ 267| 4.06k| DAV1D_PIXEL_LAYOUT_I422 : 268| 7.89k| DAV1D_PIXEL_LAYOUT_I444; 269| 7.89k| break; 270| 28.8k| } 271| 28.8k| hdr->chr = (hdr->ss_hor & hdr->ss_ver) ? ------------------ | Branch (271:20): [True: 12.5k, False: 16.3k] ------------------ 272| 16.3k| dav1d_get_bits(gb, 2) : DAV1D_CHR_UNKNOWN; 273| 28.8k| } 274| 38.5k| if (strict_std_compliance && ------------------ | Branch (274:9): [True: 0, False: 38.5k] ------------------ 275| 0| hdr->mtrx == DAV1D_MC_IDENTITY && hdr->layout != DAV1D_PIXEL_LAYOUT_I444) ------------------ | Branch (275:9): [True: 0, False: 0] | Branch (275:43): [True: 0, False: 0] ------------------ 276| 0| { 277| 0| goto error; 278| 0| } 279| 38.5k| if (!hdr->monochrome) ------------------ | Branch (279:9): [True: 29.3k, False: 9.26k] ------------------ 280| 29.3k| hdr->separate_uv_delta_q = dav1d_get_bit(gb); 281| |#if DEBUG_SEQ_HDR 282| | printf("SEQHDR: post-colorinfo: off=%u\n", 283| | dav1d_get_bits_pos(gb) - init_bit_pos); 284| |#endif 285| | 286| 38.5k| hdr->film_grain_present = dav1d_get_bit(gb); 287| |#if DEBUG_SEQ_HDR 288| | printf("SEQHDR: post-filmgrain: off=%u\n", 289| | dav1d_get_bits_pos(gb) - init_bit_pos); 290| |#endif 291| | 292| | // We needn't bother flushing the OBU here: we'll check we didn't 293| | // overrun in the caller and will then discard gb, so there's no 294| | // point in setting its position properly. 295| | 296| 38.5k| return check_trailing_bits(gb, strict_std_compliance); 297| | 298| 5.49k|error: 299| 5.49k| return DAV1D_ERR(EINVAL); ------------------ | | 58| 5.49k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 300| 38.5k|} obu.c:parse_frame_hdr: 409| 417k|static int parse_frame_hdr(Dav1dContext *const c, GetBits *const gb) { 410| 417k|#define DEBUG_FRAME_HDR 0 411| | 412| |#if DEBUG_FRAME_HDR 413| | const uint8_t *const init_ptr = gb->ptr; 414| |#endif 415| 417k| const Dav1dSequenceHeader *const seqhdr = c->seq_hdr; 416| 417k| Dav1dFrameHeader *const hdr = c->frame_hdr; 417| | 418| 417k| if (!seqhdr->reduced_still_picture_header) ------------------ | Branch (418:9): [True: 166k, False: 250k] ------------------ 419| 166k| hdr->show_existing_frame = dav1d_get_bit(gb); 420| |#if DEBUG_FRAME_HDR 421| | printf("HDR: post-show_existing_frame: off=%td\n", 422| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 423| |#endif 424| 417k| if (hdr->show_existing_frame) { ------------------ | Branch (424:9): [True: 32.8k, False: 384k] ------------------ 425| 32.8k| hdr->existing_frame_idx = dav1d_get_bits(gb, 3); 426| 32.8k| if (seqhdr->decoder_model_info_present && !seqhdr->equal_picture_interval) ------------------ | Branch (426:13): [True: 604, False: 32.1k] | Branch (426:51): [True: 381, False: 223] ------------------ 427| 381| hdr->frame_presentation_delay = dav1d_get_bits(gb, seqhdr->frame_presentation_delay_length); 428| 32.8k| if (seqhdr->frame_id_numbers_present) { ------------------ | Branch (428:13): [True: 962, False: 31.8k] ------------------ 429| 962| hdr->frame_id = dav1d_get_bits(gb, seqhdr->frame_id_n_bits); 430| 962| Dav1dFrameHeader *const ref_frame_hdr = c->refs[hdr->existing_frame_idx].p.p.frame_hdr; 431| 962| if (!ref_frame_hdr || ref_frame_hdr->frame_id != hdr->frame_id) goto error; ------------------ | Branch (431:17): [True: 400, False: 562] | Branch (431:35): [True: 162, False: 400] ------------------ 432| 962| } 433| 32.2k| return 0; 434| 32.8k| } 435| | 436| 384k| if (seqhdr->reduced_still_picture_header) { ------------------ | Branch (436:9): [True: 250k, False: 133k] ------------------ 437| 250k| hdr->frame_type = DAV1D_FRAME_TYPE_KEY; 438| 250k| hdr->show_frame = 1; 439| 250k| } else { 440| 133k| hdr->frame_type = dav1d_get_bits(gb, 2); 441| 133k| hdr->show_frame = dav1d_get_bit(gb); 442| 133k| } 443| 384k| if (hdr->show_frame) { ------------------ | Branch (443:9): [True: 312k, False: 72.6k] ------------------ 444| 312k| if (seqhdr->decoder_model_info_present && !seqhdr->equal_picture_interval) ------------------ | Branch (444:13): [True: 9.83k, False: 302k] | Branch (444:51): [True: 9.25k, False: 580] ------------------ 445| 9.25k| hdr->frame_presentation_delay = dav1d_get_bits(gb, seqhdr->frame_presentation_delay_length); 446| 312k| hdr->showable_frame = hdr->frame_type != DAV1D_FRAME_TYPE_KEY; 447| 312k| } else 448| 72.6k| hdr->showable_frame = dav1d_get_bit(gb); 449| 384k| hdr->error_resilient_mode = 450| 384k| (hdr->frame_type == DAV1D_FRAME_TYPE_KEY && hdr->show_frame) || ------------------ | Branch (450:10): [True: 268k, False: 116k] | Branch (450:53): [True: 265k, False: 3.09k] ------------------ 451| 119k| hdr->frame_type == DAV1D_FRAME_TYPE_SWITCH || ------------------ | Branch (451:9): [True: 2.86k, False: 116k] ------------------ 452| 116k| seqhdr->reduced_still_picture_header || dav1d_get_bit(gb); ------------------ | Branch (452:9): [True: 0, False: 116k] | Branch (452:49): [True: 2.27k, False: 113k] ------------------ 453| |#if DEBUG_FRAME_HDR 454| | printf("HDR: post-frametype_bits: off=%td\n", 455| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 456| |#endif 457| 384k| hdr->disable_cdf_update = dav1d_get_bit(gb); 458| 384k| hdr->allow_screen_content_tools = seqhdr->screen_content_tools == DAV1D_ADAPTIVE ? ------------------ | Branch (458:39): [True: 352k, False: 32.2k] ------------------ 459| 352k| dav1d_get_bit(gb) : seqhdr->screen_content_tools; 460| 384k| if (hdr->allow_screen_content_tools) ------------------ | Branch (460:9): [True: 293k, False: 91.6k] ------------------ 461| 293k| hdr->force_integer_mv = seqhdr->force_integer_mv == DAV1D_ADAPTIVE ? ------------------ | Branch (461:33): [True: 240k, False: 52.9k] ------------------ 462| 240k| dav1d_get_bit(gb) : seqhdr->force_integer_mv; 463| | 464| 384k| if (IS_KEY_OR_INTRA(hdr)) ------------------ | | 43| 384k| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 384k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (43:5): [True: 270k, False: 114k] | | ------------------ ------------------ 465| 270k| hdr->force_integer_mv = 1; 466| | 467| 384k| if (seqhdr->frame_id_numbers_present) ------------------ | Branch (467:9): [True: 3.38k, False: 381k] ------------------ 468| 3.38k| hdr->frame_id = dav1d_get_bits(gb, seqhdr->frame_id_n_bits); 469| | 470| 384k| if (!seqhdr->reduced_still_picture_header) ------------------ | Branch (470:9): [True: 133k, False: 250k] ------------------ 471| 133k| hdr->frame_size_override = hdr->frame_type == DAV1D_FRAME_TYPE_SWITCH ? 1 : dav1d_get_bit(gb); ------------------ | Branch (471:36): [True: 2.86k, False: 131k] ------------------ 472| |#if DEBUG_FRAME_HDR 473| | printf("HDR: post-frame_size_override_flag: off=%td\n", 474| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 475| |#endif 476| 384k| if (seqhdr->order_hint) ------------------ | Branch (476:9): [True: 120k, False: 264k] ------------------ 477| 120k| hdr->frame_offset = dav1d_get_bits(gb, seqhdr->order_hint_n_bits); 478| 384k| hdr->primary_ref_frame = !hdr->error_resilient_mode && IS_INTER_OR_SWITCH(hdr) ? ------------------ | | 36| 113k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 111k, False: 2.48k] | | ------------------ ------------------ | Branch (478:30): [True: 113k, False: 270k] ------------------ 479| 273k| dav1d_get_bits(gb, 3) : DAV1D_PRIMARY_REF_NONE; ------------------ | | 45| 658k|#define DAV1D_PRIMARY_REF_NONE 7 ------------------ 480| | 481| 384k| if (seqhdr->decoder_model_info_present) { ------------------ | Branch (481:9): [True: 10.4k, False: 374k] ------------------ 482| 10.4k| hdr->buffer_removal_time_present = dav1d_get_bit(gb); 483| 10.4k| if (hdr->buffer_removal_time_present) { ------------------ | Branch (483:13): [True: 3.24k, False: 7.17k] ------------------ 484| 9.29k| for (int i = 0; i < c->seq_hdr->num_operating_points; i++) { ------------------ | Branch (484:29): [True: 6.05k, False: 3.24k] ------------------ 485| 6.05k| const struct Dav1dSequenceHeaderOperatingPoint *const seqop = &seqhdr->operating_points[i]; 486| 6.05k| struct Dav1dFrameHeaderOperatingPoint *const op = &hdr->operating_points[i]; 487| 6.05k| if (seqop->decoder_model_param_present) { ------------------ | Branch (487:21): [True: 2.01k, False: 4.03k] ------------------ 488| 2.01k| int in_temporal_layer = (seqop->idc >> hdr->temporal_id) & 1; 489| 2.01k| int in_spatial_layer = (seqop->idc >> (hdr->spatial_id + 8)) & 1; 490| 2.01k| if (!seqop->idc || (in_temporal_layer && in_spatial_layer)) ------------------ | Branch (490:25): [True: 278, False: 1.73k] | Branch (490:41): [True: 782, False: 957] | Branch (490:62): [True: 513, False: 269] ------------------ 491| 791| op->buffer_removal_time = dav1d_get_bits(gb, seqhdr->buffer_removal_delay_length); 492| 2.01k| } 493| 6.05k| } 494| 3.24k| } 495| 10.4k| } 496| | 497| 384k| if (IS_KEY_OR_INTRA(hdr)) { ------------------ | | 43| 384k| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 384k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (43:5): [True: 270k, False: 114k] | | ------------------ ------------------ 498| 270k| hdr->refresh_frame_flags = (hdr->frame_type == DAV1D_FRAME_TYPE_KEY && ------------------ | Branch (498:37): [True: 268k, False: 1.48k] ------------------ 499| 268k| hdr->show_frame) ? 0xff : dav1d_get_bits(gb, 8); ------------------ | Branch (499:37): [True: 265k, False: 3.09k] ------------------ 500| 270k| if (hdr->refresh_frame_flags != 0xff && hdr->error_resilient_mode && seqhdr->order_hint) ------------------ | Branch (500:13): [True: 4.52k, False: 265k] | Branch (500:49): [True: 2.09k, False: 2.43k] | Branch (500:78): [True: 1.19k, False: 896] ------------------ 501| 10.7k| for (int i = 0; i < 8; i++) ------------------ | Branch (501:29): [True: 9.57k, False: 1.19k] ------------------ 502| 9.57k| dav1d_get_bits(gb, seqhdr->order_hint_n_bits); 503| 270k| if (c->strict_std_compliance && ------------------ | Branch (503:13): [True: 0, False: 270k] ------------------ 504| 0| hdr->frame_type == DAV1D_FRAME_TYPE_INTRA && hdr->refresh_frame_flags == 0xff) ------------------ | Branch (504:13): [True: 0, False: 0] | Branch (504:58): [True: 0, False: 0] ------------------ 505| 0| { 506| 0| goto error; 507| 0| } 508| 270k| if (read_frame_size(c, gb, 0) < 0) goto error; ------------------ | Branch (508:13): [True: 0, False: 270k] ------------------ 509| 270k| if (hdr->allow_screen_content_tools && !hdr->super_res.enabled) ------------------ | Branch (509:13): [True: 239k, False: 30.4k] | Branch (509:48): [True: 237k, False: 2.24k] ------------------ 510| 237k| hdr->allow_intrabc = dav1d_get_bit(gb); 511| 270k| } else { 512| 114k| hdr->refresh_frame_flags = hdr->frame_type == DAV1D_FRAME_TYPE_SWITCH ? 0xff : ------------------ | Branch (512:36): [True: 2.86k, False: 111k] ------------------ 513| 114k| dav1d_get_bits(gb, 8); 514| 114k| if (hdr->error_resilient_mode && seqhdr->order_hint) ------------------ | Branch (514:13): [True: 3.05k, False: 111k] | Branch (514:42): [True: 2.40k, False: 646] ------------------ 515| 21.6k| for (int i = 0; i < 8; i++) ------------------ | Branch (515:29): [True: 19.2k, False: 2.40k] ------------------ 516| 19.2k| dav1d_get_bits(gb, seqhdr->order_hint_n_bits); 517| 114k| if (seqhdr->order_hint) { ------------------ | Branch (517:13): [True: 107k, False: 7.14k] ------------------ 518| 107k| hdr->frame_ref_short_signaling = dav1d_get_bit(gb); 519| 107k| if (hdr->frame_ref_short_signaling) { ------------------ | Branch (519:17): [True: 68.7k, False: 38.6k] ------------------ 520| 68.7k| hdr->refidx[0] = dav1d_get_bits(gb, 3); 521| 68.7k| hdr->refidx[1] = hdr->refidx[2] = -1; 522| 68.7k| hdr->refidx[3] = dav1d_get_bits(gb, 3); 523| | 524| | /* +1 allows for unconditional stores, as unused 525| | * values can be dumped into frame_offset[-1]. */ 526| 68.7k| int frame_offset_mem[8+1]; 527| 68.7k| int *const frame_offset = &frame_offset_mem[1]; 528| 68.7k| int earliest_ref = -1; 529| 614k| for (int i = 0, earliest_offset = INT_MAX; i < 8; i++) { ------------------ | Branch (529:60): [True: 546k, False: 68.1k] ------------------ 530| 546k| const Dav1dFrameHeader *const refhdr = c->refs[i].p.p.frame_hdr; 531| 546k| if (!refhdr) goto error; ------------------ | Branch (531:25): [True: 535, False: 545k] ------------------ 532| 545k| const int diff = get_poc_diff(seqhdr->order_hint_n_bits, 533| 545k| refhdr->frame_offset, 534| 545k| hdr->frame_offset); 535| 545k| frame_offset[i] = diff; 536| 545k| if (diff < earliest_offset) { ------------------ | Branch (536:25): [True: 103k, False: 441k] ------------------ 537| 103k| earliest_offset = diff; 538| 103k| earliest_ref = i; 539| 103k| } 540| 545k| } 541| 68.1k| frame_offset[hdr->refidx[0]] = INT_MIN; // = reference frame is used 542| 68.1k| frame_offset[hdr->refidx[3]] = INT_MIN; 543| 68.1k| assert(earliest_ref >= 0); ------------------ | Branch (543:17): [True: 68.1k, False: 0] ------------------ 544| | 545| 68.1k| int refidx = -1; 546| 613k| for (int i = 0, latest_offset = 0; i < 8; i++) { ------------------ | Branch (546:52): [True: 545k, False: 68.1k] ------------------ 547| 545k| const int hint = frame_offset[i]; 548| 545k| if (hint >= latest_offset) { ------------------ | Branch (548:25): [True: 255k, False: 289k] ------------------ 549| 255k| latest_offset = hint; 550| 255k| refidx = i; 551| 255k| } 552| 545k| } 553| 68.1k| frame_offset[refidx] = INT_MIN; 554| 68.1k| hdr->refidx[6] = refidx; 555| | 556| 204k| for (int i = 4; i < 6; i++) { ------------------ | Branch (556:33): [True: 136k, False: 68.1k] ------------------ 557| | /* Unsigned compares to handle negative values. */ 558| 136k| unsigned earliest_offset = UINT8_MAX; 559| 136k| refidx = -1; 560| 1.22M| for (int j = 0; j < 8; j++) { ------------------ | Branch (560:37): [True: 1.09M, False: 136k] ------------------ 561| 1.09M| const unsigned hint = frame_offset[j]; 562| 1.09M| if (hint < earliest_offset) { ------------------ | Branch (562:29): [True: 128k, False: 961k] ------------------ 563| 128k| earliest_offset = hint; 564| 128k| refidx = j; 565| 128k| } 566| 1.09M| } 567| 136k| frame_offset[refidx] = INT_MIN; 568| 136k| hdr->refidx[i] = refidx; 569| 136k| } 570| | 571| 477k| for (int i = 1; i < 7; i++) { ------------------ | Branch (571:33): [True: 409k, False: 68.1k] ------------------ 572| 409k| refidx = hdr->refidx[i]; 573| 409k| if (refidx < 0) { ------------------ | Branch (573:25): [True: 159k, False: 249k] ------------------ 574| 159k| unsigned latest_offset = ~UINT8_MAX; 575| 1.43M| for (int j = 0; j < 8; j++) { ------------------ | Branch (575:41): [True: 1.27M, False: 159k] ------------------ 576| 1.27M| const unsigned hint = frame_offset[j]; 577| 1.27M| if (hint >= latest_offset) { ------------------ | Branch (577:33): [True: 282k, False: 995k] ------------------ 578| 282k| latest_offset = hint; 579| 282k| refidx = j; 580| 282k| } 581| 1.27M| } 582| 159k| frame_offset[refidx] = INT_MIN; 583| 159k| hdr->refidx[i] = refidx >= 0 ? refidx : earliest_ref; ------------------ | Branch (583:42): [True: 109k, False: 49.9k] ------------------ 584| 159k| } 585| 409k| } 586| 68.1k| } 587| 107k| } 588| 895k| for (int i = 0; i < 7; i++) { ------------------ | Branch (588:25): [True: 783k, False: 111k] ------------------ 589| 783k| if (!hdr->frame_ref_short_signaling) ------------------ | Branch (589:17): [True: 307k, False: 476k] ------------------ 590| 307k| hdr->refidx[i] = dav1d_get_bits(gb, 3); 591| 783k| if (seqhdr->frame_id_numbers_present) { ------------------ | Branch (591:17): [True: 2.43k, False: 781k] ------------------ 592| 2.43k| const unsigned delta_ref_frame_id = dav1d_get_bits(gb, seqhdr->delta_frame_id_n_bits) + 1; 593| 2.43k| const unsigned ref_frame_id = (hdr->frame_id + (1 << seqhdr->frame_id_n_bits) - delta_ref_frame_id) & ((1 << seqhdr->frame_id_n_bits) - 1); 594| 2.43k| Dav1dFrameHeader *const ref_frame_hdr = c->refs[hdr->refidx[i]].p.p.frame_hdr; 595| 2.43k| if (!ref_frame_hdr || ref_frame_hdr->frame_id != ref_frame_id) goto error; ------------------ | Branch (595:21): [True: 477, False: 1.96k] | Branch (595:39): [True: 1.89k, False: 69] ------------------ 596| 2.43k| } 597| 783k| } 598| 111k| const int use_ref = !hdr->error_resilient_mode && ------------------ | Branch (598:29): [True: 109k, False: 2.28k] ------------------ 599| 109k| hdr->frame_size_override; ------------------ | Branch (599:29): [True: 62.0k, False: 47.2k] ------------------ 600| 111k| if (read_frame_size(c, gb, use_ref) < 0) goto error; ------------------ | Branch (600:13): [True: 259, False: 111k] ------------------ 601| 111k| if (!hdr->force_integer_mv) ------------------ | Branch (601:13): [True: 70.5k, False: 40.8k] ------------------ 602| 70.5k| hdr->hp = dav1d_get_bit(gb); 603| 111k| hdr->subpel_filter_mode = dav1d_get_bit(gb) ? DAV1D_FILTER_SWITCHABLE : ------------------ | Branch (603:35): [True: 64.4k, False: 46.9k] ------------------ 604| 111k| dav1d_get_bits(gb, 2); 605| 111k| hdr->switchable_motion_mode = dav1d_get_bit(gb); 606| 111k| if (!hdr->error_resilient_mode && seqhdr->ref_frame_mvs && ------------------ | Branch (606:13): [True: 109k, False: 2.28k] | Branch (606:43): [True: 91.9k, False: 17.1k] ------------------ 607| 91.9k| seqhdr->order_hint && IS_INTER_OR_SWITCH(hdr)) ------------------ | | 36| 91.9k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 91.9k, False: 0] | | ------------------ ------------------ | Branch (607:13): [True: 91.9k, False: 0] ------------------ 608| 91.9k| { 609| 91.9k| hdr->use_ref_frame_mvs = dav1d_get_bit(gb); 610| 91.9k| } 611| 111k| } 612| |#if DEBUG_FRAME_HDR 613| | printf("HDR: post-frametype-specific-bits: off=%td\n", 614| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 615| |#endif 616| | 617| 381k| if (!seqhdr->reduced_still_picture_header && !hdr->disable_cdf_update) ------------------ | Branch (617:9): [True: 130k, False: 250k] | Branch (617:50): [True: 122k, False: 7.86k] ------------------ 618| 122k| hdr->refresh_context = !dav1d_get_bit(gb); 619| |#if DEBUG_FRAME_HDR 620| | printf("HDR: post-refresh_context: off=%td\n", 621| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 622| |#endif 623| | 624| | // tile data 625| 381k| hdr->tiling.uniform = dav1d_get_bit(gb); 626| 381k| const int sbsz_min1 = (64 << seqhdr->sb128) - 1; 627| 381k| const int sbsz_log2 = 6 + seqhdr->sb128; 628| 381k| const int sbw = (hdr->width[0] + sbsz_min1) >> sbsz_log2; 629| 381k| const int sbh = (hdr->height + sbsz_min1) >> sbsz_log2; 630| 381k| const int max_tile_width_sb = 4096 >> sbsz_log2; 631| 381k| const int max_tile_area_sb = 4096 * 2304 >> (2 * sbsz_log2); 632| 381k| hdr->tiling.min_log2_cols = tile_log2(max_tile_width_sb, sbw); 633| 381k| hdr->tiling.max_log2_cols = tile_log2(1, imin(sbw, DAV1D_MAX_TILE_COLS)); ------------------ | | 41| 381k|#define DAV1D_MAX_TILE_COLS 64 ------------------ 634| 381k| hdr->tiling.max_log2_rows = tile_log2(1, imin(sbh, DAV1D_MAX_TILE_ROWS)); ------------------ | | 42| 381k|#define DAV1D_MAX_TILE_ROWS 64 ------------------ 635| 381k| const int min_log2_tiles = imax(tile_log2(max_tile_area_sb, sbw * sbh), 636| 381k| hdr->tiling.min_log2_cols); 637| 381k| if (hdr->tiling.uniform) { ------------------ | Branch (637:9): [True: 326k, False: 54.7k] ------------------ 638| 326k| for (hdr->tiling.log2_cols = hdr->tiling.min_log2_cols; 639| 333k| hdr->tiling.log2_cols < hdr->tiling.max_log2_cols && dav1d_get_bit(gb); ------------------ | Branch (639:14): [True: 92.6k, False: 240k] | Branch (639:67): [True: 6.69k, False: 85.9k] ------------------ 640| 326k| hdr->tiling.log2_cols++) ; 641| 326k| const int tile_w = 1 + ((sbw - 1) >> hdr->tiling.log2_cols); 642| 326k| hdr->tiling.cols = 0; 643| 676k| for (int sbx = 0; sbx < sbw; sbx += tile_w, hdr->tiling.cols++) ------------------ | Branch (643:27): [True: 350k, False: 326k] ------------------ 644| 350k| hdr->tiling.col_start_sb[hdr->tiling.cols] = sbx; 645| 326k| hdr->tiling.min_log2_rows = 646| 326k| imax(min_log2_tiles - hdr->tiling.log2_cols, 0); 647| | 648| 326k| for (hdr->tiling.log2_rows = hdr->tiling.min_log2_rows; 649| 347k| hdr->tiling.log2_rows < hdr->tiling.max_log2_rows && dav1d_get_bit(gb); ------------------ | Branch (649:14): [True: 113k, False: 233k] | Branch (649:67): [True: 20.2k, False: 93.3k] ------------------ 650| 326k| hdr->tiling.log2_rows++) ; 651| 326k| const int tile_h = 1 + ((sbh - 1) >> hdr->tiling.log2_rows); 652| 326k| hdr->tiling.rows = 0; 653| 720k| for (int sby = 0; sby < sbh; sby += tile_h, hdr->tiling.rows++) ------------------ | Branch (653:27): [True: 393k, False: 326k] ------------------ 654| 393k| hdr->tiling.row_start_sb[hdr->tiling.rows] = sby; 655| 326k| } else { 656| 54.7k| hdr->tiling.cols = 0; 657| 54.7k| int widest_tile = 0, max_tile_area_sb = sbw * sbh; 658| 204k| for (int sbx = 0; sbx < sbw && hdr->tiling.cols < DAV1D_MAX_TILE_COLS; hdr->tiling.cols++) { ------------------ | | 41| 150k|#define DAV1D_MAX_TILE_COLS 64 ------------------ | Branch (658:27): [True: 150k, False: 54.4k] | Branch (658:40): [True: 149k, False: 320] ------------------ 659| 149k| const int tile_width_sb = imin(sbw - sbx, max_tile_width_sb); 660| 149k| const int tile_w = (tile_width_sb > 1) ? 1 + dav1d_get_uniform(gb, tile_width_sb) : 1; ------------------ | Branch (660:32): [True: 117k, False: 32.3k] ------------------ 661| 149k| hdr->tiling.col_start_sb[hdr->tiling.cols] = sbx; 662| 149k| sbx += tile_w; 663| 149k| widest_tile = imax(widest_tile, tile_w); 664| 149k| } 665| 54.7k| hdr->tiling.log2_cols = tile_log2(1, hdr->tiling.cols); 666| 54.7k| if (min_log2_tiles) max_tile_area_sb >>= min_log2_tiles + 1; ------------------ | Branch (666:13): [True: 1.12k, False: 53.6k] ------------------ 667| 54.7k| const int max_tile_height_sb = imax(max_tile_area_sb / widest_tile, 1); 668| | 669| 54.7k| hdr->tiling.rows = 0; 670| 151k| for (int sby = 0; sby < sbh && hdr->tiling.rows < DAV1D_MAX_TILE_ROWS; hdr->tiling.rows++) { ------------------ | | 42| 97.0k|#define DAV1D_MAX_TILE_ROWS 64 ------------------ | Branch (670:27): [True: 97.0k, False: 54.6k] | Branch (670:40): [True: 96.9k, False: 144] ------------------ 671| 96.9k| const int tile_height_sb = imin(sbh - sby, max_tile_height_sb); 672| 96.9k| const int tile_h = (tile_height_sb > 1) ? 1 + dav1d_get_uniform(gb, tile_height_sb) : 1; ------------------ | Branch (672:32): [True: 52.8k, False: 44.0k] ------------------ 673| 96.9k| hdr->tiling.row_start_sb[hdr->tiling.rows] = sby; 674| 96.9k| sby += tile_h; 675| 96.9k| } 676| 54.7k| hdr->tiling.log2_rows = tile_log2(1, hdr->tiling.rows); 677| 54.7k| } 678| 381k| hdr->tiling.col_start_sb[hdr->tiling.cols] = sbw; 679| 381k| hdr->tiling.row_start_sb[hdr->tiling.rows] = sbh; 680| 381k| if (hdr->tiling.log2_cols || hdr->tiling.log2_rows) { ------------------ | Branch (680:9): [True: 14.4k, False: 367k] | Branch (680:34): [True: 21.6k, False: 345k] ------------------ 681| 36.1k| hdr->tiling.update = dav1d_get_bits(gb, hdr->tiling.log2_cols + hdr->tiling.log2_rows); 682| 36.1k| if (hdr->tiling.update >= hdr->tiling.cols * hdr->tiling.rows) ------------------ | Branch (682:13): [True: 417, False: 35.6k] ------------------ 683| 417| goto error; 684| 35.6k| hdr->tiling.n_bytes = dav1d_get_bits(gb, 2) + 1; 685| 35.6k| } 686| |#if DEBUG_FRAME_HDR 687| | printf("HDR: post-tiling: off=%td\n", 688| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 689| |#endif 690| | 691| | // quant data 692| 381k| hdr->quant.yac = dav1d_get_bits(gb, 8); 693| 381k| if (dav1d_get_bit(gb)) ------------------ | Branch (693:9): [True: 31.1k, False: 350k] ------------------ 694| 31.1k| hdr->quant.ydc_delta = dav1d_get_sbits(gb, 7); 695| 381k| if (!seqhdr->monochrome) { ------------------ | Branch (695:9): [True: 304k, False: 76.9k] ------------------ 696| | // If the sequence header says that delta_q might be different 697| | // for U, V, we must check whether it actually is for this 698| | // frame. 699| 304k| const int diff_uv_delta = seqhdr->separate_uv_delta_q ? dav1d_get_bit(gb) : 0; ------------------ | Branch (699:35): [True: 26.4k, False: 277k] ------------------ 700| 304k| if (dav1d_get_bit(gb)) ------------------ | Branch (700:13): [True: 16.2k, False: 288k] ------------------ 701| 16.2k| hdr->quant.udc_delta = dav1d_get_sbits(gb, 7); 702| 304k| if (dav1d_get_bit(gb)) ------------------ | Branch (702:13): [True: 15.6k, False: 288k] ------------------ 703| 15.6k| hdr->quant.uac_delta = dav1d_get_sbits(gb, 7); 704| 304k| if (diff_uv_delta) { ------------------ | Branch (704:13): [True: 5.53k, False: 298k] ------------------ 705| 5.53k| if (dav1d_get_bit(gb)) ------------------ | Branch (705:17): [True: 1.26k, False: 4.27k] ------------------ 706| 1.26k| hdr->quant.vdc_delta = dav1d_get_sbits(gb, 7); 707| 5.53k| if (dav1d_get_bit(gb)) ------------------ | Branch (707:17): [True: 1.71k, False: 3.82k] ------------------ 708| 1.71k| hdr->quant.vac_delta = dav1d_get_sbits(gb, 7); 709| 298k| } else { 710| 298k| hdr->quant.vdc_delta = hdr->quant.udc_delta; 711| 298k| hdr->quant.vac_delta = hdr->quant.uac_delta; 712| 298k| } 713| 304k| } 714| |#if DEBUG_FRAME_HDR 715| | printf("HDR: post-quant: off=%td\n", 716| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 717| |#endif 718| 381k| hdr->quant.qm = dav1d_get_bit(gb); 719| 381k| if (hdr->quant.qm) { ------------------ | Branch (719:9): [True: 28.6k, False: 352k] ------------------ 720| 28.6k| hdr->quant.qm_y = dav1d_get_bits(gb, 4); 721| 28.6k| hdr->quant.qm_u = dav1d_get_bits(gb, 4); 722| 28.6k| hdr->quant.qm_v = seqhdr->separate_uv_delta_q ? dav1d_get_bits(gb, 4) : ------------------ | Branch (722:27): [True: 12.1k, False: 16.4k] ------------------ 723| 28.6k| hdr->quant.qm_u; 724| 28.6k| } 725| |#if DEBUG_FRAME_HDR 726| | printf("HDR: post-qm: off=%td\n", 727| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 728| |#endif 729| | 730| | // segmentation data 731| 381k| hdr->segmentation.enabled = dav1d_get_bit(gb); 732| 381k| if (hdr->segmentation.enabled) { ------------------ | Branch (732:9): [True: 23.0k, False: 358k] ------------------ 733| 23.0k| if (hdr->primary_ref_frame == DAV1D_PRIMARY_REF_NONE) { ------------------ | | 45| 23.0k|#define DAV1D_PRIMARY_REF_NONE 7 ------------------ | Branch (733:13): [True: 5.56k, False: 17.4k] ------------------ 734| 5.56k| hdr->segmentation.update_map = 1; 735| 5.56k| hdr->segmentation.update_data = 1; 736| 17.4k| } else { 737| 17.4k| hdr->segmentation.update_map = dav1d_get_bit(gb); 738| 17.4k| if (hdr->segmentation.update_map) ------------------ | Branch (738:17): [True: 10.1k, False: 7.36k] ------------------ 739| 10.1k| hdr->segmentation.temporal = dav1d_get_bit(gb); 740| 17.4k| hdr->segmentation.update_data = dav1d_get_bit(gb); 741| 17.4k| } 742| | 743| 23.0k| if (hdr->segmentation.update_data) { ------------------ | Branch (743:13): [True: 7.09k, False: 15.9k] ------------------ 744| 7.09k| hdr->segmentation.seg_data.last_active_segid = -1; 745| 63.8k| for (int i = 0; i < DAV1D_MAX_SEGMENTS; i++) { ------------------ | | 43| 63.8k|#define DAV1D_MAX_SEGMENTS 8 ------------------ | Branch (745:29): [True: 56.7k, False: 7.09k] ------------------ 746| 56.7k| Dav1dSegmentationData *const seg = 747| 56.7k| &hdr->segmentation.seg_data.d[i]; 748| 56.7k| if (dav1d_get_bit(gb)) { ------------------ | Branch (748:21): [True: 13.2k, False: 43.5k] ------------------ 749| 13.2k| seg->delta_q = dav1d_get_sbits(gb, 9); 750| 13.2k| hdr->segmentation.seg_data.last_active_segid = i; 751| 13.2k| } 752| 56.7k| if (dav1d_get_bit(gb)) { ------------------ | Branch (752:21): [True: 10.5k, False: 46.2k] ------------------ 753| 10.5k| seg->delta_lf_y_v = dav1d_get_sbits(gb, 7); 754| 10.5k| hdr->segmentation.seg_data.last_active_segid = i; 755| 10.5k| } 756| 56.7k| if (dav1d_get_bit(gb)) { ------------------ | Branch (756:21): [True: 13.4k, False: 43.3k] ------------------ 757| 13.4k| seg->delta_lf_y_h = dav1d_get_sbits(gb, 7); 758| 13.4k| hdr->segmentation.seg_data.last_active_segid = i; 759| 13.4k| } 760| 56.7k| if (dav1d_get_bit(gb)) { ------------------ | Branch (760:21): [True: 11.9k, False: 44.8k] ------------------ 761| 11.9k| seg->delta_lf_u = dav1d_get_sbits(gb, 7); 762| 11.9k| hdr->segmentation.seg_data.last_active_segid = i; 763| 11.9k| } 764| 56.7k| if (dav1d_get_bit(gb)) { ------------------ | Branch (764:21): [True: 10.6k, False: 46.0k] ------------------ 765| 10.6k| seg->delta_lf_v = dav1d_get_sbits(gb, 7); 766| 10.6k| hdr->segmentation.seg_data.last_active_segid = i; 767| 10.6k| } 768| 56.7k| if (dav1d_get_bit(gb)) { ------------------ | Branch (768:21): [True: 10.8k, False: 45.8k] ------------------ 769| 10.8k| seg->ref = dav1d_get_bits(gb, 3); 770| 10.8k| hdr->segmentation.seg_data.last_active_segid = i; 771| 10.8k| hdr->segmentation.seg_data.preskip = 1; 772| 45.8k| } else { 773| 45.8k| seg->ref = -1; 774| 45.8k| } 775| 56.7k| if ((seg->skip = dav1d_get_bit(gb))) { ------------------ | Branch (775:21): [True: 12.2k, False: 44.4k] ------------------ 776| 12.2k| hdr->segmentation.seg_data.last_active_segid = i; 777| 12.2k| hdr->segmentation.seg_data.preskip = 1; 778| 12.2k| } 779| 56.7k| if ((seg->globalmv = dav1d_get_bit(gb))) { ------------------ | Branch (779:21): [True: 11.9k, False: 44.8k] ------------------ 780| 11.9k| hdr->segmentation.seg_data.last_active_segid = i; 781| 11.9k| hdr->segmentation.seg_data.preskip = 1; 782| 11.9k| } 783| 56.7k| } 784| 15.9k| } else { 785| | // segmentation.update_data was false so we should copy 786| | // segmentation data from the reference frame. 787| 15.9k| assert(hdr->primary_ref_frame != DAV1D_PRIMARY_REF_NONE); ------------------ | Branch (787:13): [True: 15.9k, False: 0] ------------------ 788| 15.9k| const int pri_ref = hdr->refidx[hdr->primary_ref_frame]; 789| 15.9k| if (!c->refs[pri_ref].p.p.frame_hdr) goto error; ------------------ | Branch (789:17): [True: 239, False: 15.6k] ------------------ 790| 15.6k| hdr->segmentation.seg_data = 791| 15.6k| c->refs[pri_ref].p.p.frame_hdr->segmentation.seg_data; 792| 15.6k| } 793| 358k| } else { 794| 3.22M| for (int i = 0; i < DAV1D_MAX_SEGMENTS; i++) ------------------ | | 43| 3.22M|#define DAV1D_MAX_SEGMENTS 8 ------------------ | Branch (794:25): [True: 2.86M, False: 358k] ------------------ 795| 2.86M| hdr->segmentation.seg_data.d[i].ref = -1; 796| 358k| } 797| |#if DEBUG_FRAME_HDR 798| | printf("HDR: post-segmentation: off=%td\n", 799| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 800| |#endif 801| | 802| | // delta q 803| 381k| if (hdr->quant.yac) { ------------------ | Branch (803:9): [True: 339k, False: 41.8k] ------------------ 804| 339k| hdr->delta.q.present = dav1d_get_bit(gb); 805| 339k| if (hdr->delta.q.present) { ------------------ | Branch (805:13): [True: 81.0k, False: 258k] ------------------ 806| 81.0k| hdr->delta.q.res_log2 = dav1d_get_bits(gb, 2); 807| 81.0k| if (!hdr->allow_intrabc) { ------------------ | Branch (807:17): [True: 16.5k, False: 64.4k] ------------------ 808| 16.5k| hdr->delta.lf.present = dav1d_get_bit(gb); 809| 16.5k| if (hdr->delta.lf.present) { ------------------ | Branch (809:21): [True: 8.21k, False: 8.37k] ------------------ 810| 8.21k| hdr->delta.lf.res_log2 = dav1d_get_bits(gb, 2); 811| 8.21k| hdr->delta.lf.multi = dav1d_get_bit(gb); 812| 8.21k| } 813| 16.5k| } 814| 81.0k| } 815| 339k| } 816| |#if DEBUG_FRAME_HDR 817| | printf("HDR: post-delta_q_lf_flags: off=%td\n", 818| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 819| |#endif 820| | 821| | // derive lossless flags 822| 381k| const int delta_lossless = !hdr->quant.ydc_delta && !hdr->quant.udc_delta && ------------------ | Branch (822:32): [True: 351k, False: 29.7k] | Branch (822:57): [True: 344k, False: 6.30k] ------------------ 823| 344k| !hdr->quant.uac_delta && !hdr->quant.vdc_delta && !hdr->quant.vac_delta; ------------------ | Branch (823:9): [True: 338k, False: 6.10k] | Branch (823:34): [True: 338k, False: 552] | Branch (823:59): [True: 338k, False: 17] ------------------ 824| 381k| hdr->all_lossless = 1; 825| 3.42M| for (int i = 0; i < DAV1D_MAX_SEGMENTS; i++) { ------------------ | | 43| 3.42M|#define DAV1D_MAX_SEGMENTS 8 ------------------ | Branch (825:21): [True: 3.04M, False: 381k] ------------------ 826| 3.04M| hdr->segmentation.qidx[i] = hdr->segmentation.enabled ? ------------------ | Branch (826:37): [True: 182k, False: 2.86M] ------------------ 827| 182k| iclip_u8(hdr->quant.yac + hdr->segmentation.seg_data.d[i].delta_q) : 828| 3.04M| hdr->quant.yac; 829| 3.04M| hdr->segmentation.lossless[i] = 830| 3.04M| !hdr->segmentation.qidx[i] && delta_lossless; ------------------ | Branch (830:13): [True: 340k, False: 2.70M] | Branch (830:43): [True: 305k, False: 34.9k] ------------------ 831| 3.04M| hdr->all_lossless &= hdr->segmentation.lossless[i]; 832| 3.04M| } 833| | 834| | // loopfilter 835| 381k| if (hdr->all_lossless || hdr->allow_intrabc) { ------------------ | Branch (835:9): [True: 37.8k, False: 343k] | Branch (835:30): [True: 231k, False: 111k] ------------------ 836| 269k| hdr->loopfilter.mode_ref_delta_enabled = 1; 837| 269k| hdr->loopfilter.mode_ref_delta_update = 1; 838| 269k| hdr->loopfilter.mode_ref_deltas = default_mode_ref_deltas; 839| 269k| } else { 840| 111k| hdr->loopfilter.level_y[0] = dav1d_get_bits(gb, 6); 841| 111k| hdr->loopfilter.level_y[1] = dav1d_get_bits(gb, 6); 842| 111k| if (!seqhdr->monochrome && ------------------ | Branch (842:13): [True: 44.4k, False: 67.0k] ------------------ 843| 44.4k| (hdr->loopfilter.level_y[0] || hdr->loopfilter.level_y[1])) ------------------ | Branch (843:14): [True: 19.4k, False: 25.0k] | Branch (843:44): [True: 9.63k, False: 15.4k] ------------------ 844| 29.0k| { 845| 29.0k| hdr->loopfilter.level_u = dav1d_get_bits(gb, 6); 846| 29.0k| hdr->loopfilter.level_v = dav1d_get_bits(gb, 6); 847| 29.0k| } 848| 111k| hdr->loopfilter.sharpness = dav1d_get_bits(gb, 3); 849| | 850| 111k| if (hdr->primary_ref_frame == DAV1D_PRIMARY_REF_NONE) { ------------------ | | 45| 111k|#define DAV1D_PRIMARY_REF_NONE 7 ------------------ | Branch (850:13): [True: 49.3k, False: 62.1k] ------------------ 851| 49.3k| hdr->loopfilter.mode_ref_deltas = default_mode_ref_deltas; 852| 62.1k| } else { 853| 62.1k| const int ref = hdr->refidx[hdr->primary_ref_frame]; 854| 62.1k| if (!c->refs[ref].p.p.frame_hdr) goto error; ------------------ | Branch (854:17): [True: 438, False: 61.7k] ------------------ 855| 61.7k| hdr->loopfilter.mode_ref_deltas = 856| 61.7k| c->refs[ref].p.p.frame_hdr->loopfilter.mode_ref_deltas; 857| 61.7k| } 858| 111k| hdr->loopfilter.mode_ref_delta_enabled = dav1d_get_bit(gb); 859| 111k| if (hdr->loopfilter.mode_ref_delta_enabled) { ------------------ | Branch (859:13): [True: 65.1k, False: 45.9k] ------------------ 860| 65.1k| hdr->loopfilter.mode_ref_delta_update = dav1d_get_bit(gb); 861| 65.1k| if (hdr->loopfilter.mode_ref_delta_update) { ------------------ | Branch (861:17): [True: 8.56k, False: 56.5k] ------------------ 862| 77.0k| for (int i = 0; i < 8; i++) ------------------ | Branch (862:33): [True: 68.4k, False: 8.56k] ------------------ 863| 68.4k| if (dav1d_get_bit(gb)) ------------------ | Branch (863:25): [True: 24.1k, False: 44.3k] ------------------ 864| 24.1k| hdr->loopfilter.mode_ref_deltas.ref_delta[i] = 865| 24.1k| dav1d_get_sbits(gb, 7); 866| 25.6k| for (int i = 0; i < 2; i++) ------------------ | Branch (866:33): [True: 17.1k, False: 8.56k] ------------------ 867| 17.1k| if (dav1d_get_bit(gb)) ------------------ | Branch (867:25): [True: 3.79k, False: 13.3k] ------------------ 868| 3.79k| hdr->loopfilter.mode_ref_deltas.mode_delta[i] = 869| 3.79k| dav1d_get_sbits(gb, 7); 870| 8.56k| } 871| 65.1k| } 872| 111k| } 873| |#if DEBUG_FRAME_HDR 874| | printf("HDR: post-lpf: off=%td\n", 875| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 876| |#endif 877| | 878| | // cdef 879| 380k| if (!hdr->all_lossless && seqhdr->cdef && !hdr->allow_intrabc) { ------------------ | Branch (879:9): [True: 342k, False: 37.8k] | Branch (879:31): [True: 229k, False: 113k] | Branch (879:47): [True: 64.5k, False: 165k] ------------------ 880| 64.5k| hdr->cdef.damping = dav1d_get_bits(gb, 2) + 3; 881| 64.5k| hdr->cdef.n_bits = dav1d_get_bits(gb, 2); 882| 168k| for (int i = 0; i < (1 << hdr->cdef.n_bits); i++) { ------------------ | Branch (882:25): [True: 103k, False: 64.5k] ------------------ 883| 103k| hdr->cdef.y_strength[i] = dav1d_get_bits(gb, 6); 884| 103k| if (!seqhdr->monochrome) ------------------ | Branch (884:17): [True: 46.8k, False: 57.0k] ------------------ 885| 46.8k| hdr->cdef.uv_strength[i] = dav1d_get_bits(gb, 6); 886| 103k| } 887| 64.5k| } 888| |#if DEBUG_FRAME_HDR 889| | printf("HDR: post-cdef: off=%td\n", 890| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 891| |#endif 892| | 893| | // restoration 894| 380k| if ((!hdr->all_lossless || hdr->super_res.enabled) && ------------------ | Branch (894:10): [True: 342k, False: 37.8k] | Branch (894:32): [True: 25.0k, False: 12.7k] ------------------ 895| 367k| seqhdr->restoration && !hdr->allow_intrabc) ------------------ | Branch (895:9): [True: 105k, False: 262k] | Branch (895:32): [True: 90.9k, False: 14.4k] ------------------ 896| 90.9k| { 897| 90.9k| hdr->restoration.type[0] = dav1d_get_bits(gb, 2); 898| 90.9k| if (!seqhdr->monochrome) { ------------------ | Branch (898:13): [True: 37.7k, False: 53.2k] ------------------ 899| 37.7k| hdr->restoration.type[1] = dav1d_get_bits(gb, 2); 900| 37.7k| hdr->restoration.type[2] = dav1d_get_bits(gb, 2); 901| 37.7k| } 902| | 903| 90.9k| if (hdr->restoration.type[0] || hdr->restoration.type[1] || ------------------ | Branch (903:13): [True: 55.1k, False: 35.8k] | Branch (903:41): [True: 3.81k, False: 32.0k] ------------------ 904| 32.0k| hdr->restoration.type[2]) ------------------ | Branch (904:13): [True: 944, False: 31.1k] ------------------ 905| 59.8k| { 906| | // Log2 of the restoration unit size. 907| 59.8k| hdr->restoration.unit_size[0] = 6 + seqhdr->sb128; 908| 59.8k| if (dav1d_get_bit(gb)) { ------------------ | Branch (908:17): [True: 11.2k, False: 48.6k] ------------------ 909| 11.2k| hdr->restoration.unit_size[0]++; 910| 11.2k| if (!seqhdr->sb128) ------------------ | Branch (910:21): [True: 3.01k, False: 8.19k] ------------------ 911| 3.01k| hdr->restoration.unit_size[0] += dav1d_get_bit(gb); 912| 11.2k| } 913| 59.8k| hdr->restoration.unit_size[1] = hdr->restoration.unit_size[0]; 914| 59.8k| if ((hdr->restoration.type[1] || hdr->restoration.type[2]) && ------------------ | Branch (914:18): [True: 10.9k, False: 48.9k] | Branch (914:46): [True: 4.46k, False: 44.4k] ------------------ 915| 15.4k| seqhdr->ss_hor == 1 && seqhdr->ss_ver == 1) ------------------ | Branch (915:17): [True: 10.3k, False: 5.07k] | Branch (915:40): [True: 9.55k, False: 774] ------------------ 916| 9.55k| { 917| 9.55k| hdr->restoration.unit_size[1] -= dav1d_get_bit(gb); 918| 9.55k| } 919| 59.8k| } else { 920| 31.1k| hdr->restoration.unit_size[0] = 8; 921| 31.1k| } 922| 90.9k| } 923| |#if DEBUG_FRAME_HDR 924| | printf("HDR: post-restoration: off=%td\n", 925| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 926| |#endif 927| | 928| 380k| if (!hdr->all_lossless) ------------------ | Branch (928:9): [True: 342k, False: 37.8k] ------------------ 929| 342k| hdr->txfm_mode = dav1d_get_bit(gb) ? DAV1D_TX_SWITCHABLE : DAV1D_TX_LARGEST; ------------------ | Branch (929:26): [True: 192k, False: 149k] ------------------ 930| |#if DEBUG_FRAME_HDR 931| | printf("HDR: post-txfmmode: off=%td\n", 932| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 933| |#endif 934| 380k| if (IS_INTER_OR_SWITCH(hdr)) ------------------ | | 36| 380k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 110k, False: 270k] | | ------------------ ------------------ 935| 110k| hdr->switchable_comp_refs = dav1d_get_bit(gb); 936| |#if DEBUG_FRAME_HDR 937| | printf("HDR: post-refmode: off=%td\n", 938| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 939| |#endif 940| 380k| if (hdr->switchable_comp_refs && IS_INTER_OR_SWITCH(hdr) && seqhdr->order_hint) { ------------------ | | 36| 437k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 56.5k, False: 0] | | ------------------ ------------------ | Branch (940:9): [True: 56.5k, False: 324k] | Branch (940:65): [True: 54.8k, False: 1.75k] ------------------ 941| 54.8k| const int poc = hdr->frame_offset; 942| 54.8k| int off_before = -1, off_after = -1; 943| 54.8k| int off_before_idx, off_after_idx; 944| 437k| for (int i = 0; i < 7; i++) { ------------------ | Branch (944:25): [True: 383k, False: 54.7k] ------------------ 945| 383k| if (!c->refs[hdr->refidx[i]].p.p.frame_hdr) goto error; ------------------ | Branch (945:17): [True: 85, False: 383k] ------------------ 946| 383k| const int refpoc = c->refs[hdr->refidx[i]].p.p.frame_hdr->frame_offset; 947| | 948| 383k| const int diff = get_poc_diff(seqhdr->order_hint_n_bits, refpoc, poc); 949| 383k| if (diff > 0) { ------------------ | Branch (949:17): [True: 156k, False: 226k] ------------------ 950| 156k| if (off_after < 0 || get_poc_diff(seqhdr->order_hint_n_bits, ------------------ | Branch (950:21): [True: 45.1k, False: 111k] | Branch (950:38): [True: 9.00k, False: 102k] ------------------ 951| 111k| off_after, refpoc) > 0) 952| 54.1k| { 953| 54.1k| off_after = refpoc; 954| 54.1k| off_after_idx = i; 955| 54.1k| } 956| 226k| } else if (diff < 0 && (off_before < 0 || ------------------ | Branch (956:24): [True: 169k, False: 57.0k] | Branch (956:37): [True: 49.6k, False: 119k] ------------------ 957| 119k| get_poc_diff(seqhdr->order_hint_n_bits, ------------------ | Branch (957:37): [True: 6.87k, False: 112k] ------------------ 958| 119k| refpoc, off_before) > 0)) 959| 56.5k| { 960| 56.5k| off_before = refpoc; 961| 56.5k| off_before_idx = i; 962| 56.5k| } 963| 383k| } 964| | 965| 54.7k| if ((off_before | off_after) >= 0) { ------------------ | Branch (965:13): [True: 41.2k, False: 13.5k] ------------------ 966| 41.2k| hdr->skip_mode_refs[0] = imin(off_before_idx, off_after_idx); 967| 41.2k| hdr->skip_mode_refs[1] = imax(off_before_idx, off_after_idx); 968| 41.2k| hdr->skip_mode_allowed = 1; 969| 41.2k| } else if (off_before >= 0) { ------------------ | Branch (969:20): [True: 8.46k, False: 5.05k] ------------------ 970| 8.46k| int off_before2 = -1; 971| 8.46k| int off_before2_idx; 972| 67.6k| for (int i = 0; i < 7; i++) { ------------------ | Branch (972:29): [True: 59.2k, False: 8.46k] ------------------ 973| 59.2k| if (!c->refs[hdr->refidx[i]].p.p.frame_hdr) goto error; ------------------ | Branch (973:21): [True: 0, False: 59.2k] ------------------ 974| 59.2k| const int refpoc = c->refs[hdr->refidx[i]].p.p.frame_hdr->frame_offset; 975| 59.2k| if (get_poc_diff(seqhdr->order_hint_n_bits, ------------------ | Branch (975:21): [True: 16.0k, False: 43.1k] ------------------ 976| 59.2k| refpoc, off_before) < 0) { 977| 16.0k| if (off_before2 < 0 || get_poc_diff(seqhdr->order_hint_n_bits, ------------------ | Branch (977:25): [True: 4.56k, False: 11.4k] | Branch (977:44): [True: 483, False: 10.9k] ------------------ 978| 11.4k| refpoc, off_before2) > 0) 979| 5.04k| { 980| 5.04k| off_before2 = refpoc; 981| 5.04k| off_before2_idx = i; 982| 5.04k| } 983| 16.0k| } 984| 59.2k| } 985| | 986| 8.46k| if (off_before2 >= 0) { ------------------ | Branch (986:17): [True: 4.56k, False: 3.89k] ------------------ 987| 4.56k| hdr->skip_mode_refs[0] = imin(off_before_idx, off_before2_idx); 988| 4.56k| hdr->skip_mode_refs[1] = imax(off_before_idx, off_before2_idx); 989| 4.56k| hdr->skip_mode_allowed = 1; 990| 4.56k| } 991| 8.46k| } 992| 54.7k| } 993| 380k| if (hdr->skip_mode_allowed) ------------------ | Branch (993:9): [True: 45.7k, False: 334k] ------------------ 994| 45.7k| hdr->skip_mode_enabled = dav1d_get_bit(gb); 995| |#if DEBUG_FRAME_HDR 996| | printf("HDR: post-extskip: off=%td\n", 997| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 998| |#endif 999| 380k| if (!hdr->error_resilient_mode && IS_INTER_OR_SWITCH(hdr) && seqhdr->warped_motion) ------------------ | | 36| 491k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 108k, False: 2.42k] | | ------------------ ------------------ | Branch (999:9): [True: 110k, False: 269k] | Branch (999:66): [True: 89.9k, False: 18.2k] ------------------ 1000| 89.9k| hdr->warp_motion = dav1d_get_bit(gb); 1001| |#if DEBUG_FRAME_HDR 1002| | printf("HDR: post-warpmotionbit: off=%td\n", 1003| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 1004| |#endif 1005| 380k| hdr->reduced_txtp_set = dav1d_get_bit(gb); 1006| |#if DEBUG_FRAME_HDR 1007| | printf("HDR: post-reducedtxtpset: off=%td\n", 1008| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 1009| |#endif 1010| | 1011| 3.04M| for (int i = 0; i < 7; i++) ------------------ | Branch (1011:21): [True: 2.66M, False: 380k] ------------------ 1012| 2.66M| hdr->gmv[i] = dav1d_default_wm_params; 1013| | 1014| 380k| if (IS_INTER_OR_SWITCH(hdr)) { ------------------ | | 36| 380k| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (36:5): [True: 110k, False: 270k] | | ------------------ ------------------ 1015| 881k| for (int i = 0; i < 7; i++) { ------------------ | Branch (1015:25): [True: 771k, False: 109k] ------------------ 1016| 771k| hdr->gmv[i].type = !dav1d_get_bit(gb) ? DAV1D_WM_TYPE_IDENTITY : ------------------ | Branch (1016:32): [True: 731k, False: 39.8k] ------------------ 1017| 771k| dav1d_get_bit(gb) ? DAV1D_WM_TYPE_ROT_ZOOM : ------------------ | Branch (1017:33): [True: 22.4k, False: 17.4k] ------------------ 1018| 39.8k| dav1d_get_bit(gb) ? DAV1D_WM_TYPE_TRANSLATION : ------------------ | Branch (1018:33): [True: 3.45k, False: 13.9k] ------------------ 1019| 17.4k| DAV1D_WM_TYPE_AFFINE; 1020| | 1021| 771k| if (hdr->gmv[i].type == DAV1D_WM_TYPE_IDENTITY) continue; ------------------ | Branch (1021:17): [True: 731k, False: 39.8k] ------------------ 1022| | 1023| 39.8k| const Dav1dWarpedMotionParams *ref_gmv; 1024| 39.8k| if (hdr->primary_ref_frame == DAV1D_PRIMARY_REF_NONE) { ------------------ | | 45| 39.8k|#define DAV1D_PRIMARY_REF_NONE 7 ------------------ | Branch (1024:17): [True: 2.69k, False: 37.1k] ------------------ 1025| 2.69k| ref_gmv = &dav1d_default_wm_params; 1026| 37.1k| } else { 1027| 37.1k| const int pri_ref = hdr->refidx[hdr->primary_ref_frame]; 1028| 37.1k| if (!c->refs[pri_ref].p.p.frame_hdr) goto error; ------------------ | Branch (1028:21): [True: 488, False: 36.6k] ------------------ 1029| 36.6k| ref_gmv = &c->refs[pri_ref].p.p.frame_hdr->gmv[i]; 1030| 36.6k| } 1031| 39.3k| int32_t *const mat = hdr->gmv[i].matrix; 1032| 39.3k| const int32_t *const ref_mat = ref_gmv->matrix; 1033| 39.3k| int bits, shift; 1034| | 1035| 39.3k| if (hdr->gmv[i].type >= DAV1D_WM_TYPE_ROT_ZOOM) { ------------------ | Branch (1035:17): [True: 36.0k, False: 3.35k] ------------------ 1036| 36.0k| mat[2] = (1 << 16) + 2 * 1037| 36.0k| dav1d_get_bits_subexp(gb, (ref_mat[2] - (1 << 16)) >> 1, 12); 1038| 36.0k| mat[3] = 2 * dav1d_get_bits_subexp(gb, ref_mat[3] >> 1, 12); 1039| | 1040| 36.0k| bits = 12; 1041| 36.0k| shift = 10; 1042| 36.0k| } else { 1043| 3.35k| bits = 9 - !hdr->hp; 1044| 3.35k| shift = 13 + !hdr->hp; 1045| 3.35k| } 1046| | 1047| 39.3k| if (hdr->gmv[i].type == DAV1D_WM_TYPE_AFFINE) { ------------------ | Branch (1047:17): [True: 13.9k, False: 25.4k] ------------------ 1048| 13.9k| mat[4] = 2 * dav1d_get_bits_subexp(gb, ref_mat[4] >> 1, 12); 1049| 13.9k| mat[5] = (1 << 16) + 2 * 1050| 13.9k| dav1d_get_bits_subexp(gb, (ref_mat[5] - (1 << 16)) >> 1, 12); 1051| 25.4k| } else { 1052| 25.4k| mat[4] = -mat[3]; 1053| 25.4k| mat[5] = mat[2]; 1054| 25.4k| } 1055| | 1056| 39.3k| mat[0] = dav1d_get_bits_subexp(gb, ref_mat[0] >> shift, bits) * (1 << shift); 1057| 39.3k| mat[1] = dav1d_get_bits_subexp(gb, ref_mat[1] >> shift, bits) * (1 << shift); 1058| 39.3k| } 1059| 110k| } 1060| |#if DEBUG_FRAME_HDR 1061| | printf("HDR: post-gmv: off=%td\n", 1062| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 1063| |#endif 1064| | 1065| 380k| if (seqhdr->film_grain_present && (hdr->show_frame || hdr->showable_frame)) { ------------------ | Branch (1065:9): [True: 292k, False: 87.5k] | Branch (1065:40): [True: 247k, False: 45.0k] | Branch (1065:59): [True: 1.36k, False: 43.6k] ------------------ 1066| 248k| hdr->film_grain.present = dav1d_get_bit(gb); 1067| 248k| if (hdr->film_grain.present) { ------------------ | Branch (1067:13): [True: 25.9k, False: 222k] ------------------ 1068| 25.9k| const unsigned seed = dav1d_get_bits(gb, 16); 1069| 25.9k| hdr->film_grain.update = hdr->frame_type != DAV1D_FRAME_TYPE_INTER || dav1d_get_bit(gb); ------------------ | Branch (1069:38): [True: 6.95k, False: 19.0k] | Branch (1069:83): [True: 747, False: 18.2k] ------------------ 1070| 25.9k| if (!hdr->film_grain.update) { ------------------ | Branch (1070:17): [True: 18.2k, False: 7.70k] ------------------ 1071| 18.2k| const int refidx = dav1d_get_bits(gb, 3); 1072| 18.2k| int i; 1073| 45.0k| for (i = 0; i < 7; i++) ------------------ | Branch (1073:29): [True: 44.8k, False: 254] ------------------ 1074| 44.8k| if (hdr->refidx[i] == refidx) ------------------ | Branch (1074:25): [True: 18.0k, False: 26.8k] ------------------ 1075| 18.0k| break; 1076| 18.2k| if (i == 7 || !c->refs[refidx].p.p.frame_hdr) goto error; ------------------ | Branch (1076:21): [True: 254, False: 18.0k] | Branch (1076:31): [True: 73, False: 17.9k] ------------------ 1077| 17.9k| hdr->film_grain.data = c->refs[refidx].p.p.frame_hdr->film_grain.data; 1078| 17.9k| hdr->film_grain.data.seed = seed; 1079| 17.9k| } else { 1080| 7.70k| Dav1dFilmGrainData *const fgd = &hdr->film_grain.data; 1081| 7.70k| fgd->seed = seed; 1082| | 1083| 7.70k| fgd->num_y_points = dav1d_get_bits(gb, 4); 1084| 7.70k| if (fgd->num_y_points > 14) goto error; ------------------ | Branch (1084:21): [True: 301, False: 7.40k] ------------------ 1085| 11.5k| for (int i = 0; i < fgd->num_y_points; i++) { ------------------ | Branch (1085:33): [True: 4.98k, False: 6.61k] ------------------ 1086| 4.98k| fgd->y_points[i][0] = dav1d_get_bits(gb, 8); 1087| 4.98k| if (i && fgd->y_points[i - 1][0] >= fgd->y_points[i][0]) ------------------ | Branch (1087:25): [True: 2.64k, False: 2.34k] | Branch (1087:30): [True: 791, False: 1.84k] ------------------ 1088| 791| goto error; 1089| 4.19k| fgd->y_points[i][1] = dav1d_get_bits(gb, 8); 1090| 4.19k| } 1091| | 1092| 6.61k| if (!seqhdr->monochrome) ------------------ | Branch (1092:21): [True: 5.76k, False: 847] ------------------ 1093| 5.76k| fgd->chroma_scaling_from_luma = dav1d_get_bit(gb); 1094| 6.61k| if (seqhdr->monochrome || fgd->chroma_scaling_from_luma || ------------------ | Branch (1094:21): [True: 847, False: 5.76k] | Branch (1094:43): [True: 2.38k, False: 3.37k] ------------------ 1095| 3.37k| (seqhdr->ss_ver == 1 && seqhdr->ss_hor == 1 && !fgd->num_y_points)) ------------------ | Branch (1095:22): [True: 755, False: 2.62k] | Branch (1095:45): [True: 755, False: 0] | Branch (1095:68): [True: 213, False: 542] ------------------ 1096| 3.44k| { 1097| 3.44k| fgd->num_uv_points[0] = fgd->num_uv_points[1] = 0; 1098| 8.64k| } else for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1098:41): [True: 5.97k, False: 2.66k] ------------------ 1099| 5.97k| fgd->num_uv_points[pl] = dav1d_get_bits(gb, 4); 1100| 5.97k| if (fgd->num_uv_points[pl] > 10) goto error; ------------------ | Branch (1100:25): [True: 85, False: 5.89k] ------------------ 1101| 10.8k| for (int i = 0; i < fgd->num_uv_points[pl]; i++) { ------------------ | Branch (1101:37): [True: 5.41k, False: 5.47k] ------------------ 1102| 5.41k| fgd->uv_points[pl][i][0] = dav1d_get_bits(gb, 8); 1103| 5.41k| if (i && fgd->uv_points[pl][i - 1][0] >= fgd->uv_points[pl][i][0]) ------------------ | Branch (1103:29): [True: 3.35k, False: 2.06k] | Branch (1103:34): [True: 418, False: 2.93k] ------------------ 1104| 418| goto error; 1105| 4.99k| fgd->uv_points[pl][i][1] = dav1d_get_bits(gb, 8); 1106| 4.99k| } 1107| 5.89k| } 1108| | 1109| 6.10k| if (seqhdr->ss_hor == 1 && seqhdr->ss_ver == 1 && ------------------ | Branch (1109:21): [True: 4.33k, False: 1.76k] | Branch (1109:44): [True: 3.62k, False: 717] ------------------ 1110| 3.62k| !!fgd->num_uv_points[0] != !!fgd->num_uv_points[1]) ------------------ | Branch (1110:21): [True: 70, False: 3.55k] ------------------ 1111| 70| { 1112| 70| goto error; 1113| 70| } 1114| | 1115| 6.03k| fgd->scaling_shift = dav1d_get_bits(gb, 2) + 8; 1116| 6.03k| fgd->ar_coeff_lag = dav1d_get_bits(gb, 2); 1117| 6.03k| const int num_y_pos = 2 * fgd->ar_coeff_lag * (fgd->ar_coeff_lag + 1); 1118| 6.03k| if (fgd->num_y_points) ------------------ | Branch (1118:21): [True: 1.42k, False: 4.61k] ------------------ 1119| 21.0k| for (int i = 0; i < num_y_pos; i++) ------------------ | Branch (1119:37): [True: 19.6k, False: 1.42k] ------------------ 1120| 19.6k| fgd->ar_coeffs_y[i] = dav1d_get_bits(gb, 8) - 128; 1121| 18.1k| for (int pl = 0; pl < 2; pl++) ------------------ | Branch (1121:34): [True: 12.0k, False: 6.03k] ------------------ 1122| 12.0k| if (fgd->num_uv_points[pl] || fgd->chroma_scaling_from_luma) { ------------------ | Branch (1122:25): [True: 1.56k, False: 10.5k] | Branch (1122:51): [True: 4.76k, False: 5.73k] ------------------ 1123| 6.33k| const int num_uv_pos = num_y_pos + !!fgd->num_y_points; 1124| 28.4k| for (int i = 0; i < num_uv_pos; i++) ------------------ | Branch (1124:41): [True: 22.1k, False: 6.33k] ------------------ 1125| 22.1k| fgd->ar_coeffs_uv[pl][i] = dav1d_get_bits(gb, 8) - 128; 1126| 6.33k| if (!fgd->num_y_points) ------------------ | Branch (1126:29): [True: 5.00k, False: 1.32k] ------------------ 1127| 5.00k| fgd->ar_coeffs_uv[pl][num_uv_pos] = 0; 1128| 6.33k| } 1129| 6.03k| fgd->ar_coeff_shift = dav1d_get_bits(gb, 2) + 6; 1130| 6.03k| fgd->grain_scale_shift = dav1d_get_bits(gb, 2); 1131| 18.1k| for (int pl = 0; pl < 2; pl++) ------------------ | Branch (1131:34): [True: 12.0k, False: 6.03k] ------------------ 1132| 12.0k| if (fgd->num_uv_points[pl]) { ------------------ | Branch (1132:25): [True: 1.56k, False: 10.5k] ------------------ 1133| 1.56k| fgd->uv_mult[pl] = dav1d_get_bits(gb, 8) - 128; 1134| 1.56k| fgd->uv_luma_mult[pl] = dav1d_get_bits(gb, 8) - 128; 1135| 1.56k| fgd->uv_offset[pl] = dav1d_get_bits(gb, 9) - 256; 1136| 1.56k| } 1137| 6.03k| fgd->overlap_flag = dav1d_get_bit(gb); 1138| 6.03k| fgd->clip_to_restricted_range = dav1d_get_bit(gb); 1139| 6.03k| } 1140| 25.9k| } 1141| 248k| } 1142| |#if DEBUG_FRAME_HDR 1143| | printf("HDR: post-filmgrain: off=%td\n", 1144| | (gb->ptr - init_ptr) * 8 - gb->bits_left); 1145| |#endif 1146| | 1147| 378k| return 0; 1148| | 1149| 7.38k|error: 1150| 7.38k| dav1d_log(c, "Error parsing frame header\n"); ------------------ | | 44| 7.38k|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 7.38k] | | ------------------ ------------------ 1151| 7.38k| return DAV1D_ERR(EINVAL); ------------------ | | 58| 7.38k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 1152| 380k|} obu.c:read_frame_size: 343| 381k|{ 344| 381k| const Dav1dSequenceHeader *const seqhdr = c->seq_hdr; 345| 381k| Dav1dFrameHeader *const hdr = c->frame_hdr; 346| | 347| 381k| if (use_ref) { ------------------ | Branch (347:9): [True: 62.0k, False: 319k] ------------------ 348| 111k| for (int i = 0; i < 7; i++) { ------------------ | Branch (348:25): [True: 110k, False: 807] ------------------ 349| 110k| if (dav1d_get_bit(gb)) { ------------------ | Branch (349:17): [True: 61.2k, False: 49.6k] ------------------ 350| 61.2k| const Dav1dThreadPicture *const ref = 351| 61.2k| &c->refs[c->frame_hdr->refidx[i]].p; 352| 61.2k| if (!ref->p.frame_hdr) return -1; ------------------ | Branch (352:21): [True: 259, False: 61.0k] ------------------ 353| 61.0k| hdr->width[1] = ref->p.frame_hdr->width[1]; 354| 61.0k| hdr->height = ref->p.frame_hdr->height; 355| 61.0k| hdr->render_width = ref->p.frame_hdr->render_width; 356| 61.0k| hdr->render_height = ref->p.frame_hdr->render_height; 357| 61.0k| hdr->super_res.enabled = seqhdr->super_res && dav1d_get_bit(gb); ------------------ | Branch (357:42): [True: 6.34k, False: 54.6k] | Branch (357:63): [True: 3.59k, False: 2.75k] ------------------ 358| 61.0k| if (hdr->super_res.enabled) { ------------------ | Branch (358:21): [True: 3.59k, False: 57.4k] ------------------ 359| 3.59k| const int d = hdr->super_res.width_scale_denominator = 360| 3.59k| 9 + dav1d_get_bits(gb, 3); 361| 3.59k| hdr->width[0] = imax((hdr->width[1] * 8 + (d >> 1)) / d, 362| 3.59k| imin(16, hdr->width[1])); 363| 57.4k| } else { 364| 57.4k| hdr->super_res.width_scale_denominator = 8; 365| 57.4k| hdr->width[0] = hdr->width[1]; 366| 57.4k| } 367| 61.0k| return 0; 368| 61.2k| } 369| 110k| } 370| 62.0k| } 371| | 372| 320k| if (hdr->frame_size_override) { ------------------ | Branch (372:9): [True: 9.45k, False: 311k] ------------------ 373| 9.45k| hdr->width[1] = dav1d_get_bits(gb, seqhdr->width_n_bits) + 1; 374| 9.45k| hdr->height = dav1d_get_bits(gb, seqhdr->height_n_bits) + 1; 375| 311k| } else { 376| 311k| hdr->width[1] = seqhdr->max_width; 377| 311k| hdr->height = seqhdr->max_height; 378| 311k| } 379| 320k| hdr->super_res.enabled = seqhdr->super_res && dav1d_get_bit(gb); ------------------ | Branch (379:30): [True: 287k, False: 33.6k] | Branch (379:51): [True: 48.2k, False: 238k] ------------------ 380| 320k| if (hdr->super_res.enabled) { ------------------ | Branch (380:9): [True: 48.2k, False: 272k] ------------------ 381| 48.2k| const int d = hdr->super_res.width_scale_denominator = 9 + dav1d_get_bits(gb, 3); 382| 48.2k| hdr->width[0] = imax((hdr->width[1] * 8 + (d >> 1)) / d, imin(16, hdr->width[1])); 383| 272k| } else { 384| 272k| hdr->super_res.width_scale_denominator = 8; 385| 272k| hdr->width[0] = hdr->width[1]; 386| 272k| } 387| 320k| hdr->have_render_size = dav1d_get_bit(gb); 388| 320k| if (hdr->have_render_size) { ------------------ | Branch (388:9): [True: 8.69k, False: 311k] ------------------ 389| 8.69k| hdr->render_width = dav1d_get_bits(gb, 16) + 1; 390| 8.69k| hdr->render_height = dav1d_get_bits(gb, 16) + 1; 391| 311k| } else { 392| 311k| hdr->render_width = hdr->width[1]; 393| 311k| hdr->render_height = hdr->height; 394| 311k| } 395| 320k| return 0; 396| 381k|} obu.c:tile_log2: 398| 1.63M|static inline int tile_log2(const int sz, const int tgt) { 399| 1.63M| int k; 400| 2.38M| for (k = 0; (sz << k) < tgt; k++) ; ------------------ | Branch (400:17): [True: 745k, False: 1.63M] ------------------ 401| 1.63M| return k; 402| 1.63M|} obu.c:check_trailing_bits: 50| 75.4k|{ 51| 75.4k| const int trailing_one_bit = dav1d_get_bit(gb); 52| | 53| 75.4k| if (gb->error) ------------------ | Branch (53:9): [True: 3.75k, False: 71.7k] ------------------ 54| 3.75k| return DAV1D_ERR(EINVAL); ------------------ | | 58| 3.75k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 55| | 56| 71.7k| if (!strict_std_compliance) ------------------ | Branch (56:9): [True: 71.7k, False: 0] ------------------ 57| 71.7k| return 0; 58| | 59| 0| if (!trailing_one_bit || gb->state) ------------------ | Branch (59:9): [True: 0, False: 0] | Branch (59:30): [True: 0, False: 0] ------------------ 60| 0| return DAV1D_ERR(EINVAL); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 61| | 62| 0| ptrdiff_t size = gb->ptr_end - gb->ptr; 63| 0| while (size > 0 && gb->ptr[size - 1] == 0) ------------------ | Branch (63:12): [True: 0, False: 0] | Branch (63:24): [True: 0, False: 0] ------------------ 64| 0| size--; 65| | 66| 0| if (size) ------------------ | Branch (66:9): [True: 0, False: 0] ------------------ 67| 0| return DAV1D_ERR(EINVAL); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 68| | 69| 0| return 0; 70| 0|} obu.c:parse_tile_hdr: 1154| 374k|static void parse_tile_hdr(Dav1dContext *const c, GetBits *const gb) { 1155| 374k| const int n_tiles = c->frame_hdr->tiling.cols * c->frame_hdr->tiling.rows; 1156| 374k| const int have_tile_pos = n_tiles > 1 ? dav1d_get_bit(gb) : 0; ------------------ | Branch (1156:31): [True: 34.6k, False: 339k] ------------------ 1157| | 1158| 374k| if (have_tile_pos) { ------------------ | Branch (1158:9): [True: 2.61k, False: 371k] ------------------ 1159| 2.61k| const int n_bits = c->frame_hdr->tiling.log2_cols + 1160| 2.61k| c->frame_hdr->tiling.log2_rows; 1161| 2.61k| c->tile[c->n_tile_data].start = dav1d_get_bits(gb, n_bits); 1162| 2.61k| c->tile[c->n_tile_data].end = dav1d_get_bits(gb, n_bits); 1163| 371k| } else { 1164| 371k| c->tile[c->n_tile_data].start = 0; 1165| 371k| c->tile[c->n_tile_data].end = n_tiles - 1; 1166| 371k| } 1167| 374k|} dav1d_pal_dsp_init: 71| 9.41k|COLD void dav1d_pal_dsp_init(Dav1dPalDSPContext *const c) { 72| 9.41k| c->pal_idx_finish = pal_idx_finish_c; 73| | 74| 9.41k|#if HAVE_ASM 75| |#if ARCH_RISCV 76| | pal_dsp_init_riscv(c); 77| |#elif ARCH_X86 78| | pal_dsp_init_x86(c); 79| 9.41k|#endif 80| 9.41k|#endif 81| 9.41k|} dav1d_default_picture_alloc: 46| 405k|int dav1d_default_picture_alloc(Dav1dPicture *const p, void *const cookie) { 47| 405k| const int hbd = p->p.bpc > 8; 48| 405k| const int aligned_w = (p->p.w + 127) & ~127; 49| 405k| const int aligned_h = (p->p.h + 127) & ~127; 50| 405k| const int has_chroma = p->p.layout != DAV1D_PIXEL_LAYOUT_I400; 51| 405k| const int ss_ver = p->p.layout == DAV1D_PIXEL_LAYOUT_I420; 52| 405k| const int ss_hor = p->p.layout != DAV1D_PIXEL_LAYOUT_I444; 53| 405k| ptrdiff_t y_stride = aligned_w << hbd; 54| 405k| ptrdiff_t uv_stride = has_chroma ? y_stride >> ss_hor : 0; ------------------ | Branch (54:27): [True: 328k, False: 77.0k] ------------------ 55| | /* Due to how mapping of addresses to sets works in most L1 and L2 cache 56| | * implementations, strides of multiples of certain power-of-two numbers 57| | * may cause multiple rows of the same superblock to map to the same set, 58| | * causing evictions of previous rows resulting in a reduction in cache 59| | * hit rate. Avoid that by slightly padding the stride when necessary. */ 60| 405k| if (!(y_stride & 1023)) ------------------ | Branch (60:9): [True: 39.1k, False: 366k] ------------------ 61| 39.1k| y_stride += DAV1D_PICTURE_ALIGNMENT; ------------------ | | 44| 39.1k|#define DAV1D_PICTURE_ALIGNMENT 64 ------------------ 62| 405k| if (!(uv_stride & 1023) && has_chroma) ------------------ | Branch (62:9): [True: 113k, False: 291k] | Branch (62:32): [True: 36.6k, False: 77.0k] ------------------ 63| 36.6k| uv_stride += DAV1D_PICTURE_ALIGNMENT; ------------------ | | 44| 36.6k|#define DAV1D_PICTURE_ALIGNMENT 64 ------------------ 64| 405k| p->stride[0] = y_stride; 65| 405k| p->stride[1] = uv_stride; 66| 405k| const size_t y_sz = y_stride * aligned_h; 67| 405k| const size_t uv_sz = uv_stride * (aligned_h >> ss_ver); 68| 405k| const size_t pic_size = y_sz + 2 * uv_sz; 69| | 70| 405k| uint8_t *const buf = dav1d_mem_pool_pop(cookie, pic_size + DAV1D_PICTURE_ALIGNMENT); ------------------ | | 44| 405k|#define DAV1D_PICTURE_ALIGNMENT 64 ------------------ 71| 405k| if (!buf) return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (71:9): [True: 0, False: 405k] ------------------ 72| 405k| p->allocator_data = buf; 73| 405k| p->data[0] = buf; 74| 405k| p->data[1] = has_chroma ? buf + y_sz : NULL; ------------------ | Branch (74:18): [True: 328k, False: 77.0k] ------------------ 75| 405k| p->data[2] = has_chroma ? buf + y_sz + uv_sz : NULL; ------------------ | Branch (75:18): [True: 328k, False: 77.0k] ------------------ 76| | 77| 405k| return 0; 78| 405k|} dav1d_default_picture_release: 80| 405k|void dav1d_default_picture_release(Dav1dPicture *const p, void *const cookie) { 81| 405k| dav1d_mem_pool_push(cookie, p->allocator_data); 82| 405k|} dav1d_picture_free_itut_t35: 99| 972|void dav1d_picture_free_itut_t35(const uint8_t *const data, void *const user_data) { 100| 972| struct itut_t35_ctx_context *itut_t35_ctx = user_data; 101| | 102| 3.05k| for (size_t i = 0; i < itut_t35_ctx->n_itut_t35; i++) ------------------ | Branch (102:24): [True: 2.08k, False: 972] ------------------ 103| 2.08k| dav1d_free(itut_t35_ctx->itut_t35[i].payload); ------------------ | | 135| 2.08k|#define dav1d_free(ptr) free(ptr) ------------------ 104| 972| dav1d_free(itut_t35_ctx->itut_t35); ------------------ | | 135| 972|#define dav1d_free(ptr) free(ptr) ------------------ 105| 972| dav1d_free(itut_t35_ctx); ------------------ | | 135| 972|#define dav1d_free(ptr) free(ptr) ------------------ 106| 972|} dav1d_picture_copy_props: 164| 372k|{ 165| 372k| dav1d_data_props_copy(&p->m, props); 166| | 167| 372k| dav1d_ref_dec(&p->content_light_ref); 168| 372k| p->content_light_ref = content_light_ref; 169| 372k| p->content_light = content_light; 170| 372k| if (content_light_ref) dav1d_ref_inc(content_light_ref); ------------------ | Branch (170:9): [True: 6.62k, False: 366k] ------------------ 171| | 172| 372k| dav1d_ref_dec(&p->mastering_display_ref); 173| 372k| p->mastering_display_ref = mastering_display_ref; 174| 372k| p->mastering_display = mastering_display; 175| 372k| if (mastering_display_ref) dav1d_ref_inc(mastering_display_ref); ------------------ | Branch (175:9): [True: 778, False: 372k] ------------------ 176| | 177| 372k| dav1d_ref_dec(&p->itut_t35_ref); 178| 372k| p->itut_t35_ref = itut_t35_ref; 179| 372k| p->itut_t35 = itut_t35; 180| 372k| p->n_itut_t35 = n_itut_t35; 181| 372k| if (itut_t35_ref) dav1d_ref_inc(itut_t35_ref); ------------------ | Branch (181:9): [True: 1.15k, False: 371k] ------------------ 182| 372k|} dav1d_thread_picture_alloc: 186| 356k|{ 187| 356k| Dav1dThreadPicture *const p = &f->sr_cur; 188| | 189| 356k| const int res = picture_alloc(c, &p->p, f->frame_hdr->width[1], f->frame_hdr->height, 190| 356k| f->seq_hdr, f->seq_hdr_ref, 191| 356k| f->frame_hdr, f->frame_hdr_ref, 192| 356k| bpc, &f->tile[0].data.m, &c->allocator, 193| 356k| (void **) &p->progress); 194| 356k| if (res) return res; ------------------ | Branch (194:9): [True: 0, False: 356k] ------------------ 195| | 196| | // Don't clear these flags from c->frame_flags if the frame is not going to be output. 197| | // This way they will be added to the next visible frame too. 198| 356k| const int flags_mask = ((f->frame_hdr->show_frame || c->output_invisible_frames) && ------------------ | Branch (198:30): [True: 292k, False: 64.0k] | Branch (198:58): [True: 0, False: 64.0k] ------------------ 199| 292k| c->max_spatial_id == f->frame_hdr->spatial_id) ------------------ | Branch (199:29): [True: 281k, False: 10.9k] ------------------ 200| 356k| ? 0 : (PICTURE_FLAG_NEW_SEQUENCE | PICTURE_FLAG_NEW_OP_PARAMS_INFO); 201| 356k| p->flags = c->frame_flags; 202| 356k| c->frame_flags &= flags_mask; 203| | 204| 356k| p->visible = f->frame_hdr->show_frame; 205| 356k| p->showable = f->frame_hdr->showable_frame; 206| | 207| 356k| if (p->visible) { ------------------ | Branch (207:9): [True: 292k, False: 64.0k] ------------------ 208| | // Only add HDR10+ and T35 metadata when show frame flag is enabled 209| 292k| dav1d_picture_copy_props(&p->p, c->content_light, c->content_light_ref, 210| 292k| c->mastering_display, c->mastering_display_ref, 211| 292k| c->itut_t35, c->itut_t35_ref, c->n_itut_t35, 212| 292k| &f->tile[0].data.m); 213| | 214| | // Must be removed from the context after being attached to the frame 215| 292k| dav1d_ref_dec(&c->itut_t35_ref); 216| 292k| c->itut_t35 = NULL; 217| 292k| c->n_itut_t35 = 0; 218| 292k| } else { 219| 64.0k| dav1d_data_props_copy(&p->p.m, &f->tile[0].data.m); 220| 64.0k| } 221| | 222| 356k| if (c->n_fc > 1) { ------------------ | Branch (222:9): [True: 356k, False: 0] ------------------ 223| 356k| atomic_init(&p->progress[0], 0); 224| | atomic_init(&p->progress[1], 0); 225| 356k| } 226| 356k| return res; 227| 356k|} dav1d_picture_alloc_copy: 231| 49.2k|{ 232| 49.2k| struct pic_ctx_context *const pic_ctx = (struct pic_ctx_context*)src->ref->const_data; 233| 49.2k| const int res = picture_alloc(c, dst, w, src->p.h, 234| 49.2k| src->seq_hdr, src->seq_hdr_ref, 235| 49.2k| src->frame_hdr, src->frame_hdr_ref, 236| 49.2k| src->p.bpc, &src->m, &pic_ctx->allocator, 237| 49.2k| NULL); 238| 49.2k| if (res) return res; ------------------ | Branch (238:9): [True: 0, False: 49.2k] ------------------ 239| | 240| 49.2k| dav1d_picture_copy_props(dst, src->content_light, src->content_light_ref, 241| 49.2k| src->mastering_display, src->mastering_display_ref, 242| 49.2k| src->itut_t35, src->itut_t35_ref, src->n_itut_t35, 243| 49.2k| &src->m); 244| | 245| 49.2k| return 0; 246| 49.2k|} dav1d_picture_ref: 248| 4.07M|void dav1d_picture_ref(Dav1dPicture *const dst, const Dav1dPicture *const src) { 249| 4.07M| assert(dst != NULL); ------------------ | Branch (249:5): [True: 4.07M, False: 0] ------------------ 250| 4.07M| assert(dst->data[0] == NULL); ------------------ | Branch (250:5): [True: 4.07M, False: 0] ------------------ 251| 4.07M| assert(src != NULL); ------------------ | Branch (251:5): [True: 4.07M, False: 0] ------------------ 252| | 253| 4.07M| if (src->ref) { ------------------ | Branch (253:9): [True: 4.07M, False: 0] ------------------ 254| 4.07M| assert(src->data[0] != NULL); ------------------ | Branch (254:9): [True: 4.07M, False: 0] ------------------ 255| 4.07M| dav1d_ref_inc(src->ref); 256| 4.07M| } 257| 4.07M| if (src->frame_hdr_ref) dav1d_ref_inc(src->frame_hdr_ref); ------------------ | Branch (257:9): [True: 4.07M, False: 0] ------------------ 258| 4.07M| if (src->seq_hdr_ref) dav1d_ref_inc(src->seq_hdr_ref); ------------------ | Branch (258:9): [True: 4.07M, False: 0] ------------------ 259| 4.07M| if (src->m.user_data.ref) dav1d_ref_inc(src->m.user_data.ref); ------------------ | Branch (259:9): [True: 0, False: 4.07M] ------------------ 260| 4.07M| if (src->content_light_ref) dav1d_ref_inc(src->content_light_ref); ------------------ | Branch (260:9): [True: 44.0k, False: 4.02M] ------------------ 261| 4.07M| if (src->mastering_display_ref) dav1d_ref_inc(src->mastering_display_ref); ------------------ | Branch (261:9): [True: 3.24k, False: 4.06M] ------------------ 262| 4.07M| if (src->itut_t35_ref) dav1d_ref_inc(src->itut_t35_ref); ------------------ | Branch (262:9): [True: 5.84k, False: 4.06M] ------------------ 263| 4.07M| *dst = *src; 264| 4.07M|} dav1d_picture_move_ref: 266| 166k|void dav1d_picture_move_ref(Dav1dPicture *const dst, Dav1dPicture *const src) { 267| 166k| assert(dst != NULL); ------------------ | Branch (267:5): [True: 166k, False: 0] ------------------ 268| 166k| assert(dst->data[0] == NULL); ------------------ | Branch (268:5): [True: 166k, False: 0] ------------------ 269| 166k| assert(src != NULL); ------------------ | Branch (269:5): [True: 166k, False: 0] ------------------ 270| | 271| 166k| if (src->ref) ------------------ | Branch (271:9): [True: 166k, False: 0] ------------------ 272| 166k| assert(src->data[0] != NULL); ------------------ | Branch (272:9): [True: 166k, False: 0] ------------------ 273| | 274| 166k| *dst = *src; 275| 166k| memset(src, 0, sizeof(*src)); 276| 166k|} dav1d_thread_picture_ref: 280| 3.75M|{ 281| 3.75M| dav1d_picture_ref(&dst->p, &src->p); 282| 3.75M| dst->visible = src->visible; 283| 3.75M| dst->showable = src->showable; 284| 3.75M| dst->progress = src->progress; 285| 3.75M| dst->flags = src->flags; 286| 3.75M|} dav1d_picture_unref_internal: 299| 4.72M|void dav1d_picture_unref_internal(Dav1dPicture *const p) { 300| 4.72M| validate_input(p != NULL); ------------------ | | 59| 4.72M|#define validate_input(x) validate_input_or_ret(x, ) | | ------------------ | | | | 52| 4.72M| if (!(x)) { \ | | | | ------------------ | | | | | Branch (52:9): [True: 0, False: 4.72M] | | | | ------------------ | | | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | | | ------------------ | | | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | | | ------------------ | | | | 54| 0| #x, __func__); \ | | | | 55| 0| debug_abort(); \ | | | | ------------------ | | | | | | 39| 0|#define debug_abort abort | | | | ------------------ | | | | 56| 0| return r; \ | | | | 57| 0| } | | ------------------ ------------------ 301| | 302| 4.72M| if (p->ref) { ------------------ | Branch (302:9): [True: 4.47M, False: 246k] ------------------ 303| 4.47M| validate_input(p->data[0] != NULL); ------------------ | | 59| 4.47M|#define validate_input(x) validate_input_or_ret(x, ) | | ------------------ | | | | 52| 4.47M| if (!(x)) { \ | | | | ------------------ | | | | | Branch (52:9): [True: 0, False: 4.47M] | | | | ------------------ | | | | 53| 0| debug_print("Input validation check \'%s\' failed in %s!\n", \ | | | | ------------------ | | | | | | 38| 0|#define debug_print(...) fprintf(stderr, __VA_ARGS__) | | | | ------------------ | | | | 54| 0| #x, __func__); \ | | | | 55| 0| debug_abort(); \ | | | | ------------------ | | | | | | 39| 0|#define debug_abort abort | | | | ------------------ | | | | 56| 0| return r; \ | | | | 57| 0| } | | ------------------ ------------------ 304| 4.47M| dav1d_ref_dec(&p->ref); 305| 4.47M| } 306| 4.72M| dav1d_ref_dec(&p->seq_hdr_ref); 307| 4.72M| dav1d_ref_dec(&p->frame_hdr_ref); 308| 4.72M| dav1d_ref_dec(&p->m.user_data.ref); 309| 4.72M| dav1d_ref_dec(&p->content_light_ref); 310| 4.72M| dav1d_ref_dec(&p->mastering_display_ref); 311| 4.72M| dav1d_ref_dec(&p->itut_t35_ref); 312| 4.72M| memset(p, 0, sizeof(*p)); 313| 4.72M| dav1d_data_props_set_defaults(&p->m); 314| 4.72M|} dav1d_thread_picture_unref: 316| 4.15M|void dav1d_thread_picture_unref(Dav1dThreadPicture *const p) { 317| 4.15M| dav1d_picture_unref_internal(&p->p); 318| | 319| | p->progress = NULL; 320| 4.15M|} dav1d_picture_get_event_flags: 322| 176k|enum Dav1dEventFlags dav1d_picture_get_event_flags(const Dav1dThreadPicture *const p) { 323| 176k| if (!p->flags) ------------------ | Branch (323:9): [True: 147k, False: 29.2k] ------------------ 324| 147k| return 0; 325| | 326| 29.2k| enum Dav1dEventFlags flags = 0; 327| 29.2k| if (p->flags & PICTURE_FLAG_NEW_SEQUENCE) ------------------ | Branch (327:9): [True: 14.9k, False: 14.2k] ------------------ 328| 14.9k| flags |= DAV1D_EVENT_FLAG_NEW_SEQUENCE; 329| 29.2k| if (p->flags & PICTURE_FLAG_NEW_OP_PARAMS_INFO) ------------------ | Branch (329:9): [True: 18, False: 29.2k] ------------------ 330| 18| flags |= DAV1D_EVENT_FLAG_NEW_OP_PARAMS_INFO; 331| | 332| 29.2k| return flags; 333| 176k|} picture.c:picture_alloc: 117| 405k|{ 118| 405k| if (p->data[0]) { ------------------ | Branch (118:9): [True: 0, False: 405k] ------------------ 119| 0| dav1d_log(c, "Picture already allocated!\n"); ------------------ | | 44| 0|#define dav1d_log(...) do { } while(0) | | ------------------ | | | Branch (44:37): [Folded, False: 0] | | ------------------ ------------------ 120| 0| return -1; 121| 0| } 122| 405k| assert(bpc > 0 && bpc <= 16); ------------------ | Branch (122:5): [True: 405k, False: 0] | Branch (122:5): [True: 405k, False: 0] ------------------ 123| | 124| 405k| size_t extra = c->n_fc > 1 ? sizeof(atomic_int) * 2 : 0; ------------------ | Branch (124:20): [True: 405k, False: 0] ------------------ 125| 405k| struct pic_ctx_context *pic_ctx = dav1d_mem_pool_pop(c->pic_ctx_pool, extra + 126| 405k| sizeof(struct pic_ctx_context)); 127| 405k| if (!pic_ctx) ------------------ | Branch (127:9): [True: 0, False: 405k] ------------------ 128| 0| return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 129| | 130| 405k| p->p.w = w; 131| 405k| p->p.h = h; 132| 405k| p->seq_hdr = seq_hdr; 133| 405k| p->frame_hdr = frame_hdr; 134| 405k| p->p.layout = seq_hdr->layout; 135| 405k| p->p.bpc = bpc; 136| 405k| dav1d_data_props_set_defaults(&p->m); 137| 405k| const int res = p_allocator->alloc_picture_callback(p, p_allocator->cookie); 138| 405k| if (res < 0) { ------------------ | Branch (138:9): [True: 0, False: 405k] ------------------ 139| 0| dav1d_mem_pool_push(c->pic_ctx_pool, pic_ctx); 140| 0| return res; 141| 0| } 142| | 143| 405k| pic_ctx->allocator = *p_allocator; 144| 405k| pic_ctx->pic = *p; 145| 405k| p->ref = dav1d_ref_init(&pic_ctx->ref, pic_ctx, free_buffer, c->pic_ctx_pool, 0); 146| | 147| 405k| p->seq_hdr_ref = seq_hdr_ref; 148| 405k| if (seq_hdr_ref) dav1d_ref_inc(seq_hdr_ref); ------------------ | Branch (148:9): [True: 405k, False: 0] ------------------ 149| | 150| 405k| p->frame_hdr_ref = frame_hdr_ref; 151| 405k| if (frame_hdr_ref) dav1d_ref_inc(frame_hdr_ref); ------------------ | Branch (151:9): [True: 405k, False: 0] ------------------ 152| | 153| 405k| if (extra && extra_ptr) ------------------ | Branch (153:9): [True: 405k, False: 0] | Branch (153:18): [True: 356k, False: 49.2k] ------------------ 154| 356k| *extra_ptr = &pic_ctx->extra_data; 155| | 156| 405k| return 0; 157| 405k|} picture.c:free_buffer: 91| 405k|static void free_buffer(const uint8_t *const data, void *const user_data) { 92| 405k| struct pic_ctx_context *pic_ctx = (struct pic_ctx_context*)data; 93| | 94| 405k| pic_ctx->allocator.release_picture_callback(&pic_ctx->pic, 95| 405k| pic_ctx->allocator.cookie); 96| 405k| dav1d_mem_pool_push(user_data, pic_ctx); 97| 405k|} dav1d_init_qm_tables: 4684| 1|COLD void dav1d_init_qm_tables(void) { 4685| | // This function is guaranteed to be called only once 4686| | 4687| 16| for (int i = 0; i < 15; i++) ------------------ | Branch (4687:21): [True: 15, False: 1] ------------------ 4688| 45| for (int j = 0; j < 2; j++) { ------------------ | Branch (4688:25): [True: 30, False: 15] ------------------ 4689| | // note that the w/h in the assignment is inverted, this is on purpose 4690| | // because we store coefficients transposed 4691| 30| dav1d_qm_tbl[i][j][RTX_4X8 ] = qm_tbl_8x4[i][j]; 4692| 30| dav1d_qm_tbl[i][j][RTX_8X4 ] = qm_tbl_4x8[i][j]; 4693| 30| dav1d_qm_tbl[i][j][RTX_4X16 ] = qm_tbl_16x4[i][j]; 4694| 30| dav1d_qm_tbl[i][j][RTX_16X4 ] = qm_tbl_4x16[i][j]; 4695| 30| dav1d_qm_tbl[i][j][RTX_8X16 ] = qm_tbl_16x8[i][j]; 4696| 30| dav1d_qm_tbl[i][j][RTX_16X8 ] = qm_tbl_8x16[i][j]; 4697| 30| dav1d_qm_tbl[i][j][RTX_8X32 ] = qm_tbl_32x8[i][j]; 4698| 30| dav1d_qm_tbl[i][j][RTX_32X8 ] = qm_tbl_8x32[i][j]; 4699| 30| dav1d_qm_tbl[i][j][RTX_16X32] = qm_tbl_32x16[i][j]; 4700| 30| dav1d_qm_tbl[i][j][RTX_32X16] = qm_tbl_16x32[i][j]; 4701| | 4702| 30| dav1d_qm_tbl[i][j][ TX_4X4 ] = qm_tbl_4x4[i][j]; 4703| 30| dav1d_qm_tbl[i][j][ TX_8X8 ] = qm_tbl_8x8[i][j]; 4704| 30| dav1d_qm_tbl[i][j][ TX_16X16] = qm_tbl_16x16[i][j]; 4705| 30| dav1d_qm_tbl[i][j][ TX_32X32] = qm_tbl_32x32[i][j]; 4706| | 4707| 30| dav1d_qm_tbl[i][j][ TX_64X64] = dav1d_qm_tbl[i][j][ TX_32X32]; 4708| 30| dav1d_qm_tbl[i][j][RTX_64X32] = dav1d_qm_tbl[i][j][ TX_32X32]; 4709| 30| dav1d_qm_tbl[i][j][RTX_64X16] = dav1d_qm_tbl[i][j][RTX_32X16]; 4710| 30| dav1d_qm_tbl[i][j][RTX_32X64] = dav1d_qm_tbl[i][j][ TX_32X32]; 4711| 30| dav1d_qm_tbl[i][j][RTX_16X64] = dav1d_qm_tbl[i][j][RTX_16X32]; 4712| 30| } 4713| | 4714| | // dav1d_qm_tbl[15][*][*] == NULL 4715| 1|} dav1d_read_coef_blocks_8bpc: 826| 5.77M|{ 827| 5.77M| const Dav1dFrameContext *const f = t->f; 828| 5.77M| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 829| 5.77M| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 830| 5.77M| const int bx4 = t->bx & 31, by4 = t->by & 31; 831| 5.77M| const int cbx4 = bx4 >> ss_hor, cby4 = by4 >> ss_ver; 832| 5.77M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 833| 5.77M| const int bw4 = b_dim[0], bh4 = b_dim[1]; 834| 5.77M| const int cbw4 = (bw4 + ss_hor) >> ss_hor, cbh4 = (bh4 + ss_ver) >> ss_ver; 835| 5.77M| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 && ------------------ | Branch (835:28): [True: 4.10M, False: 1.66M] ------------------ 836| 4.10M| (bw4 > ss_hor || t->bx & 1) && ------------------ | Branch (836:29): [True: 3.81M, False: 294k] | Branch (836:45): [True: 146k, False: 147k] ------------------ 837| 3.96M| (bh4 > ss_ver || t->by & 1); ------------------ | Branch (837:29): [True: 3.71M, False: 247k] | Branch (837:45): [True: 123k, False: 123k] ------------------ 838| | 839| 5.77M| if (b->skip) { ------------------ | Branch (839:9): [True: 2.99M, False: 2.78M] ------------------ 840| 2.99M| BlockContext *const a = t->a; 841| 2.99M| dav1d_memset_pow2[b_dim[2]](&a->lcoef[bx4], 0x40); 842| 2.99M| dav1d_memset_pow2[b_dim[3]](&t->l.lcoef[by4], 0x40); 843| 2.99M| if (has_chroma) { ------------------ | Branch (843:13): [True: 1.46M, False: 1.52M] ------------------ 844| 1.46M| dav1d_memset_pow2_fn memset_cw = dav1d_memset_pow2[ulog2(cbw4)]; 845| 1.46M| dav1d_memset_pow2_fn memset_ch = dav1d_memset_pow2[ulog2(cbh4)]; 846| 1.46M| memset_cw(&a->ccoef[0][cbx4], 0x40); 847| 1.46M| memset_cw(&a->ccoef[1][cbx4], 0x40); 848| 1.46M| memset_ch(&t->l.ccoef[0][cby4], 0x40); 849| 1.46M| memset_ch(&t->l.ccoef[1][cby4], 0x40); 850| 1.46M| } 851| 2.99M| return; 852| 2.99M| } 853| | 854| 2.78M| Dav1dTileState *const ts = t->ts; 855| 2.78M| const int w4 = imin(bw4, f->bw - t->bx), h4 = imin(bh4, f->bh - t->by); 856| 2.78M| const int cw4 = (w4 + ss_hor) >> ss_hor, ch4 = (h4 + ss_ver) >> ss_ver; 857| 2.78M| assert(t->frame_thread.pass == 1); ------------------ | Branch (857:5): [True: 2.78M, False: 18.4E] ------------------ 858| 2.78M| assert(!b->skip); ------------------ | Branch (858:5): [True: 2.78M, False: 18.4E] ------------------ 859| 2.78M| const TxfmInfo *const uv_t_dim = &dav1d_txfm_dimensions[b->uvtx]; 860| 2.78M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[b->intra ? b->tx : b->max_ytx]; ------------------ | Branch (860:58): [True: 2.10M, False: 680k] ------------------ 861| 2.78M| const uint16_t tx_split[2] = { b->tx_split0, b->tx_split1 }; 862| | 863| 5.60M| for (int init_y = 0; init_y < h4; init_y += 16) { ------------------ | Branch (863:26): [True: 2.82M, False: 2.78M] ------------------ 864| 2.82M| const int sub_h4 = imin(h4, 16 + init_y); 865| 5.74M| for (int init_x = 0; init_x < w4; init_x += 16) { ------------------ | Branch (865:30): [True: 2.92M, False: 2.82M] ------------------ 866| 2.92M| const int sub_w4 = imin(w4, init_x + 16); 867| 2.92M| int y_off = !!init_y, y, x; 868| 5.98M| for (y = init_y, t->by += init_y; y < sub_h4; ------------------ | Branch (868:47): [True: 3.06M, False: 2.92M] ------------------ 869| 3.06M| y += t_dim->h, t->by += t_dim->h, y_off++) 870| 3.06M| { 871| 3.06M| int x_off = !!init_x; 872| 6.76M| for (x = init_x, t->bx += init_x; x < sub_w4; ------------------ | Branch (872:51): [True: 3.69M, False: 3.06M] ------------------ 873| 3.69M| x += t_dim->w, t->bx += t_dim->w, x_off++) 874| 3.69M| { 875| 3.69M| if (!b->intra) { ------------------ | Branch (875:25): [True: 749k, False: 2.94M] ------------------ 876| 749k| read_coef_tree(t, bs, b, b->max_ytx, 0, tx_split, 877| 749k| x_off, y_off, NULL); 878| 2.94M| } else { 879| 2.94M| uint8_t cf_ctx = 0x40; 880| 2.94M| enum TxfmType txtp; 881| 2.94M| const int eob = 882| 2.94M| decode_coefs(t, &t->a->lcoef[bx4 + x], 883| 2.94M| &t->l.lcoef[by4 + y], b->tx, bs, b, 1, 884| 2.94M| 0, ts->frame_thread[1].cf, &txtp, &cf_ctx); 885| 2.94M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 2.94M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 2.94M] | | ------------------ | | 35| 2.94M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 2.94M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 886| 0| printf("Post-y-cf-blk[tx=%d,txtp=%d,eob=%d]: r=%d\n", 887| 0| b->tx, txtp, eob, ts->msac.rng); 888| 2.94M| *ts->frame_thread[1].cbi++ = eob * (1 << 5) + txtp; 889| 2.94M| ts->frame_thread[1].cf += imin(t_dim->w, 8) * imin(t_dim->h, 8) * 16; 890| 2.94M| dav1d_memset_likely_pow2(&t->a->lcoef[bx4 + x], cf_ctx, imin(t_dim->w, f->bw - t->bx)); 891| 2.94M| dav1d_memset_likely_pow2(&t->l.lcoef[by4 + y], cf_ctx, imin(t_dim->h, f->bh - t->by)); 892| 2.94M| } 893| 3.69M| } 894| 3.06M| t->bx -= x; 895| 3.06M| } 896| 2.92M| t->by -= y; 897| | 898| 2.92M| if (!has_chroma) continue; ------------------ | Branch (898:17): [True: 426k, False: 2.49M] ------------------ 899| | 900| 2.49M| const int sub_ch4 = imin(ch4, (init_y + 16) >> ss_ver); 901| 2.49M| const int sub_cw4 = imin(cw4, (init_x + 16) >> ss_hor); 902| 7.47M| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (902:30): [True: 4.97M, False: 2.49M] ------------------ 903| 10.3M| for (y = init_y >> ss_ver, t->by += init_y; y < sub_ch4; ------------------ | Branch (903:61): [True: 5.37M, False: 4.97M] ------------------ 904| 5.37M| y += uv_t_dim->h, t->by += uv_t_dim->h << ss_ver) 905| 5.37M| { 906| 11.7M| for (x = init_x >> ss_hor, t->bx += init_x; x < sub_cw4; ------------------ | Branch (906:65): [True: 6.41M, False: 5.37M] ------------------ 907| 6.41M| x += uv_t_dim->w, t->bx += uv_t_dim->w << ss_hor) 908| 6.41M| { 909| 6.41M| uint8_t cf_ctx = 0x40; 910| 6.41M| enum TxfmType txtp; 911| 6.41M| if (!b->intra) ------------------ | Branch (911:29): [True: 1.08M, False: 5.33M] ------------------ 912| 1.08M| txtp = t->scratch.txtp_map[(by4 + (y << ss_ver)) * 32 + 913| 1.08M| bx4 + (x << ss_hor)]; 914| 6.41M| const int eob = 915| 6.41M| decode_coefs(t, &t->a->ccoef[pl][cbx4 + x], 916| 6.41M| &t->l.ccoef[pl][cby4 + y], b->uvtx, bs, 917| 6.41M| b, b->intra, 1 + pl, ts->frame_thread[1].cf, 918| 6.41M| &txtp, &cf_ctx); 919| 6.41M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 6.41M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 6.41M] | | ------------------ | | 35| 6.41M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 6.41M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 920| 0| printf("Post-uv-cf-blk[pl=%d,tx=%d," 921| 0| "txtp=%d,eob=%d]: r=%d\n", 922| 0| pl, b->uvtx, txtp, eob, ts->msac.rng); 923| 6.41M| *ts->frame_thread[1].cbi++ = eob * (1 << 5) + txtp; 924| 6.41M| ts->frame_thread[1].cf += uv_t_dim->w * uv_t_dim->h * 16; 925| 6.41M| int ctw = imin(uv_t_dim->w, (f->bw - t->bx + ss_hor) >> ss_hor); 926| 6.41M| int cth = imin(uv_t_dim->h, (f->bh - t->by + ss_ver) >> ss_ver); 927| 6.41M| dav1d_memset_likely_pow2(&t->a->ccoef[pl][cbx4 + x], cf_ctx, ctw); 928| 6.41M| dav1d_memset_likely_pow2(&t->l.ccoef[pl][cby4 + y], cf_ctx, cth); 929| 6.41M| } 930| 5.37M| t->bx -= x << ss_hor; 931| 5.37M| } 932| 4.97M| t->by -= y << ss_ver; 933| 4.97M| } 934| 2.49M| } 935| 2.82M| } 936| 2.78M|} dav1d_recon_b_intra_8bpc: 1179| 847k|{ 1180| 847k| Dav1dTileState *const ts = t->ts; 1181| 847k| const Dav1dFrameContext *const f = t->f; 1182| 847k| const Dav1dDSPContext *const dsp = f->dsp; 1183| 847k| const int bx4 = t->bx & 31, by4 = t->by & 31; 1184| 847k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 1185| 847k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 1186| 847k| const int cbx4 = bx4 >> ss_hor, cby4 = by4 >> ss_ver; 1187| 847k| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 1188| 847k| const int bw4 = b_dim[0], bh4 = b_dim[1]; 1189| 847k| const int w4 = imin(bw4, f->bw - t->bx), h4 = imin(bh4, f->bh - t->by); 1190| 847k| const int cw4 = (w4 + ss_hor) >> ss_hor, ch4 = (h4 + ss_ver) >> ss_ver; 1191| 847k| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 && ------------------ | Branch (1191:28): [True: 605k, False: 241k] ------------------ 1192| 605k| (bw4 > ss_hor || t->bx & 1) && ------------------ | Branch (1192:29): [True: 497k, False: 108k] | Branch (1192:45): [True: 54.2k, False: 54.0k] ------------------ 1193| 552k| (bh4 > ss_ver || t->by & 1); ------------------ | Branch (1193:29): [True: 460k, False: 91.9k] | Branch (1193:45): [True: 46.0k, False: 45.9k] ------------------ 1194| 847k| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[b->tx]; 1195| 847k| const TxfmInfo *const uv_t_dim = &dav1d_txfm_dimensions[b->uvtx]; 1196| | 1197| | // coefficient coding 1198| 847k| pixel *const edge = bitfn(t->scratch.edge) + 128; ------------------ | | 51| 847k|#define bitfn(x) x##_8bpc ------------------ 1199| 847k| const int cbw4 = (bw4 + ss_hor) >> ss_hor, cbh4 = (bh4 + ss_ver) >> ss_ver; 1200| | 1201| 847k| const int intra_edge_filter_flag = f->seq_hdr->intra_edge_filter << 10; 1202| | 1203| 1.77M| for (int init_y = 0; init_y < h4; init_y += 16) { ------------------ | Branch (1203:26): [True: 924k, False: 848k] ------------------ 1204| 924k| const int sub_h4 = imin(h4, 16 + init_y); 1205| 924k| const int sub_ch4 = imin(ch4, (init_y + 16) >> ss_ver); 1206| 1.99M| for (int init_x = 0; init_x < w4; init_x += 16) { ------------------ | Branch (1206:30): [True: 1.07M, False: 925k] ------------------ 1207| 1.07M| if (b->pal_sz[0]) { ------------------ | Branch (1207:17): [True: 13.9k, False: 1.05M] ------------------ 1208| 13.9k| pixel *dst = ((pixel *) f->cur.data[0]) + 1209| 13.9k| 4 * (t->by * PXSTRIDE(f->cur.stride[0]) + t->bx); ------------------ | | 53| 13.9k|#define PXSTRIDE(x) (x) ------------------ 1210| 13.9k| const uint8_t *pal_idx; 1211| 13.9k| if (t->frame_thread.pass) { ------------------ | Branch (1211:21): [True: 13.9k, False: 0] ------------------ 1212| 13.9k| const int p = t->frame_thread.pass & 1; 1213| 13.9k| assert(ts->frame_thread[p].pal_idx); ------------------ | Branch (1213:21): [True: 13.9k, False: 0] ------------------ 1214| 13.9k| pal_idx = ts->frame_thread[p].pal_idx; 1215| 13.9k| ts->frame_thread[p].pal_idx += bw4 * bh4 * 8; 1216| 13.9k| } else { 1217| 0| pal_idx = t->scratch.pal_idx_y; 1218| 0| } 1219| 13.9k| const pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (1219:42): [True: 13.9k, False: 0] ------------------ 1220| 13.9k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 1221| 13.9k| ((t->bx >> 1) + (t->by & 1))][0] : 1222| 13.9k| bytefn(t->scratch.pal)[0]; ------------------ | | 87| 0|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 13.9k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 1223| 13.9k| f->dsp->ipred.pal_pred(dst, f->cur.stride[0], pal, 1224| 13.9k| pal_idx, bw4 * 4, bh4 * 4); 1225| 13.9k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 13.9k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 13.9k] | | ------------------ | | 35| 13.9k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 13.9k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1226| 0| hex_dump(dst, PXSTRIDE(f->cur.stride[0]), ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 1227| 0| bw4 * 4, bh4 * 4, "y-pal-pred"); 1228| 13.9k| } 1229| | 1230| 1.07M| const int intra_flags = (sm_flag(t->a, bx4) | 1231| 1.07M| sm_flag(&t->l, by4) | 1232| 1.07M| intra_edge_filter_flag); 1233| 1.07M| const int sb_has_tr = init_x + 16 < w4 ? 1 : init_y ? 0 : ------------------ | Branch (1233:35): [True: 148k, False: 924k] | Branch (1233:58): [True: 76.8k, False: 847k] ------------------ 1234| 924k| intra_edge_flags & EDGE_I444_TOP_HAS_RIGHT; 1235| 1.07M| const int sb_has_bl = init_x ? 0 : init_y + 16 < h4 ? 1 : ------------------ | Branch (1235:35): [True: 148k, False: 924k] | Branch (1235:48): [True: 76.8k, False: 847k] ------------------ 1236| 924k| intra_edge_flags & EDGE_I444_LEFT_HAS_BOTTOM; 1237| 1.07M| int y, x; 1238| 1.07M| const int sub_w4 = imin(w4, init_x + 16); 1239| 2.76M| for (y = init_y, t->by += init_y; y < sub_h4; ------------------ | Branch (1239:47): [True: 1.69M, False: 1.07M] ------------------ 1240| 1.69M| y += t_dim->h, t->by += t_dim->h) 1241| 1.69M| { 1242| 1.69M| pixel *dst = ((pixel *) f->cur.data[0]) + 1243| 1.69M| 4 * (t->by * PXSTRIDE(f->cur.stride[0]) + ------------------ | | 53| 1.69M|#define PXSTRIDE(x) (x) ------------------ 1244| 1.69M| t->bx + init_x); 1245| 8.69M| for (x = init_x, t->bx += init_x; x < sub_w4; ------------------ | Branch (1245:51): [True: 6.99M, False: 1.69M] ------------------ 1246| 6.99M| x += t_dim->w, t->bx += t_dim->w) 1247| 6.99M| { 1248| 6.99M| if (b->pal_sz[0]) goto skip_y_pred; ------------------ | Branch (1248:25): [True: 15.1k, False: 6.97M] ------------------ 1249| | 1250| 6.97M| int angle = b->y_angle; 1251| 6.97M| const enum EdgeFlags edge_flags = 1252| 6.97M| (((y > init_y || !sb_has_tr) && (x + t_dim->w >= sub_w4)) ? ------------------ | Branch (1252:28): [True: 5.46M, False: 1.51M] | Branch (1252:42): [True: 409k, False: 1.10M] | Branch (1252:57): [True: 923k, False: 4.95M] ------------------ 1253| 6.05M| 0 : EDGE_I444_TOP_HAS_RIGHT) | 1254| 6.97M| ((x > init_x || (!sb_has_bl && y + t_dim->h >= sub_h4)) ? ------------------ | Branch (1254:27): [True: 5.29M, False: 1.68M] | Branch (1254:42): [True: 1.20M, False: 476k] | Branch (1254:56): [True: 747k, False: 457k] ------------------ 1255| 6.04M| 0 : EDGE_I444_LEFT_HAS_BOTTOM); 1256| 6.97M| const pixel *top_sb_edge = NULL; 1257| 6.97M| if (!(t->by & (f->sb_step - 1))) { ------------------ | Branch (1257:25): [True: 629k, False: 6.34M] ------------------ 1258| 629k| top_sb_edge = f->ipred_edge[0]; 1259| 629k| const int sby = t->by >> f->sb_shift; 1260| 629k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1261| 629k| } 1262| 6.97M| const enum IntraPredMode m = 1263| 6.97M| bytefn(dav1d_prepare_intra_edges)(t->bx, ------------------ | | 87| 6.97M|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 6.97M|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 1264| 6.97M| t->bx > ts->tiling.col_start, 1265| 6.97M| t->by, 1266| 6.97M| t->by > ts->tiling.row_start, 1267| 6.97M| ts->tiling.col_end, 1268| 6.97M| ts->tiling.row_end, 1269| 6.97M| edge_flags, dst, 1270| 6.97M| f->cur.stride[0], top_sb_edge, 1271| 6.97M| b->y_mode, &angle, 1272| 6.97M| t_dim->w, t_dim->h, 1273| 6.97M| f->seq_hdr->intra_edge_filter, 1274| 6.97M| edge HIGHBD_CALL_SUFFIX); 1275| 6.97M| dsp->ipred.intra_pred[m](dst, f->cur.stride[0], edge, 1276| 6.97M| t_dim->w * 4, t_dim->h * 4, 1277| 6.97M| angle | intra_flags, 1278| 6.97M| 4 * f->bw - 4 * t->bx, 1279| 6.97M| 4 * f->bh - 4 * t->by 1280| 6.97M| HIGHBD_CALL_SUFFIX); 1281| | 1282| 6.97M| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 6.97M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 6.97M] | | ------------------ | | 35| 6.97M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 6.97M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1283| 0| hex_dump(edge - t_dim->h * 4, t_dim->h * 4, 1284| 0| t_dim->h * 4, 2, "l"); 1285| 0| hex_dump(edge, 0, 1, 1, "tl"); 1286| 0| hex_dump(edge + 1, t_dim->w * 4, 1287| 0| t_dim->w * 4, 2, "t"); 1288| 0| hex_dump(dst, f->cur.stride[0], 1289| 0| t_dim->w * 4, t_dim->h * 4, "y-intra-pred"); 1290| 0| } 1291| | 1292| 6.99M| skip_y_pred: {} 1293| 6.99M| if (!b->skip) { ------------------ | Branch (1293:25): [True: 688k, False: 6.30M] ------------------ 1294| 688k| coef *cf; 1295| 688k| int eob; 1296| 688k| enum TxfmType txtp; 1297| 688k| if (t->frame_thread.pass) { ------------------ | Branch (1297:29): [True: 688k, False: 18.4E] ------------------ 1298| 688k| const int p = t->frame_thread.pass & 1; 1299| 688k| const int cbi = *ts->frame_thread[p].cbi++; 1300| 688k| cf = ts->frame_thread[p].cf; 1301| 688k| ts->frame_thread[p].cf += imin(t_dim->w, 8) * imin(t_dim->h, 8) * 16; 1302| 688k| eob = cbi >> 5; 1303| 688k| txtp = cbi & 0x1f; 1304| 18.4E| } else { 1305| 18.4E| uint8_t cf_ctx; 1306| 18.4E| cf = bitfn(t->cf); ------------------ | | 51| 18.4E|#define bitfn(x) x##_8bpc ------------------ 1307| 18.4E| eob = decode_coefs(t, &t->a->lcoef[bx4 + x], 1308| 18.4E| &t->l.lcoef[by4 + y], b->tx, bs, 1309| 18.4E| b, 1, 0, cf, &txtp, &cf_ctx); 1310| 18.4E| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 18.4E|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 18.4E] | | ------------------ | | 35| 18.4E| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 18.4E| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1311| 0| printf("Post-y-cf-blk[tx=%d,txtp=%d,eob=%d]: r=%d\n", 1312| 0| b->tx, txtp, eob, ts->msac.rng); 1313| 18.4E| dav1d_memset_likely_pow2(&t->a->lcoef[bx4 + x], cf_ctx, imin(t_dim->w, f->bw - t->bx)); 1314| 18.4E| dav1d_memset_likely_pow2(&t->l.lcoef[by4 + y], cf_ctx, imin(t_dim->h, f->bh - t->by)); 1315| 18.4E| } 1316| 688k| if (eob >= 0) { ------------------ | Branch (1316:29): [True: 503k, False: 184k] ------------------ 1317| 503k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 503k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 503k] | | ------------------ | | 35| 503k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 503k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1318| 0| coef_dump(cf, imin(t_dim->h, 8) * 4, 1319| 0| imin(t_dim->w, 8) * 4, 3, "dq"); 1320| 503k| dsp->itx.itxfm_add[b->tx] 1321| 503k| [txtp](dst, 1322| 503k| f->cur.stride[0], 1323| 503k| cf, eob HIGHBD_CALL_SUFFIX); 1324| 503k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 503k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 503k] | | ------------------ | | 35| 503k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 503k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1325| 0| hex_dump(dst, f->cur.stride[0], 1326| 0| t_dim->w * 4, t_dim->h * 4, "recon"); 1327| 503k| } 1328| 6.30M| } else if (!t->frame_thread.pass) { ------------------ | Branch (1328:32): [True: 0, False: 6.30M] ------------------ 1329| 0| dav1d_memset_pow2[t_dim->lw](&t->a->lcoef[bx4 + x], 0x40); 1330| 0| dav1d_memset_pow2[t_dim->lh](&t->l.lcoef[by4 + y], 0x40); 1331| 0| } 1332| 6.99M| dst += 4 * t_dim->w; 1333| 6.99M| } 1334| 1.69M| t->bx -= x; 1335| 1.69M| } 1336| 1.07M| t->by -= y; 1337| | 1338| 1.07M| if (!has_chroma) continue; ------------------ | Branch (1338:17): [True: 408k, False: 665k] ------------------ 1339| | 1340| 665k| const ptrdiff_t stride = f->cur.stride[1]; 1341| | 1342| 665k| if (b->uv_mode == CFL_PRED) { ------------------ | Branch (1342:17): [True: 46.3k, False: 619k] ------------------ 1343| 46.3k| assert(!init_x && !init_y); ------------------ | Branch (1343:17): [True: 46.3k, False: 0] | Branch (1343:17): [True: 46.3k, False: 0] ------------------ 1344| | 1345| 46.3k| int16_t *const ac = t->scratch.ac; 1346| 46.3k| pixel *y_src = ((pixel *) f->cur.data[0]) + 4 * (t->bx & ~ss_hor) + 1347| 46.3k| 4 * (t->by & ~ss_ver) * PXSTRIDE(f->cur.stride[0]); ------------------ | | 53| 46.3k|#define PXSTRIDE(x) (x) ------------------ 1348| 46.3k| const ptrdiff_t uv_off = 4 * ((t->bx >> ss_hor) + 1349| 46.3k| (t->by >> ss_ver) * PXSTRIDE(stride)); ------------------ | | 53| 46.3k|#define PXSTRIDE(x) (x) ------------------ 1350| 46.3k| pixel *const uv_dst[2] = { ((pixel *) f->cur.data[1]) + uv_off, 1351| 46.3k| ((pixel *) f->cur.data[2]) + uv_off }; 1352| | 1353| 46.3k| const int furthest_r = 1354| 46.3k| ((cw4 << ss_hor) + t_dim->w - 1) & ~(t_dim->w - 1); 1355| 46.3k| const int furthest_b = 1356| 46.3k| ((ch4 << ss_ver) + t_dim->h - 1) & ~(t_dim->h - 1); 1357| 46.3k| dsp->ipred.cfl_ac[f->cur.p.layout - 1](ac, y_src, f->cur.stride[0], 1358| 46.3k| cbw4 - (furthest_r >> ss_hor), 1359| 46.3k| cbh4 - (furthest_b >> ss_ver), 1360| 46.3k| cbw4 * 4, cbh4 * 4); 1361| 139k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1361:34): [True: 92.7k, False: 46.3k] ------------------ 1362| 92.7k| if (!b->cfl_alpha[pl]) continue; ------------------ | Branch (1362:25): [True: 18.0k, False: 74.6k] ------------------ 1363| 74.6k| int angle = 0; 1364| 74.6k| const pixel *top_sb_edge = NULL; 1365| 74.6k| if (!((t->by & ~ss_ver) & (f->sb_step - 1))) { ------------------ | Branch (1365:25): [True: 10.4k, False: 64.1k] ------------------ 1366| 10.4k| top_sb_edge = f->ipred_edge[pl + 1]; 1367| 10.4k| const int sby = t->by >> f->sb_shift; 1368| 10.4k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1369| 10.4k| } 1370| 74.6k| const int xpos = t->bx >> ss_hor, ypos = t->by >> ss_ver; 1371| 74.6k| const int xstart = ts->tiling.col_start >> ss_hor; 1372| 74.6k| const int ystart = ts->tiling.row_start >> ss_ver; 1373| 74.6k| const enum IntraPredMode m = 1374| 74.6k| bytefn(dav1d_prepare_intra_edges)(xpos, xpos > xstart, ------------------ | | 87| 74.6k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 74.6k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 1375| 74.6k| ypos, ypos > ystart, 1376| 74.6k| ts->tiling.col_end >> ss_hor, 1377| 74.6k| ts->tiling.row_end >> ss_ver, 1378| 74.6k| 0, uv_dst[pl], stride, 1379| 74.6k| top_sb_edge, DC_PRED, &angle, 1380| 74.6k| uv_t_dim->w, uv_t_dim->h, 0, 1381| 74.6k| edge HIGHBD_CALL_SUFFIX); 1382| 74.6k| dsp->ipred.cfl_pred[m](uv_dst[pl], stride, edge, 1383| 74.6k| uv_t_dim->w * 4, 1384| 74.6k| uv_t_dim->h * 4, 1385| 74.6k| ac, b->cfl_alpha[pl] 1386| 74.6k| HIGHBD_CALL_SUFFIX); 1387| 74.6k| } 1388| 46.3k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 46.3k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 46.3k] | | ------------------ | | 35| 46.3k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 46.3k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1389| 0| ac_dump(ac, 4*cbw4, 4*cbh4, "ac"); 1390| 0| hex_dump(uv_dst[0], stride, cbw4 * 4, cbh4 * 4, "u-cfl-pred"); 1391| 0| hex_dump(uv_dst[1], stride, cbw4 * 4, cbh4 * 4, "v-cfl-pred"); 1392| 0| } 1393| 619k| } else if (b->pal_sz[1]) { ------------------ | Branch (1393:24): [True: 5.78k, False: 613k] ------------------ 1394| 5.78k| const ptrdiff_t uv_dstoff = 4 * ((t->bx >> ss_hor) + 1395| 5.78k| (t->by >> ss_ver) * PXSTRIDE(f->cur.stride[1])); ------------------ | | 53| 5.78k|#define PXSTRIDE(x) (x) ------------------ 1396| 5.78k| const pixel (*pal)[8]; 1397| 5.78k| const uint8_t *pal_idx; 1398| 5.78k| if (t->frame_thread.pass) { ------------------ | Branch (1398:21): [True: 5.78k, False: 0] ------------------ 1399| 5.78k| const int p = t->frame_thread.pass & 1; 1400| 5.78k| assert(ts->frame_thread[p].pal_idx); ------------------ | Branch (1400:21): [True: 5.78k, False: 0] ------------------ 1401| 5.78k| pal = f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 1402| 5.78k| ((t->bx >> 1) + (t->by & 1))]; 1403| 5.78k| pal_idx = ts->frame_thread[p].pal_idx; 1404| 5.78k| ts->frame_thread[p].pal_idx += cbw4 * cbh4 * 8; 1405| 5.78k| } else { 1406| 0| pal = bytefn(t->scratch.pal); ------------------ | | 87| 0|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 0|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 1407| 0| pal_idx = t->scratch.pal_idx_uv; 1408| 0| } 1409| | 1410| 5.78k| f->dsp->ipred.pal_pred(((pixel *) f->cur.data[1]) + uv_dstoff, 1411| 5.78k| f->cur.stride[1], pal[1], 1412| 5.78k| pal_idx, cbw4 * 4, cbh4 * 4); 1413| 5.78k| f->dsp->ipred.pal_pred(((pixel *) f->cur.data[2]) + uv_dstoff, 1414| 5.78k| f->cur.stride[1], pal[2], 1415| 5.78k| pal_idx, cbw4 * 4, cbh4 * 4); 1416| 5.78k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 5.78k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 5.78k] | | ------------------ | | 35| 5.78k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 5.78k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1417| 0| hex_dump(((pixel *) f->cur.data[1]) + uv_dstoff, 1418| 0| PXSTRIDE(f->cur.stride[1]), ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 1419| 0| cbw4 * 4, cbh4 * 4, "u-pal-pred"); 1420| 0| hex_dump(((pixel *) f->cur.data[2]) + uv_dstoff, 1421| 0| PXSTRIDE(f->cur.stride[1]), ------------------ | | 53| 0|#define PXSTRIDE(x) (x) ------------------ 1422| 0| cbw4 * 4, cbh4 * 4, "v-pal-pred"); 1423| 0| } 1424| 5.78k| } 1425| | 1426| 665k| const int sm_uv_fl = sm_uv_flag(t->a, cbx4) | 1427| 665k| sm_uv_flag(&t->l, cby4); 1428| 665k| const int uv_sb_has_tr = 1429| 665k| ((init_x + 16) >> ss_hor) < cw4 ? 1 : init_y ? 0 : ------------------ | Branch (1429:17): [True: 103k, False: 561k] | Branch (1429:55): [True: 54.7k, False: 507k] ------------------ 1430| 561k| intra_edge_flags & (EDGE_I420_TOP_HAS_RIGHT >> (f->cur.p.layout - 1)); 1431| 665k| const int uv_sb_has_bl = 1432| 665k| init_x ? 0 : ((init_y + 16) >> ss_ver) < ch4 ? 1 : ------------------ | Branch (1432:17): [True: 103k, False: 561k] | Branch (1432:30): [True: 54.7k, False: 507k] ------------------ 1433| 561k| intra_edge_flags & (EDGE_I420_LEFT_HAS_BOTTOM >> (f->cur.p.layout - 1)); 1434| 665k| const int sub_cw4 = imin(cw4, (init_x + 16) >> ss_hor); 1435| 1.99M| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1435:30): [True: 1.32M, False: 665k] ------------------ 1436| 3.55M| for (y = init_y >> ss_ver, t->by += init_y; y < sub_ch4; ------------------ | Branch (1436:61): [True: 2.22M, False: 1.32M] ------------------ 1437| 2.22M| y += uv_t_dim->h, t->by += uv_t_dim->h << ss_ver) 1438| 2.22M| { 1439| 2.22M| pixel *dst = ((pixel *) f->cur.data[1 + pl]) + 1440| 2.22M| 4 * ((t->by >> ss_ver) * PXSTRIDE(stride) + ------------------ | | 53| 2.22M|#define PXSTRIDE(x) (x) ------------------ 1441| 2.22M| ((t->bx + init_x) >> ss_hor)); 1442| 8.78M| for (x = init_x >> ss_hor, t->bx += init_x; x < sub_cw4; ------------------ | Branch (1442:65): [True: 6.56M, False: 2.22M] ------------------ 1443| 6.56M| x += uv_t_dim->w, t->bx += uv_t_dim->w << ss_hor) 1444| 6.56M| { 1445| 6.56M| if ((b->uv_mode == CFL_PRED && b->cfl_alpha[pl]) || ------------------ | Branch (1445:30): [True: 92.7k, False: 6.47M] | Branch (1445:56): [True: 74.6k, False: 18.0k] ------------------ 1446| 6.49M| b->pal_sz[1]) ------------------ | Branch (1446:29): [True: 12.4k, False: 6.47M] ------------------ 1447| 87.1k| { 1448| 87.1k| goto skip_uv_pred; 1449| 87.1k| } 1450| | 1451| 6.47M| int angle = b->uv_angle; 1452| | // this probably looks weird because we're using 1453| | // luma flags in a chroma loop, but that's because 1454| | // prepare_intra_edges() expects luma flags as input 1455| 6.47M| const enum EdgeFlags edge_flags = 1456| 6.47M| (((y > (init_y >> ss_ver) || !uv_sb_has_tr) && ------------------ | Branch (1456:32): [True: 4.83M, False: 1.64M] | Branch (1456:58): [True: 491k, False: 1.15M] ------------------ 1457| 5.32M| (x + uv_t_dim->w >= sub_cw4)) ? ------------------ | Branch (1457:31): [True: 1.32M, False: 4.00M] ------------------ 1458| 5.15M| 0 : EDGE_I444_TOP_HAS_RIGHT) | 1459| 6.47M| ((x > (init_x >> ss_hor) || ------------------ | Branch (1459:31): [True: 4.33M, False: 2.13M] ------------------ 1460| 2.13M| (!uv_sb_has_bl && y + uv_t_dim->h >= sub_ch4)) ? ------------------ | Branch (1460:32): [True: 1.45M, False: 679k] | Branch (1460:49): [True: 798k, False: 660k] ------------------ 1461| 5.14M| 0 : EDGE_I444_LEFT_HAS_BOTTOM); 1462| 6.47M| const pixel *top_sb_edge = NULL; 1463| 6.47M| if (!((t->by & ~ss_ver) & (f->sb_step - 1))) { ------------------ | Branch (1463:29): [True: 573k, False: 5.90M] ------------------ 1464| 573k| top_sb_edge = f->ipred_edge[1 + pl]; 1465| 573k| const int sby = t->by >> f->sb_shift; 1466| 573k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1467| 573k| } 1468| 6.47M| const enum IntraPredMode uv_mode = 1469| 6.47M| b->uv_mode == CFL_PRED ? DC_PRED : b->uv_mode; ------------------ | Branch (1469:30): [True: 18.0k, False: 6.45M] ------------------ 1470| 6.47M| const int xpos = t->bx >> ss_hor, ypos = t->by >> ss_ver; 1471| 6.47M| const int xstart = ts->tiling.col_start >> ss_hor; 1472| 6.47M| const int ystart = ts->tiling.row_start >> ss_ver; 1473| 6.47M| const enum IntraPredMode m = 1474| 6.47M| bytefn(dav1d_prepare_intra_edges)(xpos, xpos > xstart, ------------------ | | 87| 6.47M|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 6.47M|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 1475| 6.47M| ypos, ypos > ystart, 1476| 6.47M| ts->tiling.col_end >> ss_hor, 1477| 6.47M| ts->tiling.row_end >> ss_ver, 1478| 6.47M| edge_flags, dst, stride, 1479| 6.47M| top_sb_edge, uv_mode, 1480| 6.47M| &angle, uv_t_dim->w, 1481| 6.47M| uv_t_dim->h, 1482| 6.47M| f->seq_hdr->intra_edge_filter, 1483| 6.47M| edge HIGHBD_CALL_SUFFIX); 1484| 6.47M| angle |= intra_edge_filter_flag; 1485| 6.47M| dsp->ipred.intra_pred[m](dst, stride, edge, 1486| 6.47M| uv_t_dim->w * 4, 1487| 6.47M| uv_t_dim->h * 4, 1488| 6.47M| angle | sm_uv_fl, 1489| 6.47M| (4 * f->bw + ss_hor - 1490| 6.47M| 4 * (t->bx & ~ss_hor)) >> ss_hor, 1491| 6.47M| (4 * f->bh + ss_ver - 1492| 6.47M| 4 * (t->by & ~ss_ver)) >> ss_ver 1493| 6.47M| HIGHBD_CALL_SUFFIX); 1494| 6.47M| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 6.47M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 6.47M] | | ------------------ | | 35| 6.47M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 6.47M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1495| 0| hex_dump(edge - uv_t_dim->h * 4, uv_t_dim->h * 4, 1496| 0| uv_t_dim->h * 4, 2, "l"); 1497| 0| hex_dump(edge, 0, 1, 1, "tl"); 1498| 0| hex_dump(edge + 1, uv_t_dim->w * 4, 1499| 0| uv_t_dim->w * 4, 2, "t"); 1500| 0| hex_dump(dst, stride, uv_t_dim->w * 4, 1501| 0| uv_t_dim->h * 4, pl ? "v-intra-pred" : "u-intra-pred"); ------------------ | Branch (1501:55): [True: 0, False: 0] ------------------ 1502| 0| } 1503| | 1504| 6.56M| skip_uv_pred: {} 1505| 6.56M| if (!b->skip) { ------------------ | Branch (1505:29): [True: 741k, False: 5.82M] ------------------ 1506| 741k| enum TxfmType txtp; 1507| 741k| int eob; 1508| 741k| coef *cf; 1509| 741k| if (t->frame_thread.pass) { ------------------ | Branch (1509:33): [True: 741k, False: 18.4E] ------------------ 1510| 741k| const int p = t->frame_thread.pass & 1; 1511| 741k| const int cbi = *ts->frame_thread[p].cbi++; 1512| 741k| cf = ts->frame_thread[p].cf; 1513| 741k| ts->frame_thread[p].cf += uv_t_dim->w * uv_t_dim->h * 16; 1514| 741k| eob = cbi >> 5; 1515| 741k| txtp = cbi & 0x1f; 1516| 18.4E| } else { 1517| 18.4E| uint8_t cf_ctx; 1518| 18.4E| cf = bitfn(t->cf); ------------------ | | 51| 18.4E|#define bitfn(x) x##_8bpc ------------------ 1519| 18.4E| eob = decode_coefs(t, &t->a->ccoef[pl][cbx4 + x], 1520| 18.4E| &t->l.ccoef[pl][cby4 + y], 1521| 18.4E| b->uvtx, bs, b, 1, 1 + pl, cf, 1522| 18.4E| &txtp, &cf_ctx); 1523| 18.4E| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 18.4E|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 18.4E] | | ------------------ | | 35| 18.4E| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 18.4E| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1524| 0| printf("Post-uv-cf-blk[pl=%d,tx=%d," 1525| 0| "txtp=%d,eob=%d]: r=%d [x=%d,cbx4=%d]\n", 1526| 0| pl, b->uvtx, txtp, eob, ts->msac.rng, x, cbx4); 1527| 18.4E| int ctw = imin(uv_t_dim->w, (f->bw - t->bx + ss_hor) >> ss_hor); 1528| 18.4E| int cth = imin(uv_t_dim->h, (f->bh - t->by + ss_ver) >> ss_ver); 1529| 18.4E| dav1d_memset_likely_pow2(&t->a->ccoef[pl][cbx4 + x], cf_ctx, ctw); 1530| 18.4E| dav1d_memset_likely_pow2(&t->l.ccoef[pl][cby4 + y], cf_ctx, cth); 1531| 18.4E| } 1532| 741k| if (eob >= 0) { ------------------ | Branch (1532:33): [True: 281k, False: 459k] ------------------ 1533| 281k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 281k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 281k] | | ------------------ | | 35| 281k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 281k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1534| 0| coef_dump(cf, uv_t_dim->h * 4, 1535| 0| uv_t_dim->w * 4, 3, "dq"); 1536| 281k| dsp->itx.itxfm_add[b->uvtx] 1537| 281k| [txtp](dst, stride, 1538| 281k| cf, eob HIGHBD_CALL_SUFFIX); 1539| 281k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 281k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 281k] | | ------------------ | | 35| 281k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 281k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1540| 0| hex_dump(dst, stride, uv_t_dim->w * 4, 1541| 0| uv_t_dim->h * 4, "recon"); 1542| 281k| } 1543| 5.82M| } else if (!t->frame_thread.pass) { ------------------ | Branch (1543:36): [True: 0, False: 5.82M] ------------------ 1544| 0| dav1d_memset_pow2[uv_t_dim->lw](&t->a->ccoef[pl][cbx4 + x], 0x40); 1545| 0| dav1d_memset_pow2[uv_t_dim->lh](&t->l.ccoef[pl][cby4 + y], 0x40); 1546| 0| } 1547| 6.56M| dst += uv_t_dim->w * 4; 1548| 6.56M| } 1549| 2.22M| t->bx -= x << ss_hor; 1550| 2.22M| } 1551| 1.32M| t->by -= y << ss_ver; 1552| 1.32M| } 1553| 665k| } 1554| 924k| } 1555| 847k|} dav1d_recon_b_inter_8bpc: 1559| 1.79M|{ 1560| 1.79M| Dav1dTileState *const ts = t->ts; 1561| 1.79M| const Dav1dFrameContext *const f = t->f; 1562| 1.79M| const Dav1dDSPContext *const dsp = f->dsp; 1563| 1.79M| const int bx4 = t->bx & 31, by4 = t->by & 31; 1564| 1.79M| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 1565| 1.79M| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 1566| 1.79M| const int cbx4 = bx4 >> ss_hor, cby4 = by4 >> ss_ver; 1567| 1.79M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 1568| 1.79M| const int bw4 = b_dim[0], bh4 = b_dim[1]; 1569| 1.79M| const int w4 = imin(bw4, f->bw - t->bx), h4 = imin(bh4, f->bh - t->by); 1570| 1.79M| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 && ------------------ | Branch (1570:28): [True: 773k, False: 1.01M] ------------------ 1571| 773k| (bw4 > ss_hor || t->bx & 1) && ------------------ | Branch (1571:29): [True: 654k, False: 118k] | Branch (1571:45): [True: 59.1k, False: 59.4k] ------------------ 1572| 713k| (bh4 > ss_ver || t->by & 1); ------------------ | Branch (1572:29): [True: 625k, False: 88.4k] | Branch (1572:45): [True: 44.1k, False: 44.2k] ------------------ 1573| 1.79M| const int chr_layout_idx = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I400 ? 0 : ------------------ | Branch (1573:32): [True: 1.01M, False: 773k] ------------------ 1574| 1.79M| DAV1D_PIXEL_LAYOUT_I444 - f->cur.p.layout; 1575| 1.79M| int res; 1576| | 1577| | // prediction 1578| 1.79M| const int cbh4 = (bh4 + ss_ver) >> ss_ver, cbw4 = (bw4 + ss_hor) >> ss_hor; 1579| 1.79M| pixel *dst = ((pixel *) f->cur.data[0]) + 1580| 1.79M| 4 * (t->by * PXSTRIDE(f->cur.stride[0]) + t->bx); ------------------ | | 53| 1.79M|#define PXSTRIDE(x) (x) ------------------ 1581| 1.79M| const ptrdiff_t uvdstoff = 1582| 1.79M| 4 * ((t->bx >> ss_hor) + (t->by >> ss_ver) * PXSTRIDE(f->cur.stride[1])); ------------------ | | 53| 1.79M|#define PXSTRIDE(x) (x) ------------------ 1583| 1.79M| if (IS_KEY_OR_INTRA(f->frame_hdr)) { ------------------ | | 43| 1.79M| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 1.79M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (43:5): [True: 14.2k, False: 1.77M] | | ------------------ ------------------ 1584| | // intrabc 1585| 14.2k| assert(!f->frame_hdr->super_res.enabled); ------------------ | Branch (1585:9): [True: 14.2k, False: 0] ------------------ 1586| 14.2k| res = mc(t, dst, NULL, f->cur.stride[0], bw4, bh4, t->bx, t->by, 0, 1587| 14.2k| b->mv[0], &f->sr_cur, 0 /* unused */, FILTER_2D_BILINEAR); 1588| 14.2k| if (res) return res; ------------------ | Branch (1588:13): [True: 0, False: 14.2k] ------------------ 1589| 33.4k| if (has_chroma) for (int pl = 1; pl < 3; pl++) { ------------------ | Branch (1589:13): [True: 11.1k, False: 3.13k] | Branch (1589:42): [True: 22.3k, False: 11.1k] ------------------ 1590| 22.3k| res = mc(t, ((pixel *)f->cur.data[pl]) + uvdstoff, NULL, f->cur.stride[1], 1591| 22.3k| bw4 << (bw4 == ss_hor), bh4 << (bh4 == ss_ver), 1592| 22.3k| t->bx & ~ss_hor, t->by & ~ss_ver, pl, b->mv[0], 1593| 22.3k| &f->sr_cur, 0 /* unused */, FILTER_2D_BILINEAR); 1594| 22.3k| if (res) return res; ------------------ | Branch (1594:17): [True: 0, False: 22.3k] ------------------ 1595| 22.3k| } 1596| 1.77M| } else if (b->comp_type == COMP_INTER_NONE) { ------------------ | Branch (1596:16): [True: 1.68M, False: 89.0k] ------------------ 1597| 1.68M| const Dav1dThreadPicture *const refp = &f->refp[b->ref[0]]; 1598| 1.68M| const enum Filter2d filter_2d = b->filter2d; 1599| | 1600| 1.68M| if (imin(bw4, bh4) > 1 && ------------------ | Branch (1600:13): [True: 1.46M, False: 219k] ------------------ 1601| 1.46M| ((b->inter_mode == GLOBALMV && f->gmv_warp_allowed[b->ref[0]]) || ------------------ | Branch (1601:15): [True: 1.04M, False: 424k] | Branch (1601:44): [True: 617k, False: 426k] ------------------ 1602| 852k| (b->motion_mode == MM_WARP && t->warpmv.type > DAV1D_WM_TYPE_TRANSLATION))) ------------------ | Branch (1602:15): [True: 97.1k, False: 755k] | Branch (1602:44): [True: 91.0k, False: 6.12k] ------------------ 1603| 708k| { 1604| 708k| res = warp_affine(t, dst, NULL, f->cur.stride[0], b_dim, 0, refp, 1605| 708k| b->motion_mode == MM_WARP ? &t->warpmv : ------------------ | Branch (1605:31): [True: 91.0k, False: 617k] ------------------ 1606| 708k| &f->frame_hdr->gmv[b->ref[0]]); 1607| 708k| if (res) return res; ------------------ | Branch (1607:17): [True: 0, False: 708k] ------------------ 1608| 979k| } else { 1609| 979k| res = mc(t, dst, NULL, f->cur.stride[0], 1610| 979k| bw4, bh4, t->bx, t->by, 0, b->mv[0], refp, b->ref[0], filter_2d); 1611| 979k| if (res) return res; ------------------ | Branch (1611:17): [True: 0, False: 979k] ------------------ 1612| 979k| if (b->motion_mode == MM_OBMC) { ------------------ | Branch (1612:17): [True: 161k, False: 818k] ------------------ 1613| 161k| res = obmc(t, dst, f->cur.stride[0], b_dim, 0, bx4, by4, w4, h4); 1614| 161k| if (res) return res; ------------------ | Branch (1614:21): [True: 0, False: 161k] ------------------ 1615| 161k| } 1616| 979k| } 1617| 1.68M| if (b->interintra_type) { ------------------ | Branch (1617:13): [True: 60.4k, False: 1.62M] ------------------ 1618| 60.4k| pixel *const tl_edge = bitfn(t->scratch.edge) + 32; ------------------ | | 51| 60.4k|#define bitfn(x) x##_8bpc ------------------ 1619| 60.4k| enum IntraPredMode m = b->interintra_mode == II_SMOOTH_PRED ? ------------------ | Branch (1619:36): [True: 8.79k, False: 51.6k] ------------------ 1620| 51.6k| SMOOTH_PRED : b->interintra_mode; 1621| 60.4k| pixel *const tmp = bitfn(t->scratch.interintra); ------------------ | | 51| 60.4k|#define bitfn(x) x##_8bpc ------------------ 1622| 60.4k| int angle = 0; 1623| 60.4k| const pixel *top_sb_edge = NULL; 1624| 60.4k| if (!(t->by & (f->sb_step - 1))) { ------------------ | Branch (1624:17): [True: 8.87k, False: 51.5k] ------------------ 1625| 8.87k| top_sb_edge = f->ipred_edge[0]; 1626| 8.87k| const int sby = t->by >> f->sb_shift; 1627| 8.87k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1628| 8.87k| } 1629| 60.4k| m = bytefn(dav1d_prepare_intra_edges)(t->bx, t->bx > ts->tiling.col_start, ------------------ | | 87| 60.4k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 60.4k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 1630| 60.4k| t->by, t->by > ts->tiling.row_start, 1631| 60.4k| ts->tiling.col_end, ts->tiling.row_end, 1632| 60.4k| 0, dst, f->cur.stride[0], top_sb_edge, 1633| 60.4k| m, &angle, bw4, bh4, 0, tl_edge 1634| 60.4k| HIGHBD_CALL_SUFFIX); 1635| 60.4k| dsp->ipred.intra_pred[m](tmp, 4 * bw4 * sizeof(pixel), 1636| 60.4k| tl_edge, bw4 * 4, bh4 * 4, 0, 0, 0 1637| 60.4k| HIGHBD_CALL_SUFFIX); 1638| 60.4k| dsp->mc.blend(dst, f->cur.stride[0], tmp, 1639| 60.4k| bw4 * 4, bh4 * 4, II_MASK(0, bs, b)); ------------------ | | 83| 60.4k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 84| 60.4k| (size_t)((b)->interintra_type == INTER_INTRA_BLEND ? \ | | ------------------ | | | Branch (84:14): [True: 44.8k, False: 15.5k] | | ------------------ | | 85| 60.4k| dav1d_masks.offsets[c][(bs)-BS_32x32].ii[(b)->interintra_mode] : \ | | 86| 60.4k| dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[0][(b)->wedge_idx]) * 8)) ------------------ 1640| 60.4k| } 1641| | 1642| 1.68M| if (!has_chroma) goto skip_inter_chroma_pred; ------------------ | Branch (1642:13): [True: 1.11M, False: 576k] ------------------ 1643| | 1644| | // sub8x8 derivation 1645| 576k| int is_sub8x8 = bw4 == ss_hor || bh4 == ss_ver; ------------------ | Branch (1645:25): [True: 46.4k, False: 530k] | Branch (1645:42): [True: 32.8k, False: 497k] ------------------ 1646| 576k| refmvs_block *const *r; 1647| 576k| if (is_sub8x8) { ------------------ | Branch (1647:13): [True: 79.9k, False: 496k] ------------------ 1648| 79.9k| assert(ss_hor == 1); ------------------ | Branch (1648:13): [True: 79.9k, False: 1] ------------------ 1649| 79.9k| r = &t->rt.r[(t->by & 31) + 5]; 1650| 79.9k| if (bw4 == 1) is_sub8x8 &= r[0][t->bx - 1].ref.ref[0] > 0; ------------------ | Branch (1650:17): [True: 47.0k, False: 32.8k] ------------------ 1651| 79.9k| if (bh4 == ss_ver) is_sub8x8 &= r[-1][t->bx].ref.ref[0] > 0; ------------------ | Branch (1651:17): [True: 43.0k, False: 36.8k] ------------------ 1652| 79.9k| if (bw4 == 1 && bh4 == ss_ver) ------------------ | Branch (1652:17): [True: 47.0k, False: 32.8k] | Branch (1652:29): [True: 10.1k, False: 36.9k] ------------------ 1653| 10.1k| is_sub8x8 &= r[-1][t->bx - 1].ref.ref[0] > 0; 1654| 79.9k| } 1655| | 1656| | // chroma prediction 1657| 576k| if (is_sub8x8) { ------------------ | Branch (1657:13): [True: 71.6k, False: 504k] ------------------ 1658| 71.6k| assert(ss_hor == 1); ------------------ | Branch (1658:13): [True: 71.6k, False: 0] ------------------ 1659| 71.6k| ptrdiff_t h_off = 0, v_off = 0; 1660| 71.6k| if (bw4 == 1 && bh4 == ss_ver) { ------------------ | Branch (1660:17): [True: 42.1k, False: 29.4k] | Branch (1660:29): [True: 8.46k, False: 33.6k] ------------------ 1661| 25.4k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1661:34): [True: 16.9k, False: 8.46k] ------------------ 1662| 16.9k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, 1663| 16.9k| NULL, f->cur.stride[1], 1664| 16.9k| bw4, bh4, t->bx - 1, t->by - 1, 1 + pl, 1665| 16.9k| r[-1][t->bx - 1].mv.mv[0], 1666| 16.9k| &f->refp[r[-1][t->bx - 1].ref.ref[0] - 1], 1667| 16.9k| r[-1][t->bx - 1].ref.ref[0] - 1, 1668| 16.9k| t->frame_thread.pass != 2 ? t->tl_4x4_filter : ------------------ | Branch (1668:30): [True: 0, False: 16.9k] ------------------ 1669| 16.9k| f->frame_thread.b[((t->by - 1) * f->b4_stride) + t->bx - 1].filter2d); 1670| 16.9k| if (res) return res; ------------------ | Branch (1670:25): [True: 0, False: 16.9k] ------------------ 1671| 16.9k| } 1672| 8.46k| v_off = 2 * PXSTRIDE(f->cur.stride[1]); ------------------ | | 53| 8.46k|#define PXSTRIDE(x) (x) ------------------ 1673| 8.46k| h_off = 2; 1674| 8.46k| } 1675| 71.6k| if (bw4 == 1) { ------------------ | Branch (1675:17): [True: 42.1k, False: 29.4k] ------------------ 1676| 42.1k| const enum Filter2d left_filter_2d = 1677| 42.1k| dav1d_filter_2d[t->l.filter[1][by4]][t->l.filter[0][by4]]; 1678| 126k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1678:34): [True: 84.2k, False: 42.1k] ------------------ 1679| 84.2k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff + v_off, NULL, 1680| 84.2k| f->cur.stride[1], bw4, bh4, t->bx - 1, 1681| 84.2k| t->by, 1 + pl, r[0][t->bx - 1].mv.mv[0], 1682| 84.2k| &f->refp[r[0][t->bx - 1].ref.ref[0] - 1], 1683| 84.2k| r[0][t->bx - 1].ref.ref[0] - 1, 1684| 84.2k| t->frame_thread.pass != 2 ? left_filter_2d : ------------------ | Branch (1684:30): [True: 0, False: 84.2k] ------------------ 1685| 84.2k| f->frame_thread.b[(t->by * f->b4_stride) + t->bx - 1].filter2d); 1686| 84.2k| if (res) return res; ------------------ | Branch (1686:25): [True: 0, False: 84.2k] ------------------ 1687| 84.2k| } 1688| 42.1k| h_off = 2; 1689| 42.1k| } 1690| 71.6k| if (bh4 == ss_ver) { ------------------ | Branch (1690:17): [True: 37.9k, False: 33.6k] ------------------ 1691| 37.9k| const enum Filter2d top_filter_2d = 1692| 37.9k| dav1d_filter_2d[t->a->filter[1][bx4]][t->a->filter[0][bx4]]; 1693| 113k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1693:34): [True: 75.9k, False: 37.9k] ------------------ 1694| 75.9k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff + h_off, NULL, 1695| 75.9k| f->cur.stride[1], bw4, bh4, t->bx, t->by - 1, 1696| 75.9k| 1 + pl, r[-1][t->bx].mv.mv[0], 1697| 75.9k| &f->refp[r[-1][t->bx].ref.ref[0] - 1], 1698| 75.9k| r[-1][t->bx].ref.ref[0] - 1, 1699| 75.9k| t->frame_thread.pass != 2 ? top_filter_2d : ------------------ | Branch (1699:30): [True: 0, False: 75.9k] ------------------ 1700| 75.9k| f->frame_thread.b[((t->by - 1) * f->b4_stride) + t->bx].filter2d); 1701| 75.9k| if (res) return res; ------------------ | Branch (1701:25): [True: 0, False: 75.9k] ------------------ 1702| 75.9k| } 1703| 37.9k| v_off = 2 * PXSTRIDE(f->cur.stride[1]); ------------------ | | 53| 37.9k|#define PXSTRIDE(x) (x) ------------------ 1704| 37.9k| } 1705| 214k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1705:30): [True: 143k, False: 71.6k] ------------------ 1706| 143k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff + h_off + v_off, NULL, f->cur.stride[1], 1707| 143k| bw4, bh4, t->bx, t->by, 1 + pl, b->mv[0], 1708| 143k| refp, b->ref[0], filter_2d); 1709| 143k| if (res) return res; ------------------ | Branch (1709:21): [True: 0, False: 143k] ------------------ 1710| 143k| } 1711| 504k| } else { 1712| 504k| if (imin(cbw4, cbh4) > 1 && ------------------ | Branch (1712:17): [True: 254k, False: 250k] ------------------ 1713| 254k| ((b->inter_mode == GLOBALMV && f->gmv_warp_allowed[b->ref[0]]) || ------------------ | Branch (1713:19): [True: 34.3k, False: 219k] | Branch (1713:48): [True: 11.8k, False: 22.4k] ------------------ 1714| 242k| (b->motion_mode == MM_WARP && t->warpmv.type > DAV1D_WM_TYPE_TRANSLATION))) ------------------ | Branch (1714:19): [True: 43.0k, False: 199k] | Branch (1714:48): [True: 41.8k, False: 1.13k] ------------------ 1715| 53.7k| { 1716| 161k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1716:34): [True: 107k, False: 53.7k] ------------------ 1717| 107k| res = warp_affine(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, NULL, 1718| 107k| f->cur.stride[1], b_dim, 1 + pl, refp, 1719| 107k| b->motion_mode == MM_WARP ? &t->warpmv : ------------------ | Branch (1719:39): [True: 83.7k, False: 23.6k] ------------------ 1720| 107k| &f->frame_hdr->gmv[b->ref[0]]); 1721| 107k| if (res) return res; ------------------ | Branch (1721:25): [True: 0, False: 107k] ------------------ 1722| 107k| } 1723| 451k| } else { 1724| 1.35M| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1724:34): [True: 902k, False: 451k] ------------------ 1725| 902k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, 1726| 902k| NULL, f->cur.stride[1], 1727| 902k| bw4 << (bw4 == ss_hor), bh4 << (bh4 == ss_ver), 1728| 902k| t->bx & ~ss_hor, t->by & ~ss_ver, 1729| 902k| 1 + pl, b->mv[0], refp, b->ref[0], filter_2d); 1730| 902k| if (res) return res; ------------------ | Branch (1730:25): [True: 0, False: 902k] ------------------ 1731| 902k| if (b->motion_mode == MM_OBMC) { ------------------ | Branch (1731:25): [True: 296k, False: 606k] ------------------ 1732| 296k| res = obmc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, 1733| 296k| f->cur.stride[1], b_dim, 1 + pl, bx4, by4, w4, h4); 1734| 296k| if (res) return res; ------------------ | Branch (1734:29): [True: 0, False: 296k] ------------------ 1735| 296k| } 1736| 902k| } 1737| 451k| } 1738| 504k| if (b->interintra_type) { ------------------ | Branch (1738:17): [True: 57.5k, False: 447k] ------------------ 1739| | // FIXME for 8x32 with 4:2:2 subsampling, this probably does 1740| | // the wrong thing since it will select 4x16, not 4x32, as a 1741| | // transform size... 1742| 57.5k| const uint8_t *const ii_mask = II_MASK(chr_layout_idx, bs, b); ------------------ | | 83| 57.5k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 84| 57.5k| (size_t)((b)->interintra_type == INTER_INTRA_BLEND ? \ | | ------------------ | | | Branch (84:14): [True: 42.6k, False: 14.8k] | | ------------------ | | 85| 57.5k| dav1d_masks.offsets[c][(bs)-BS_32x32].ii[(b)->interintra_mode] : \ | | 86| 57.5k| dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[0][(b)->wedge_idx]) * 8)) ------------------ 1743| | 1744| 172k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1744:34): [True: 115k, False: 57.5k] ------------------ 1745| 115k| pixel *const tmp = bitfn(t->scratch.interintra); ------------------ | | 51| 115k|#define bitfn(x) x##_8bpc ------------------ 1746| 115k| pixel *const tl_edge = bitfn(t->scratch.edge) + 32; ------------------ | | 51| 115k|#define bitfn(x) x##_8bpc ------------------ 1747| 115k| enum IntraPredMode m = 1748| 115k| b->interintra_mode == II_SMOOTH_PRED ? ------------------ | Branch (1748:25): [True: 16.4k, False: 98.6k] ------------------ 1749| 98.6k| SMOOTH_PRED : b->interintra_mode; 1750| 115k| int angle = 0; 1751| 115k| pixel *const uvdst = ((pixel *) f->cur.data[1 + pl]) + uvdstoff; 1752| 115k| const pixel *top_sb_edge = NULL; 1753| 115k| if (!(t->by & (f->sb_step - 1))) { ------------------ | Branch (1753:25): [True: 17.2k, False: 97.8k] ------------------ 1754| 17.2k| top_sb_edge = f->ipred_edge[pl + 1]; 1755| 17.2k| const int sby = t->by >> f->sb_shift; 1756| 17.2k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1757| 17.2k| } 1758| 115k| m = bytefn(dav1d_prepare_intra_edges)(t->bx >> ss_hor, ------------------ | | 87| 115k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 115k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 1759| 115k| (t->bx >> ss_hor) > 1760| 115k| (ts->tiling.col_start >> ss_hor), 1761| 115k| t->by >> ss_ver, 1762| 115k| (t->by >> ss_ver) > 1763| 115k| (ts->tiling.row_start >> ss_ver), 1764| 115k| ts->tiling.col_end >> ss_hor, 1765| 115k| ts->tiling.row_end >> ss_ver, 1766| 115k| 0, uvdst, f->cur.stride[1], 1767| 115k| top_sb_edge, m, 1768| 115k| &angle, cbw4, cbh4, 0, tl_edge 1769| 115k| HIGHBD_CALL_SUFFIX); 1770| 115k| dsp->ipred.intra_pred[m](tmp, cbw4 * 4 * sizeof(pixel), 1771| 115k| tl_edge, cbw4 * 4, cbh4 * 4, 0, 0, 0 1772| 115k| HIGHBD_CALL_SUFFIX); 1773| 115k| dsp->mc.blend(uvdst, f->cur.stride[1], tmp, 1774| 115k| cbw4 * 4, cbh4 * 4, ii_mask); 1775| 115k| } 1776| 57.5k| } 1777| 504k| } 1778| | 1779| 1.68M| skip_inter_chroma_pred: {} 1780| 1.68M| t->tl_4x4_filter = filter_2d; 1781| 1.68M| } else { 1782| 89.0k| const enum Filter2d filter_2d = b->filter2d; 1783| | // Maximum super block size is 128x128 1784| 89.0k| int16_t (*tmp)[128 * 128] = t->scratch.compinter; 1785| 89.0k| int jnt_weight; 1786| 89.0k| uint8_t *const seg_mask = t->scratch.seg_mask; 1787| 89.0k| const uint8_t *mask; 1788| | 1789| 268k| for (int i = 0; i < 2; i++) { ------------------ | Branch (1789:25): [True: 179k, False: 89.0k] ------------------ 1790| 179k| const Dav1dThreadPicture *const refp = &f->refp[b->ref[i]]; 1791| | 1792| 179k| if (b->inter_mode == GLOBALMV_GLOBALMV && f->gmv_warp_allowed[b->ref[i]]) { ------------------ | Branch (1792:17): [True: 5.66k, False: 173k] | Branch (1792:55): [True: 2.42k, False: 3.24k] ------------------ 1793| 2.41k| res = warp_affine(t, NULL, tmp[i], bw4 * 4, b_dim, 0, refp, 1794| 2.41k| &f->frame_hdr->gmv[b->ref[i]]); 1795| 2.41k| if (res) return res; ------------------ | Branch (1795:21): [True: 0, False: 2.41k] ------------------ 1796| 176k| } else { 1797| 176k| res = mc(t, NULL, tmp[i], 0, bw4, bh4, t->bx, t->by, 0, 1798| 176k| b->mv[i], refp, b->ref[i], filter_2d); 1799| 176k| if (res) return res; ------------------ | Branch (1799:21): [True: 0, False: 176k] ------------------ 1800| 176k| } 1801| 179k| } 1802| 89.0k| switch (b->comp_type) { ------------------ | Branch (1802:17): [True: 89.7k, False: 18.4E] ------------------ 1803| 54.2k| case COMP_INTER_AVG: ------------------ | Branch (1803:9): [True: 54.2k, False: 34.8k] ------------------ 1804| 54.2k| dsp->mc.avg(dst, f->cur.stride[0], tmp[0], tmp[1], 1805| 54.2k| bw4 * 4, bh4 * 4 HIGHBD_CALL_SUFFIX); 1806| 54.2k| break; 1807| 12.5k| case COMP_INTER_WEIGHTED_AVG: ------------------ | Branch (1807:9): [True: 12.5k, False: 76.4k] ------------------ 1808| 12.5k| jnt_weight = f->jnt_weights[b->ref[0]][b->ref[1]]; 1809| 12.5k| dsp->mc.w_avg(dst, f->cur.stride[0], tmp[0], tmp[1], 1810| 12.5k| bw4 * 4, bh4 * 4, jnt_weight HIGHBD_CALL_SUFFIX); 1811| 12.5k| break; 1812| 14.6k| case COMP_INTER_SEG: ------------------ | Branch (1812:9): [True: 14.6k, False: 74.4k] ------------------ 1813| 14.6k| dsp->mc.w_mask[chr_layout_idx](dst, f->cur.stride[0], 1814| 14.6k| tmp[b->mask_sign], tmp[!b->mask_sign], 1815| 14.6k| bw4 * 4, bh4 * 4, seg_mask, 1816| 14.6k| b->mask_sign HIGHBD_CALL_SUFFIX); 1817| 14.6k| mask = seg_mask; 1818| 14.6k| break; 1819| 8.21k| case COMP_INTER_WEDGE: ------------------ | Branch (1819:9): [True: 8.21k, False: 80.8k] ------------------ 1820| 8.21k| mask = WEDGE_MASK(0, bs, 0, b->wedge_idx); ------------------ | | 89| 8.21k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 90| 8.21k| (size_t)dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[sign][idx] * 8)) ------------------ 1821| 8.21k| dsp->mc.mask(dst, f->cur.stride[0], 1822| 8.21k| tmp[b->mask_sign], tmp[!b->mask_sign], 1823| 8.21k| bw4 * 4, bh4 * 4, mask HIGHBD_CALL_SUFFIX); 1824| 8.21k| if (has_chroma) ------------------ | Branch (1824:17): [True: 7.61k, False: 599] ------------------ 1825| 7.61k| mask = WEDGE_MASK(chr_layout_idx, bs, b->mask_sign, b->wedge_idx); ------------------ | | 89| 7.61k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 90| 7.61k| (size_t)dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[sign][idx] * 8)) ------------------ 1826| 8.21k| break; 1827| 89.0k| } 1828| | 1829| | // chroma 1830| 248k| if (has_chroma) for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1830:13): [True: 82.9k, False: 6.56k] | Branch (1830:42): [True: 165k, False: 82.9k] ------------------ 1831| 496k| for (int i = 0; i < 2; i++) { ------------------ | Branch (1831:29): [True: 330k, False: 165k] ------------------ 1832| 330k| const Dav1dThreadPicture *const refp = &f->refp[b->ref[i]]; 1833| 330k| if (b->inter_mode == GLOBALMV_GLOBALMV && ------------------ | Branch (1833:21): [True: 8.34k, False: 322k] ------------------ 1834| 8.34k| imin(cbw4, cbh4) > 1 && f->gmv_warp_allowed[b->ref[i]]) ------------------ | Branch (1834:21): [True: 4.15k, False: 4.19k] | Branch (1834:45): [True: 1.73k, False: 2.41k] ------------------ 1835| 1.73k| { 1836| 1.73k| res = warp_affine(t, NULL, tmp[i], bw4 * 4 >> ss_hor, 1837| 1.73k| b_dim, 1 + pl, 1838| 1.73k| refp, &f->frame_hdr->gmv[b->ref[i]]); 1839| 1.73k| if (res) return res; ------------------ | Branch (1839:25): [True: 0, False: 1.73k] ------------------ 1840| 329k| } else { 1841| 329k| res = mc(t, NULL, tmp[i], 0, bw4, bh4, t->bx, t->by, 1842| 329k| 1 + pl, b->mv[i], refp, b->ref[i], filter_2d); 1843| 329k| if (res) return res; ------------------ | Branch (1843:25): [True: 0, False: 329k] ------------------ 1844| 329k| } 1845| 330k| } 1846| 165k| pixel *const uvdst = ((pixel *) f->cur.data[1 + pl]) + uvdstoff; 1847| 165k| switch (b->comp_type) { ------------------ | Branch (1847:21): [True: 166k, False: 18.4E] ------------------ 1848| 101k| case COMP_INTER_AVG: ------------------ | Branch (1848:13): [True: 101k, False: 63.9k] ------------------ 1849| 101k| dsp->mc.avg(uvdst, f->cur.stride[1], tmp[0], tmp[1], 1850| 101k| bw4 * 4 >> ss_hor, bh4 * 4 >> ss_ver 1851| 101k| HIGHBD_CALL_SUFFIX); 1852| 101k| break; 1853| 22.1k| case COMP_INTER_WEIGHTED_AVG: ------------------ | Branch (1853:13): [True: 22.1k, False: 143k] ------------------ 1854| 22.1k| dsp->mc.w_avg(uvdst, f->cur.stride[1], tmp[0], tmp[1], 1855| 22.1k| bw4 * 4 >> ss_hor, bh4 * 4 >> ss_ver, jnt_weight 1856| 22.1k| HIGHBD_CALL_SUFFIX); 1857| 22.1k| break; 1858| 15.2k| case COMP_INTER_WEDGE: ------------------ | Branch (1858:13): [True: 15.2k, False: 150k] ------------------ 1859| 42.3k| case COMP_INTER_SEG: ------------------ | Branch (1859:13): [True: 27.1k, False: 138k] ------------------ 1860| 42.3k| dsp->mc.mask(uvdst, f->cur.stride[1], 1861| 42.3k| tmp[b->mask_sign], tmp[!b->mask_sign], 1862| 42.3k| bw4 * 4 >> ss_hor, bh4 * 4 >> ss_ver, mask 1863| 42.3k| HIGHBD_CALL_SUFFIX); 1864| 42.3k| break; 1865| 165k| } 1866| 165k| } 1867| 89.4k| } 1868| | 1869| 1.79M| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 1.79M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.79M] | | ------------------ | | 35| 1.79M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.79M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1870| 0| hex_dump(dst, f->cur.stride[0], b_dim[0] * 4, b_dim[1] * 4, "y-pred"); 1871| 0| if (has_chroma) { ------------------ | Branch (1871:13): [True: 0, False: 0] ------------------ 1872| 0| hex_dump(&((pixel *) f->cur.data[1])[uvdstoff], f->cur.stride[1], 1873| 0| cbw4 * 4, cbh4 * 4, "u-pred"); 1874| 0| hex_dump(&((pixel *) f->cur.data[2])[uvdstoff], f->cur.stride[1], 1875| 0| cbw4 * 4, cbh4 * 4, "v-pred"); 1876| 0| } 1877| 0| } 1878| | 1879| 1.79M| const int cw4 = (w4 + ss_hor) >> ss_hor, ch4 = (h4 + ss_ver) >> ss_ver; 1880| | 1881| 1.79M| if (b->skip) { ------------------ | Branch (1881:9): [True: 1.39M, False: 400k] ------------------ 1882| | // reset coef contexts 1883| 1.39M| BlockContext *const a = t->a; 1884| 1.39M| dav1d_memset_pow2[b_dim[2]](&a->lcoef[bx4], 0x40); 1885| 1.39M| dav1d_memset_pow2[b_dim[3]](&t->l.lcoef[by4], 0x40); 1886| 1.39M| if (has_chroma) { ------------------ | Branch (1886:13): [True: 343k, False: 1.04M] ------------------ 1887| 343k| dav1d_memset_pow2_fn memset_cw = dav1d_memset_pow2[ulog2(cbw4)]; 1888| 343k| dav1d_memset_pow2_fn memset_ch = dav1d_memset_pow2[ulog2(cbh4)]; 1889| 343k| memset_cw(&a->ccoef[0][cbx4], 0x40); 1890| 343k| memset_cw(&a->ccoef[1][cbx4], 0x40); 1891| 343k| memset_ch(&t->l.ccoef[0][cby4], 0x40); 1892| 343k| memset_ch(&t->l.ccoef[1][cby4], 0x40); 1893| 343k| } 1894| 1.39M| return 0; 1895| 1.39M| } 1896| | 1897| 400k| const TxfmInfo *const uvtx = &dav1d_txfm_dimensions[b->uvtx]; 1898| 400k| const TxfmInfo *const ytx = &dav1d_txfm_dimensions[b->max_ytx]; 1899| 400k| const uint16_t tx_split[2] = { b->tx_split0, b->tx_split1 }; 1900| | 1901| 804k| for (int init_y = 0; init_y < bh4; init_y += 16) { ------------------ | Branch (1901:26): [True: 403k, False: 400k] ------------------ 1902| 811k| for (int init_x = 0; init_x < bw4; init_x += 16) { ------------------ | Branch (1902:30): [True: 407k, False: 403k] ------------------ 1903| | // coefficient coding & inverse transforms 1904| 407k| int y_off = !!init_y, y; 1905| 407k| dst += PXSTRIDE(f->cur.stride[0]) * 4 * init_y; ------------------ | | 53| 407k|#define PXSTRIDE(x) (x) ------------------ 1906| 820k| for (y = init_y, t->by += init_y; y < imin(h4, init_y + 16); ------------------ | Branch (1906:47): [True: 412k, False: 407k] ------------------ 1907| 412k| y += ytx->h, y_off++) 1908| 412k| { 1909| 412k| int x, x_off = !!init_x; 1910| 832k| for (x = init_x, t->bx += init_x; x < imin(w4, init_x + 16); ------------------ | Branch (1910:51): [True: 419k, False: 412k] ------------------ 1911| 419k| x += ytx->w, x_off++) 1912| 419k| { 1913| 419k| read_coef_tree(t, bs, b, b->max_ytx, 0, tx_split, 1914| 419k| x_off, y_off, &dst[x * 4]); 1915| 419k| t->bx += ytx->w; 1916| 419k| } 1917| 412k| dst += PXSTRIDE(f->cur.stride[0]) * 4 * ytx->h; ------------------ | | 53| 412k|#define PXSTRIDE(x) (x) ------------------ 1918| 412k| t->bx -= x; 1919| 412k| t->by += ytx->h; 1920| 412k| } 1921| 407k| dst -= PXSTRIDE(f->cur.stride[0]) * 4 * y; ------------------ | | 53| 407k|#define PXSTRIDE(x) (x) ------------------ 1922| 407k| t->by -= y; 1923| | 1924| | // chroma coefs and inverse transform 1925| 995k| if (has_chroma) for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1925:17): [True: 332k, False: 75.5k] | Branch (1925:46): [True: 663k, False: 332k] ------------------ 1926| 663k| pixel *uvdst = ((pixel *) f->cur.data[1 + pl]) + uvdstoff + 1927| 663k| (PXSTRIDE(f->cur.stride[1]) * init_y * 4 >> ss_ver); ------------------ | | 53| 663k|#define PXSTRIDE(x) (x) ------------------ 1928| 663k| for (y = init_y >> ss_ver, t->by += init_y; 1929| 1.33M| y < imin(ch4, (init_y + 16) >> ss_ver); y += uvtx->h) ------------------ | Branch (1929:22): [True: 668k, False: 663k] ------------------ 1930| 668k| { 1931| 668k| int x; 1932| 668k| for (x = init_x >> ss_hor, t->bx += init_x; 1933| 1.33M| x < imin(cw4, (init_x + 16) >> ss_hor); x += uvtx->w) ------------------ | Branch (1933:26): [True: 670k, False: 668k] ------------------ 1934| 670k| { 1935| 670k| coef *cf; 1936| 670k| int eob; 1937| 670k| enum TxfmType txtp; 1938| 670k| if (t->frame_thread.pass) { ------------------ | Branch (1938:29): [True: 670k, False: 18.4E] ------------------ 1939| 670k| const int p = t->frame_thread.pass & 1; 1940| 670k| const int cbi = *ts->frame_thread[p].cbi++; 1941| 670k| cf = ts->frame_thread[p].cf; 1942| 670k| ts->frame_thread[p].cf += uvtx->w * uvtx->h * 16; 1943| 670k| eob = cbi >> 5; 1944| 670k| txtp = cbi & 0x1f; 1945| 18.4E| } else { 1946| 18.4E| uint8_t cf_ctx; 1947| 18.4E| cf = bitfn(t->cf); ------------------ | | 51| 18.4E|#define bitfn(x) x##_8bpc ------------------ 1948| 18.4E| txtp = t->scratch.txtp_map[(by4 + (y << ss_ver)) * 32 + 1949| 18.4E| bx4 + (x << ss_hor)]; 1950| 18.4E| eob = decode_coefs(t, &t->a->ccoef[pl][cbx4 + x], 1951| 18.4E| &t->l.ccoef[pl][cby4 + y], 1952| 18.4E| b->uvtx, bs, b, 0, 1 + pl, 1953| 18.4E| cf, &txtp, &cf_ctx); 1954| 18.4E| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 18.4E|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 18.4E] | | ------------------ | | 35| 18.4E| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 18.4E| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1955| 0| printf("Post-uv-cf-blk[pl=%d,tx=%d," 1956| 0| "txtp=%d,eob=%d]: r=%d\n", 1957| 0| pl, b->uvtx, txtp, eob, ts->msac.rng); 1958| 18.4E| int ctw = imin(uvtx->w, (f->bw - t->bx + ss_hor) >> ss_hor); 1959| 18.4E| int cth = imin(uvtx->h, (f->bh - t->by + ss_ver) >> ss_ver); 1960| 18.4E| dav1d_memset_likely_pow2(&t->a->ccoef[pl][cbx4 + x], cf_ctx, ctw); 1961| 18.4E| dav1d_memset_likely_pow2(&t->l.ccoef[pl][cby4 + y], cf_ctx, cth); 1962| 18.4E| } 1963| 670k| if (eob >= 0) { ------------------ | Branch (1963:29): [True: 200k, False: 470k] ------------------ 1964| 200k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 200k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 200k] | | ------------------ | | 35| 200k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 200k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1965| 0| coef_dump(cf, uvtx->h * 4, uvtx->w * 4, 3, "dq"); 1966| 200k| dsp->itx.itxfm_add[b->uvtx] 1967| 200k| [txtp](&uvdst[4 * x], 1968| 200k| f->cur.stride[1], 1969| 200k| cf, eob HIGHBD_CALL_SUFFIX); 1970| 200k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 200k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 200k] | | ------------------ | | 35| 200k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 200k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1971| 0| hex_dump(&uvdst[4 * x], f->cur.stride[1], 1972| 0| uvtx->w * 4, uvtx->h * 4, "recon"); 1973| 200k| } 1974| 670k| t->bx += uvtx->w << ss_hor; 1975| 670k| } 1976| 668k| uvdst += PXSTRIDE(f->cur.stride[1]) * 4 * uvtx->h; ------------------ | | 53| 668k|#define PXSTRIDE(x) (x) ------------------ 1977| 668k| t->bx -= x << ss_hor; 1978| 668k| t->by += uvtx->h << ss_ver; 1979| 668k| } 1980| 663k| t->by -= y << ss_ver; 1981| 663k| } 1982| 407k| } 1983| 403k| } 1984| 400k| return 0; 1985| 1.79M|} dav1d_filter_sbrow_deblock_cols_8bpc: 1987| 518k|void bytefn(dav1d_filter_sbrow_deblock_cols)(Dav1dFrameContext *const f, const int sby) { 1988| 518k| if (!(f->c->inloop_filters & DAV1D_INLOOPFILTER_DEBLOCK) || ------------------ | Branch (1988:9): [True: 3, False: 518k] ------------------ 1989| 518k| (!f->frame_hdr->loopfilter.level_y[0] && !f->frame_hdr->loopfilter.level_y[1])) ------------------ | Branch (1989:10): [True: 5.04k, False: 513k] | Branch (1989:50): [True: 0, False: 5.04k] ------------------ 1990| 0| { 1991| 0| return; 1992| 0| } 1993| 518k| const int y = sby * f->sb_step * 4; 1994| 518k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 1995| 518k| pixel *const p[3] = { 1996| 518k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), ------------------ | | 53| 518k|#define PXSTRIDE(x) (x) ------------------ 1997| 518k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), ------------------ | | 53| 518k|#define PXSTRIDE(x) (x) ------------------ 1998| 518k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) ------------------ | | 53| 518k|#define PXSTRIDE(x) (x) ------------------ 1999| 518k| }; 2000| 518k| Av1Filter *mask = f->lf.mask + (sby >> !f->seq_hdr->sb128) * f->sb128w; 2001| 518k| bytefn(dav1d_loopfilter_sbrow_cols)(f, p, mask, sby, ------------------ | | 87| 518k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 518k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2002| 518k| f->lf.start_of_tile_row[sby]); 2003| 518k|} dav1d_filter_sbrow_deblock_rows_8bpc: 2005| 566k|void bytefn(dav1d_filter_sbrow_deblock_rows)(Dav1dFrameContext *const f, const int sby) { 2006| 566k| const int y = sby * f->sb_step * 4; 2007| 566k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2008| 566k| pixel *const p[3] = { 2009| 566k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), ------------------ | | 53| 566k|#define PXSTRIDE(x) (x) ------------------ 2010| 566k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), ------------------ | | 53| 566k|#define PXSTRIDE(x) (x) ------------------ 2011| 566k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) ------------------ | | 53| 566k|#define PXSTRIDE(x) (x) ------------------ 2012| 566k| }; 2013| 566k| Av1Filter *mask = f->lf.mask + (sby >> !f->seq_hdr->sb128) * f->sb128w; 2014| 566k| if (f->c->inloop_filters & DAV1D_INLOOPFILTER_DEBLOCK && ------------------ | Branch (2014:9): [True: 566k, False: 18.4E] ------------------ 2015| 566k| (f->frame_hdr->loopfilter.level_y[0] || f->frame_hdr->loopfilter.level_y[1])) ------------------ | Branch (2015:10): [True: 513k, False: 52.8k] | Branch (2015:49): [True: 5.02k, False: 47.8k] ------------------ 2016| 518k| { 2017| 518k| bytefn(dav1d_loopfilter_sbrow_rows)(f, p, mask, sby); ------------------ | | 87| 518k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 518k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2018| 518k| } 2019| 566k| if (f->seq_hdr->cdef || f->lf.restore_planes) { ------------------ | Branch (2019:9): [True: 72.3k, False: 494k] | Branch (2019:29): [True: 168k, False: 325k] ------------------ 2020| | // Store loop filtered pixels required by CDEF / LR 2021| 241k| bytefn(dav1d_copy_lpf)(f, p, sby); ------------------ | | 87| 241k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 241k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2022| 241k| } 2023| 566k|} dav1d_filter_sbrow_cdef_8bpc: 2025| 72.3k|void bytefn(dav1d_filter_sbrow_cdef)(Dav1dTaskContext *const tc, const int sby) { 2026| 72.3k| const Dav1dFrameContext *const f = tc->f; 2027| 72.3k| if (!(f->c->inloop_filters & DAV1D_INLOOPFILTER_CDEF)) return; ------------------ | Branch (2027:9): [True: 0, False: 72.3k] ------------------ 2028| 72.3k| const int sbsz = f->sb_step; 2029| 72.3k| const int y = sby * sbsz * 4; 2030| 72.3k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2031| 72.3k| pixel *const p[3] = { 2032| 72.3k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), ------------------ | | 53| 72.3k|#define PXSTRIDE(x) (x) ------------------ 2033| 72.3k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), ------------------ | | 53| 72.3k|#define PXSTRIDE(x) (x) ------------------ 2034| 72.3k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) ------------------ | | 53| 72.3k|#define PXSTRIDE(x) (x) ------------------ 2035| 72.3k| }; 2036| 72.3k| Av1Filter *prev_mask = f->lf.mask + ((sby - 1) >> !f->seq_hdr->sb128) * f->sb128w; 2037| 72.3k| Av1Filter *mask = f->lf.mask + (sby >> !f->seq_hdr->sb128) * f->sb128w; 2038| 72.3k| const int start = sby * sbsz; 2039| 72.3k| if (sby) { ------------------ | Branch (2039:9): [True: 57.5k, False: 14.7k] ------------------ 2040| 57.5k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2041| 57.5k| pixel *p_up[3] = { 2042| 57.5k| p[0] - 8 * PXSTRIDE(f->cur.stride[0]), ------------------ | | 53| 57.5k|#define PXSTRIDE(x) (x) ------------------ 2043| 57.5k| p[1] - (8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver), ------------------ | | 53| 57.5k|#define PXSTRIDE(x) (x) ------------------ 2044| 57.5k| p[2] - (8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver), ------------------ | | 53| 57.5k|#define PXSTRIDE(x) (x) ------------------ 2045| 57.5k| }; 2046| 57.5k| bytefn(dav1d_cdef_brow)(tc, p_up, prev_mask, start - 2, start, 1, sby); ------------------ | | 87| 57.5k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 57.5k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2047| 57.5k| } 2048| 72.3k| const int n_blks = sbsz - 2 * (sby + 1 < f->sbh); 2049| 72.3k| const int end = imin(start + n_blks, f->bh); 2050| 72.3k| bytefn(dav1d_cdef_brow)(tc, p, mask, start, end, 0, sby); ------------------ | | 87| 72.3k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 72.3k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2051| 72.3k|} dav1d_filter_sbrow_resize_8bpc: 2053| 14.0k|void bytefn(dav1d_filter_sbrow_resize)(Dav1dFrameContext *const f, const int sby) { 2054| 14.0k| const int sbsz = f->sb_step; 2055| 14.0k| const int y = sby * sbsz * 4; 2056| 14.0k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2057| 14.0k| const pixel *const p[3] = { 2058| 14.0k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), ------------------ | | 53| 14.0k|#define PXSTRIDE(x) (x) ------------------ 2059| 14.0k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), ------------------ | | 53| 14.0k|#define PXSTRIDE(x) (x) ------------------ 2060| 14.0k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) ------------------ | | 53| 14.0k|#define PXSTRIDE(x) (x) ------------------ 2061| 14.0k| }; 2062| 14.0k| pixel *const sr_p[3] = { 2063| 14.0k| f->lf.sr_p[0] + y * PXSTRIDE(f->sr_cur.p.stride[0]), ------------------ | | 53| 14.0k|#define PXSTRIDE(x) (x) ------------------ 2064| 14.0k| f->lf.sr_p[1] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver), ------------------ | | 53| 14.0k|#define PXSTRIDE(x) (x) ------------------ 2065| 14.0k| f->lf.sr_p[2] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver) ------------------ | | 53| 14.0k|#define PXSTRIDE(x) (x) ------------------ 2066| 14.0k| }; 2067| 14.0k| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400; 2068| 55.5k| for (int pl = 0; pl < 1 + 2 * has_chroma; pl++) { ------------------ | Branch (2068:22): [True: 41.4k, False: 14.0k] ------------------ 2069| 41.4k| const int ss_ver = pl && f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (2069:28): [True: 27.3k, False: 14.0k] | Branch (2069:34): [True: 13.8k, False: 13.4k] ------------------ 2070| 41.4k| const int h_start = 8 * !!sby >> ss_ver; 2071| 41.4k| const ptrdiff_t dst_stride = f->sr_cur.p.stride[!!pl]; 2072| 41.4k| pixel *dst = sr_p[pl] - h_start * PXSTRIDE(dst_stride); ------------------ | | 53| 41.4k|#define PXSTRIDE(x) (x) ------------------ 2073| 41.4k| const ptrdiff_t src_stride = f->cur.stride[!!pl]; 2074| 41.4k| const pixel *src = p[pl] - h_start * PXSTRIDE(src_stride); ------------------ | | 53| 41.4k|#define PXSTRIDE(x) (x) ------------------ 2075| 41.4k| const int h_end = 4 * (sbsz - 2 * (sby + 1 < f->sbh)) >> ss_ver; 2076| 41.4k| const int ss_hor = pl && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (2076:28): [True: 27.3k, False: 14.0k] | Branch (2076:34): [True: 24.3k, False: 2.98k] ------------------ 2077| 41.4k| const int dst_w = (f->sr_cur.p.p.w + ss_hor) >> ss_hor; 2078| 41.4k| const int src_w = (4 * f->bw + ss_hor) >> ss_hor; 2079| 41.4k| const int img_h = (f->cur.p.h - sbsz * 4 * sby + ss_ver) >> ss_ver; 2080| | 2081| 41.4k| f->dsp->mc.resize(dst, dst_stride, src, src_stride, dst_w, 2082| 41.4k| imin(img_h, h_end) + h_start, src_w, 2083| 41.4k| f->resize_step[!!pl], f->resize_start[!!pl] 2084| 41.4k| HIGHBD_CALL_SUFFIX); 2085| 41.4k| } 2086| 14.0k|} dav1d_filter_sbrow_lr_8bpc: 2088| 175k|void bytefn(dav1d_filter_sbrow_lr)(Dav1dFrameContext *const f, const int sby) { 2089| 175k| if (!(f->c->inloop_filters & DAV1D_INLOOPFILTER_RESTORATION)) return; ------------------ | Branch (2089:9): [True: 0, False: 175k] ------------------ 2090| 175k| const int y = sby * f->sb_step * 4; 2091| 175k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2092| 175k| pixel *const sr_p[3] = { 2093| 175k| f->lf.sr_p[0] + y * PXSTRIDE(f->sr_cur.p.stride[0]), ------------------ | | 53| 175k|#define PXSTRIDE(x) (x) ------------------ 2094| 175k| f->lf.sr_p[1] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver), ------------------ | | 53| 175k|#define PXSTRIDE(x) (x) ------------------ 2095| 175k| f->lf.sr_p[2] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver) ------------------ | | 53| 175k|#define PXSTRIDE(x) (x) ------------------ 2096| 175k| }; 2097| 175k| bytefn(dav1d_lr_sbrow)(f, sr_p, sby); ------------------ | | 87| 175k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 175k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2098| 175k|} dav1d_backup_ipred_edge_8bpc: 2111| 984k|void bytefn(dav1d_backup_ipred_edge)(Dav1dTaskContext *const t) { 2112| 984k| const Dav1dFrameContext *const f = t->f; 2113| 984k| Dav1dTileState *const ts = t->ts; 2114| 984k| const int sby = t->by >> f->sb_shift; 2115| 984k| const int sby_off = f->sb128w * 128 * sby; 2116| 984k| const int x_off = ts->tiling.col_start; 2117| | 2118| 984k| const pixel *const y = 2119| 984k| ((const pixel *) f->cur.data[0]) + x_off * 4 + 2120| 984k| ((t->by + f->sb_step) * 4 - 1) * PXSTRIDE(f->cur.stride[0]); ------------------ | | 53| 984k|#define PXSTRIDE(x) (x) ------------------ 2121| 984k| pixel_copy(&f->ipred_edge[0][sby_off + x_off * 4], y, ------------------ | | 47| 984k|#define pixel_copy memcpy ------------------ 2122| 984k| 4 * (ts->tiling.col_end - x_off)); 2123| | 2124| 984k| if (f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) { ------------------ | Branch (2124:9): [True: 101k, False: 883k] ------------------ 2125| 101k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2126| 101k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 2127| | 2128| 101k| const ptrdiff_t uv_off = (x_off * 4 >> ss_hor) + 2129| 101k| (((t->by + f->sb_step) * 4 >> ss_ver) - 1) * PXSTRIDE(f->cur.stride[1]); ------------------ | | 53| 101k|#define PXSTRIDE(x) (x) ------------------ 2130| 302k| for (int pl = 1; pl <= 2; pl++) ------------------ | Branch (2130:26): [True: 201k, False: 101k] ------------------ 2131| 201k| pixel_copy(&f->ipred_edge[pl][sby_off + (x_off * 4 >> ss_hor)], ------------------ | | 47| 201k|#define pixel_copy memcpy ------------------ 2132| 201k| &((const pixel *) f->cur.data[pl])[uv_off], 2133| 201k| 4 * (ts->tiling.col_end - x_off) >> ss_hor); 2134| 101k| } 2135| 984k|} dav1d_copy_pal_block_y_8bpc: 2141| 112k|{ 2142| 112k| const Dav1dFrameContext *const f = t->f; 2143| 112k| pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (2143:24): [True: 112k, False: 1] ------------------ 2144| 112k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2145| 112k| ((t->bx >> 1) + (t->by & 1))][0] : 2146| 112k| bytefn(t->scratch.pal)[0]; ------------------ | | 87| 1|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 112k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2147| 583k| for (int x = 0; x < bw4; x++) ------------------ | Branch (2147:21): [True: 470k, False: 112k] ------------------ 2148| 470k| memcpy(bytefn(t->al_pal)[0][bx4 + x][0], pal, 8 * sizeof(pixel)); ------------------ | | 87| 470k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 470k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2149| 483k| for (int y = 0; y < bh4; y++) ------------------ | Branch (2149:21): [True: 370k, False: 112k] ------------------ 2150| 370k| memcpy(bytefn(t->al_pal)[1][by4 + y][0], pal, 8 * sizeof(pixel)); ------------------ | | 87| 370k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 370k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2151| 112k|} dav1d_copy_pal_block_uv_8bpc: 2157| 20.1k|{ 2158| 20.1k| const Dav1dFrameContext *const f = t->f; 2159| 20.1k| const pixel (*const pal)[8] = t->frame_thread.pass ? ------------------ | Branch (2159:35): [True: 20.1k, False: 0] ------------------ 2160| 20.1k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2161| 20.1k| ((t->bx >> 1) + (t->by & 1))] : 2162| 20.1k| bytefn(t->scratch.pal); ------------------ | | 87| 0|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 0|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2163| | // see aomedia bug 2183 for why we use luma coordinates here 2164| 60.3k| for (int pl = 1; pl <= 2; pl++) { ------------------ | Branch (2164:22): [True: 40.2k, False: 20.1k] ------------------ 2165| 260k| for (int x = 0; x < bw4; x++) ------------------ | Branch (2165:25): [True: 220k, False: 40.2k] ------------------ 2166| 220k| memcpy(bytefn(t->al_pal)[0][bx4 + x][pl], pal[pl], 8 * sizeof(pixel)); ------------------ | | 87| 220k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 220k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2167| 260k| for (int y = 0; y < bh4; y++) ------------------ | Branch (2167:25): [True: 220k, False: 40.2k] ------------------ 2168| 220k| memcpy(bytefn(t->al_pal)[1][by4 + y][pl], pal[pl], 8 * sizeof(pixel)); ------------------ | | 87| 220k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 220k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2169| 40.2k| } 2170| 20.1k|} dav1d_read_pal_plane_8bpc: 2175| 132k|{ 2176| 132k| Dav1dTileState *const ts = t->ts; 2177| 132k| const Dav1dFrameContext *const f = t->f; 2178| 132k| const int pal_sz = b->pal_sz[pl] = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 132k|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 2179| 132k| ts->cdf.m.pal_sz[pl][sz_ctx], 6) + 2; 2180| 132k| pixel cache[16], used_cache[8]; 2181| 132k| int l_cache = pl ? t->pal_sz_uv[1][by4] : t->l.pal_sz[by4]; ------------------ | Branch (2181:19): [True: 20.1k, False: 112k] ------------------ 2182| 132k| int n_cache = 0; 2183| | // don't reuse above palette outside SB64 boundaries 2184| 132k| int a_cache = by4 & 15 ? pl ? t->pal_sz_uv[0][bx4] : t->a->pal_sz[bx4] : 0; ------------------ | Branch (2184:19): [True: 105k, False: 26.6k] | Branch (2184:30): [True: 11.8k, False: 93.9k] ------------------ 2185| 132k| const pixel *l = bytefn(t->al_pal)[1][by4][pl]; ------------------ | | 87| 132k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 132k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2186| 132k| const pixel *a = bytefn(t->al_pal)[0][bx4][pl]; ------------------ | | 87| 132k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 132k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2187| | 2188| | // fill/sort cache 2189| 260k| while (l_cache && a_cache) { ------------------ | Branch (2189:12): [True: 168k, False: 91.6k] | Branch (2189:23): [True: 128k, False: 40.8k] ------------------ 2190| 128k| if (*l < *a) { ------------------ | Branch (2190:13): [True: 36.4k, False: 91.5k] ------------------ 2191| 36.4k| if (!n_cache || cache[n_cache - 1] != *l) ------------------ | Branch (2191:17): [True: 7.46k, False: 29.0k] | Branch (2191:29): [True: 28.9k, False: 99] ------------------ 2192| 36.3k| cache[n_cache++] = *l; 2193| 36.4k| l++; 2194| 36.4k| l_cache--; 2195| 91.5k| } else { 2196| 91.5k| if (*a == *l) { ------------------ | Branch (2196:17): [True: 47.9k, False: 43.5k] ------------------ 2197| 47.9k| l++; 2198| 47.9k| l_cache--; 2199| 47.9k| } 2200| 91.5k| if (!n_cache || cache[n_cache - 1] != *a) ------------------ | Branch (2200:17): [True: 14.6k, False: 76.8k] | Branch (2200:29): [True: 71.8k, False: 5.03k] ------------------ 2201| 86.4k| cache[n_cache++] = *a; 2202| 91.5k| a++; 2203| 91.5k| a_cache--; 2204| 91.5k| } 2205| 128k| } 2206| 132k| if (l_cache) { ------------------ | Branch (2206:9): [True: 40.8k, False: 91.6k] ------------------ 2207| 162k| do { 2208| 162k| if (!n_cache || cache[n_cache - 1] != *l) ------------------ | Branch (2208:17): [True: 33.3k, False: 128k] | Branch (2208:29): [True: 109k, False: 18.9k] ------------------ 2209| 143k| cache[n_cache++] = *l; 2210| 162k| l++; 2211| 162k| } while (--l_cache > 0); ------------------ | Branch (2211:18): [True: 121k, False: 40.8k] ------------------ 2212| 91.6k| } else if (a_cache) { ------------------ | Branch (2212:16): [True: 41.3k, False: 50.2k] ------------------ 2213| 175k| do { 2214| 175k| if (!n_cache || cache[n_cache - 1] != *a) ------------------ | Branch (2214:17): [True: 27.8k, False: 147k] | Branch (2214:29): [True: 91.7k, False: 55.4k] ------------------ 2215| 119k| cache[n_cache++] = *a; 2216| 175k| a++; 2217| 175k| } while (--a_cache > 0); ------------------ | Branch (2217:18): [True: 133k, False: 41.3k] ------------------ 2218| 41.3k| } 2219| | 2220| | // find reused cache entries 2221| 132k| int i = 0; 2222| 453k| for (int n = 0; n < n_cache && i < pal_sz; n++) ------------------ | Branch (2222:21): [True: 338k, False: 115k] | Branch (2222:36): [True: 320k, False: 17.4k] ------------------ 2223| 320k| if (dav1d_msac_decode_bool_equi(&ts->msac)) ------------------ | | 53| 320k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (2223:13): [True: 174k, False: 146k] ------------------ 2224| 174k| used_cache[i++] = cache[n]; 2225| 132k| const int n_used_cache = i; 2226| | 2227| | // parse new entries 2228| 132k| pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (2228:24): [True: 132k, False: 8] ------------------ 2229| 132k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2230| 132k| ((t->bx >> 1) + (t->by & 1))][pl] : 2231| 132k| bytefn(t->scratch.pal)[pl]; ------------------ | | 87| 8|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 132k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2232| 132k| if (i < pal_sz) { ------------------ | Branch (2232:9): [True: 112k, False: 19.7k] ------------------ 2233| 112k| const int bpc = BITDEPTH == 8 ? 8 : f->cur.p.bpc; ------------------ | Branch (2233:25): [True: 112k, Folded] ------------------ 2234| 112k| int prev = pal[i++] = dav1d_msac_decode_bools(&ts->msac, bpc); 2235| | 2236| 112k| if (i < pal_sz) { ------------------ | Branch (2236:13): [True: 94.4k, False: 18.3k] ------------------ 2237| 94.4k| int bits = bpc - 3 + dav1d_msac_decode_bools(&ts->msac, 2); 2238| 94.4k| const int max = (1 << bpc) - 1; 2239| | 2240| 206k| do { 2241| 206k| const int delta = dav1d_msac_decode_bools(&ts->msac, bits); 2242| 206k| prev = pal[i++] = imin(prev + delta + !pl, max); 2243| 206k| if (prev + !pl >= max) { ------------------ | Branch (2243:21): [True: 56.8k, False: 149k] ------------------ 2244| 146k| for (; i < pal_sz; i++) ------------------ | Branch (2244:28): [True: 90.0k, False: 56.8k] ------------------ 2245| 90.0k| pal[i] = max; 2246| 56.8k| break; 2247| 56.8k| } 2248| 149k| bits = imin(bits, 1 + ulog2(max - prev - !pl)); 2249| 149k| } while (i < pal_sz); ------------------ | Branch (2249:22): [True: 111k, False: 37.6k] ------------------ 2250| 94.4k| } 2251| | 2252| | // merge cache+new entries 2253| 112k| int n = 0, m = n_used_cache; 2254| 642k| for (i = 0; i < pal_sz; i++) { ------------------ | Branch (2254:21): [True: 529k, False: 112k] ------------------ 2255| 529k| if (n < n_used_cache && (m >= pal_sz || used_cache[n] <= pal[m])) { ------------------ | Branch (2255:17): [True: 214k, False: 315k] | Branch (2255:38): [True: 42.9k, False: 171k] | Branch (2255:53): [True: 77.6k, False: 94.1k] ------------------ 2256| 120k| pal[i] = used_cache[n++]; 2257| 409k| } else { 2258| 409k| assert(m < pal_sz); ------------------ | Branch (2258:17): [True: 409k, False: 18.4E] ------------------ 2259| 409k| pal[i] = pal[m++]; 2260| 409k| } 2261| 529k| } 2262| 112k| } else { 2263| 19.7k| memcpy(pal, used_cache, n_used_cache * sizeof(*used_cache)); 2264| 19.7k| } 2265| | 2266| 132k| if (DEBUG_BLOCK_INFO) { ------------------ | | 34| 132k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 132k] | | ------------------ | | 35| 132k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 132k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2267| 0| printf("Post-pal[pl=%d,sz=%d,cache_size=%d,used_cache=%d]: r=%d, cache=", 2268| 0| pl, pal_sz, n_cache, n_used_cache, ts->msac.rng); 2269| 0| for (int n = 0; n < n_cache; n++) ------------------ | Branch (2269:25): [True: 0, False: 0] ------------------ 2270| 0| printf("%c%02x", n ? ' ' : '[', cache[n]); ------------------ | Branch (2270:30): [True: 0, False: 0] ------------------ 2271| 0| printf("%s, pal=", n_cache ? "]" : "[]"); ------------------ | Branch (2271:28): [True: 0, False: 0] ------------------ 2272| 0| for (int n = 0; n < pal_sz; n++) ------------------ | Branch (2272:25): [True: 0, False: 0] ------------------ 2273| 0| printf("%c%02x", n ? ' ' : '[', pal[n]); ------------------ | Branch (2273:30): [True: 0, False: 0] ------------------ 2274| 0| printf("]\n"); 2275| 0| } 2276| 132k|} dav1d_read_pal_uv_8bpc: 2280| 20.1k|{ 2281| 20.1k| bytefn(dav1d_read_pal_plane)(t, b, 1, sz_ctx, bx4, by4); ------------------ | | 87| 20.1k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 20.1k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2282| | 2283| | // V pal coding 2284| 20.1k| Dav1dTileState *const ts = t->ts; 2285| 20.1k| const Dav1dFrameContext *const f = t->f; 2286| 20.1k| pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (2286:24): [True: 20.1k, False: 1] ------------------ 2287| 20.1k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2288| 20.1k| ((t->bx >> 1) + (t->by & 1))][2] : 2289| 20.1k| bytefn(t->scratch.pal)[2]; ------------------ | | 87| 1|#define bytefn(x) bitfn(x) | | ------------------ | | | | 51| 20.1k|#define bitfn(x) x##_8bpc | | ------------------ ------------------ 2290| 20.1k| const int bpc = BITDEPTH == 8 ? 8 : f->cur.p.bpc; ------------------ | Branch (2290:21): [True: 20.1k, Folded] ------------------ 2291| 20.1k| if (dav1d_msac_decode_bool_equi(&ts->msac)) { ------------------ | | 53| 20.1k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (2291:9): [True: 11.2k, False: 8.87k] ------------------ 2292| 11.2k| const int bits = bpc - 4 + dav1d_msac_decode_bools(&ts->msac, 2); 2293| 11.2k| int prev = pal[0] = dav1d_msac_decode_bools(&ts->msac, bpc); 2294| 11.2k| const int max = (1 << bpc) - 1; 2295| 45.2k| for (int i = 1; i < b->pal_sz[1]; i++) { ------------------ | Branch (2295:25): [True: 33.9k, False: 11.2k] ------------------ 2296| 33.9k| int delta = dav1d_msac_decode_bools(&ts->msac, bits); 2297| 33.9k| if (delta && dav1d_msac_decode_bool_equi(&ts->msac)) delta = -delta; ------------------ | | 53| 33.5k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (2297:17): [True: 33.5k, False: 423] | Branch (2297:26): [True: 18.6k, False: 14.8k] ------------------ 2298| 33.9k| prev = pal[i] = (prev + delta) & max; 2299| 33.9k| } 2300| 11.2k| } else { 2301| 49.2k| for (int i = 0; i < b->pal_sz[1]; i++) ------------------ | Branch (2301:25): [True: 40.3k, False: 8.87k] ------------------ 2302| 40.3k| pal[i] = dav1d_msac_decode_bools(&ts->msac, bpc); 2303| 8.87k| } 2304| 20.1k| if (DEBUG_BLOCK_INFO) { ------------------ | | 34| 20.1k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 20.1k] | | ------------------ | | 35| 20.1k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 20.1k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2305| 0| printf("Post-pal[pl=2]: r=%d ", ts->msac.rng); 2306| 0| for (int n = 0; n < b->pal_sz[1]; n++) ------------------ | Branch (2306:25): [True: 0, False: 0] ------------------ 2307| 0| printf("%c%02x", n ? ' ' : '[', pal[n]); ------------------ | Branch (2307:30): [True: 0, False: 0] ------------------ 2308| 0| printf("]\n"); 2309| 0| } 2310| 20.1k|} recon_tmpl.c:read_coef_tree: 736| 7.57M|{ 737| 7.57M| const Dav1dFrameContext *const f = t->f; 738| 7.57M| Dav1dTileState *const ts = t->ts; 739| 7.57M| const Dav1dDSPContext *const dsp = f->dsp; 740| 7.57M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[ytx]; 741| 7.57M| const int txw = t_dim->w, txh = t_dim->h; 742| | 743| | /* y_off can be larger than 3 since lossless blocks use TX_4X4 but can't 744| | * be splitted. Aviods an undefined left shift. */ 745| 7.57M| if (depth < 2 && tx_split[depth] && ------------------ | Branch (745:9): [True: 7.24M, False: 327k] | Branch (745:22): [True: 428k, False: 6.81M] ------------------ 746| 428k| tx_split[depth] & (1 << (y_off * 4 + x_off))) ------------------ | Branch (746:9): [True: 350k, False: 78.6k] ------------------ 747| 350k| { 748| 350k| const enum RectTxfmSize sub = t_dim->sub; 749| 350k| const TxfmInfo *const sub_t_dim = &dav1d_txfm_dimensions[sub]; 750| 350k| const int txsw = sub_t_dim->w, txsh = sub_t_dim->h; 751| | 752| 350k| read_coef_tree(t, bs, b, sub, depth + 1, tx_split, 753| 350k| x_off * 2 + 0, y_off * 2 + 0, dst); 754| 350k| t->bx += txsw; 755| 350k| if (txw >= txh && t->bx < f->bw) ------------------ | Branch (755:13): [True: 269k, False: 80.3k] | Branch (755:27): [True: 268k, False: 1.02k] ------------------ 756| 268k| read_coef_tree(t, bs, b, sub, depth + 1, tx_split, x_off * 2 + 1, 757| 268k| y_off * 2 + 0, dst ? &dst[4 * txsw] : NULL); ------------------ | Branch (757:43): [True: 100k, False: 168k] ------------------ 758| 350k| t->bx -= txsw; 759| 350k| t->by += txsh; 760| 350k| if (txh >= txw && t->by < f->bh) { ------------------ | Branch (760:13): [True: 256k, False: 93.7k] | Branch (760:27): [True: 254k, False: 1.53k] ------------------ 761| 254k| if (dst) ------------------ | Branch (761:17): [True: 96.7k, False: 158k] ------------------ 762| 96.7k| dst += 4 * txsh * PXSTRIDE(f->cur.stride[0]); ------------------ | | 53| 96.7k|#define PXSTRIDE(x) (x) ------------------ 763| 254k| read_coef_tree(t, bs, b, sub, depth + 1, tx_split, 764| 254k| x_off * 2 + 0, y_off * 2 + 1, dst); 765| 254k| t->bx += txsw; 766| 254k| if (txw >= txh && t->bx < f->bw) ------------------ | Branch (766:17): [True: 175k, False: 79.8k] | Branch (766:31): [True: 174k, False: 999] ------------------ 767| 174k| read_coef_tree(t, bs, b, sub, depth + 1, tx_split, x_off * 2 + 1, 768| 174k| y_off * 2 + 1, dst ? &dst[4 * txsw] : NULL); ------------------ | Branch (768:47): [True: 62.9k, False: 111k] ------------------ 769| 254k| t->bx -= txsw; 770| 254k| } 771| 350k| t->by -= txsh; 772| 7.22M| } else { 773| 7.22M| const int bx4 = t->bx & 31, by4 = t->by & 31; 774| 7.22M| enum TxfmType txtp; 775| 7.22M| uint8_t cf_ctx; 776| 7.22M| int eob; 777| 7.22M| coef *cf; 778| | 779| 7.22M| if (t->frame_thread.pass) { ------------------ | Branch (779:13): [True: 7.22M, False: 18.4E] ------------------ 780| 7.22M| const int p = t->frame_thread.pass & 1; 781| 7.22M| assert(ts->frame_thread[p].cf); ------------------ | Branch (781:13): [True: 7.22M, False: 18.4E] ------------------ 782| 7.22M| cf = ts->frame_thread[p].cf; 783| 7.22M| ts->frame_thread[p].cf += imin(t_dim->w, 8) * imin(t_dim->h, 8) * 16; 784| 18.4E| } else { 785| 18.4E| cf = bitfn(t->cf); ------------------ | | 51| 18.4E|#define bitfn(x) x##_8bpc ------------------ 786| 18.4E| } 787| 7.22M| if (t->frame_thread.pass != 2) { ------------------ | Branch (787:13): [True: 6.36M, False: 859k] ------------------ 788| 6.36M| eob = decode_coefs(t, &t->a->lcoef[bx4], &t->l.lcoef[by4], 789| 6.36M| ytx, bs, b, 0, 0, cf, &txtp, &cf_ctx); 790| 6.36M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 6.36M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 6.36M] | | ------------------ | | 35| 6.36M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 6.36M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 791| 0| printf("Post-y-cf-blk[tx=%d,txtp=%d,eob=%d]: r=%d\n", 792| 0| ytx, txtp, eob, ts->msac.rng); 793| 6.36M| dav1d_memset_likely_pow2(&t->a->lcoef[bx4], cf_ctx, imin(txw, f->bw - t->bx)); 794| 6.36M| dav1d_memset_likely_pow2(&t->l.lcoef[by4], cf_ctx, imin(txh, f->bh - t->by)); 795| 6.36M|#define set_ctx(rep_macro) \ 796| 6.36M| for (int y = 0; y < txh; y++) { \ 797| 6.36M| rep_macro(txtp_map, 0, txtp); \ 798| 6.36M| txtp_map += 32; \ 799| 6.36M| } 800| 6.36M| uint8_t *txtp_map = &t->scratch.txtp_map[by4 * 32 + bx4]; 801| 6.36M| case_set_upto16(t_dim->lw); ------------------ | | 80| 6.36M| switch (var) { \ | | 81| 4.49M| case 0: set_ctx(set_ctx1); break; \ | | ------------------ | | | | 796| 9.29M| for (int y = 0; y < txh; y++) { \ | | | | ------------------ | | | | | Branch (796:29): [True: 4.79M, False: 4.49M] | | | | ------------------ | | | | 797| 4.79M| rep_macro(txtp_map, 0, txtp); \ | | | | ------------------ | | | | | | 81| 4.79M| case 0: set_ctx(set_ctx1); break; \ | | | | | | ------------------ | | | | | | | | 56| 4.79M| ((union alias8 *) &(var)[off])->u8 = (val) * 0x01 | | | | | | ------------------ | | | | ------------------ | | | | 798| 4.79M| txtp_map += 32; \ | | | | 799| 4.79M| } | | ------------------ | | | Branch (81:5): [True: 4.49M, False: 1.86M] | | ------------------ | | 82| 790k| case 1: set_ctx(set_ctx2); break; \ | | ------------------ | | | | 796| 2.76M| for (int y = 0; y < txh; y++) { \ | | | | ------------------ | | | | | Branch (796:29): [True: 1.97M, False: 790k] | | | | ------------------ | | | | 797| 1.97M| rep_macro(txtp_map, 0, txtp); \ | | | | ------------------ | | | | | | 82| 1.97M| case 1: set_ctx(set_ctx2); break; \ | | | | | | ------------------ | | | | | | | | 58| 1.97M| ((union alias16 *) &(var)[off])->u16 = (val) * 0x0101 | | | | | | ------------------ | | | | ------------------ | | | | 798| 1.97M| txtp_map += 32; \ | | | | 799| 1.97M| } | | ------------------ | | | Branch (82:5): [True: 790k, False: 5.57M] | | ------------------ | | 83| 690k| case 2: set_ctx(set_ctx4); break; \ | | ------------------ | | | | 796| 2.98M| for (int y = 0; y < txh; y++) { \ | | | | ------------------ | | | | | Branch (796:29): [True: 2.29M, False: 690k] | | | | ------------------ | | | | 797| 2.29M| rep_macro(txtp_map, 0, txtp); \ | | | | ------------------ | | | | | | 83| 2.29M| case 2: set_ctx(set_ctx4); break; \ | | | | | | ------------------ | | | | | | | | 60| 2.29M| ((union alias32 *) &(var)[off])->u32 = (val) * 0x01010101U | | | | | | ------------------ | | | | ------------------ | | | | 798| 2.29M| txtp_map += 32; \ | | | | 799| 2.29M| } | | ------------------ | | | Branch (83:5): [True: 690k, False: 5.67M] | | ------------------ | | 84| 226k| case 3: set_ctx(set_ctx8); break; \ | | ------------------ | | | | 796| 1.54M| for (int y = 0; y < txh; y++) { \ | | | | ------------------ | | | | | Branch (796:29): [True: 1.32M, False: 226k] | | | | ------------------ | | | | 797| 1.32M| rep_macro(txtp_map, 0, txtp); \ | | | | ------------------ | | | | | | 84| 1.32M| case 3: set_ctx(set_ctx8); break; \ | | | | | | ------------------ | | | | | | | | 62| 1.32M| ((union alias64 *) &(var)[off])->u64 = (val) * 0x0101010101010101ULL | | | | | | ------------------ | | | | ------------------ | | | | 798| 1.32M| txtp_map += 32; \ | | | | 799| 1.32M| } | | ------------------ | | | Branch (84:5): [True: 226k, False: 6.13M] | | ------------------ | | 85| 164k| case 4: set_ctx(set_ctx16); break; \ | | ------------------ | | | | 796| 2.45M| for (int y = 0; y < txh; y++) { \ | | | | ------------------ | | | | | Branch (796:29): [True: 2.28M, False: 164k] | | | | ------------------ | | | | 797| 2.28M| rep_macro(txtp_map, 0, txtp); \ | | | | ------------------ | | | | | | 85| 2.28M| case 4: set_ctx(set_ctx16); break; \ | | | | | | ------------------ | | | | | | | | 63| 2.28M|#define set_ctx16(var, off, val) do { \ | | | | | | | | 64| 2.28M| memset(&(var)[off], val, 16); \ | | | | | | | | 65| 2.28M| } while (0) | | | | | | | | ------------------ | | | | | | | | | Branch (65:14): [Folded, False: 2.28M] | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 798| 2.28M| txtp_map += 32; \ | | | | 799| 2.28M| } | | ------------------ | | | Branch (85:5): [True: 164k, False: 6.19M] | | ------------------ | | 86| 0| default: assert(0); \ | | ------------------ | | | Branch (86:5): [True: 0, False: 6.36M] | | ------------------ | | 87| 6.36M| } ------------------ | Branch (801:13): [Folded, False: 0] ------------------ 802| 6.36M|#undef set_ctx 803| 6.36M| if (t->frame_thread.pass == 1) ------------------ | Branch (803:17): [True: 6.36M, False: 66] ------------------ 804| 6.36M| *ts->frame_thread[1].cbi++ = eob * (1 << 5) + txtp; 805| 6.36M| } else { 806| 859k| const int cbi = *ts->frame_thread[0].cbi++; 807| 859k| eob = cbi >> 5; 808| 859k| txtp = cbi & 0x1f; 809| 859k| } 810| 7.22M| if (!(t->frame_thread.pass & 1)) { ------------------ | Branch (810:13): [True: 868k, False: 6.35M] ------------------ 811| 868k| assert(dst); ------------------ | Branch (811:13): [True: 868k, False: 18.4E] ------------------ 812| 868k| if (eob >= 0) { ------------------ | Branch (812:17): [True: 700k, False: 168k] ------------------ 813| 700k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 700k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 700k] | | ------------------ | | 35| 700k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 700k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 814| 0| coef_dump(cf, imin(t_dim->h, 8) * 4, imin(t_dim->w, 8) * 4, 3, "dq"); 815| 700k| dsp->itx.itxfm_add[ytx][txtp](dst, f->cur.stride[0], cf, eob 816| 700k| HIGHBD_CALL_SUFFIX); 817| 700k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 700k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 700k] | | ------------------ | | 35| 700k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 700k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 818| 0| hex_dump(dst, f->cur.stride[0], t_dim->w * 4, t_dim->h * 4, "recon"); 819| 700k| } 820| 868k| } 821| 7.22M| } 822| 7.57M|} recon_tmpl.c:decode_coefs: 327| 49.9M|{ 328| 49.9M| Dav1dTileState *const ts = t->ts; 329| 49.9M| const int chroma = !!plane; 330| 49.9M| const Dav1dFrameContext *const f = t->f; 331| 49.9M| const int lossless = f->frame_hdr->segmentation.lossless[b->seg_id]; 332| 49.9M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[tx]; 333| 49.9M| const int dbg = DEBUG_BLOCK_INFO && plane && 0; ------------------ | | 34| 49.9M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 49.9M] | | ------------------ | | 35| 49.9M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 49.9M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ | Branch (333:41): [True: 0, False: 0] | Branch (333:50): [Folded, False: 0] ------------------ 334| | 335| 49.9M| if (dbg) ------------------ | Branch (335:9): [Folded, False: 49.9M] ------------------ 336| 0| printf("Start: r=%d\n", ts->msac.rng); 337| | 338| | // does this block have any non-zero coefficients 339| 49.9M| const int sctx = get_skip_ctx(t_dim, bs, a, l, chroma, f->cur.p.layout); 340| 49.9M| const int all_skip = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 49.9M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 341| 49.9M| ts->cdf.coef.skip[t_dim->ctx][sctx]); 342| 49.9M| if (dbg) ------------------ | Branch (342:9): [Folded, False: 49.9M] ------------------ 343| 0| printf("Post-non-zero[%d][%d][%d]: r=%d\n", 344| 0| t_dim->ctx, sctx, all_skip, ts->msac.rng); 345| 49.9M| if (all_skip) { ------------------ | Branch (345:9): [True: 32.2M, False: 17.7M] ------------------ 346| 32.2M| *res_ctx = 0x40; 347| 32.2M| *txtp = lossless * WHT_WHT; /* lossless ? WHT_WHT : DCT_DCT */ 348| 32.2M| return -1; 349| 32.2M| } 350| | 351| | // transform type (chroma: derived, luma: explicitly coded) 352| 17.7M| if (lossless) { ------------------ | Branch (352:9): [True: 7.60M, False: 10.1M] ------------------ 353| 7.60M| assert(t_dim->max == TX_4X4); ------------------ | Branch (353:9): [True: 7.60M, False: 18.4E] ------------------ 354| 7.60M| *txtp = WHT_WHT; 355| 10.1M| } else if (t_dim->max + intra >= TX_64X64) { ------------------ | Branch (355:16): [True: 2.72M, False: 7.45M] ------------------ 356| 2.72M| *txtp = DCT_DCT; 357| 7.45M| } else if (chroma) { ------------------ | Branch (357:16): [True: 2.50M, False: 4.95M] ------------------ 358| | // inferred from either the luma txtp (inter) or a LUT (intra) 359| 2.50M| *txtp = intra ? dav1d_txtp_from_uvmode[b->uv_mode] : ------------------ | Branch (359:17): [True: 1.65M, False: 851k] ------------------ 360| 2.50M| get_uv_inter_txtp(t_dim, *txtp); 361| 4.95M| } else if (!f->frame_hdr->segmentation.qidx[b->seg_id]) { ------------------ | Branch (361:16): [True: 6.60k, False: 4.94M] ------------------ 362| | // In libaom, lossless is checked by a literal qidx == 0, but not all 363| | // such blocks are actually lossless. The remainder gets an implicit 364| | // transform type (for luma) 365| 6.60k| *txtp = DCT_DCT; 366| 4.94M| } else { 367| 4.94M| unsigned idx; 368| 4.94M| if (intra) { ------------------ | Branch (368:13): [True: 3.49M, False: 1.44M] ------------------ 369| 3.49M| const enum IntraPredMode y_mode_nofilt = b->y_mode == FILTER_PRED ? ------------------ | Branch (369:54): [True: 663k, False: 2.83M] ------------------ 370| 2.83M| dav1d_filter_mode_to_y_mode[b->y_angle] : b->y_mode; 371| 3.49M| if (f->frame_hdr->reduced_txtp_set || t_dim->min == TX_16X16) { ------------------ | Branch (371:17): [True: 547k, False: 2.94M] | Branch (371:51): [True: 471k, False: 2.47M] ------------------ 372| 1.02M| idx = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 1.02M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 373| 1.02M| ts->cdf.m.txtp_intra2[t_dim->min][y_mode_nofilt], 4); 374| 1.02M| *txtp = dav1d_tx_types_per_set[idx + 0]; 375| 2.47M| } else { 376| 2.47M| idx = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 2.47M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 377| 2.47M| ts->cdf.m.txtp_intra1[t_dim->min][y_mode_nofilt], 6); 378| 2.47M| *txtp = dav1d_tx_types_per_set[idx + 5]; 379| 2.47M| } 380| 3.49M| if (dbg) ------------------ | Branch (380:17): [Folded, False: 3.49M] ------------------ 381| 0| printf("Post-txtp-intra[%d->%d][%d][%d->%d]: r=%d\n", 382| 0| tx, t_dim->min, y_mode_nofilt, idx, *txtp, ts->msac.rng); 383| 3.49M| } else { 384| 1.75M| if (f->frame_hdr->reduced_txtp_set || t_dim->max == TX_32X32) { ------------------ | Branch (384:17): [True: 18.4E, False: 1.75M] | Branch (384:51): [True: 267k, False: 1.49M] ------------------ 385| 379k| idx = dav1d_msac_decode_bool_adapt(&ts->msac, ------------------ | | 52| 379k|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 386| 379k| ts->cdf.m.txtp_inter3[t_dim->min]); 387| 379k| *txtp = (idx - 1) & IDTX; /* idx ? DCT_DCT : IDTX */ 388| 1.07M| } else if (t_dim->min == TX_16X16) { ------------------ | Branch (388:24): [True: 215k, False: 854k] ------------------ 389| 215k| idx = dav1d_msac_decode_symbol_adapt16(&ts->msac, ------------------ | | 57| 215k|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 390| 215k| ts->cdf.m.txtp_inter2, 11); 391| 215k| *txtp = dav1d_tx_types_per_set[idx + 12]; 392| 854k| } else { 393| 854k| idx = dav1d_msac_decode_symbol_adapt16(&ts->msac, ------------------ | | 57| 854k|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) ------------------ 394| 854k| ts->cdf.m.txtp_inter1[t_dim->min], 15); 395| 854k| *txtp = dav1d_tx_types_per_set[idx + 24]; 396| 854k| } 397| 1.44M| if (dbg) ------------------ | Branch (397:17): [Folded, False: 1.44M] ------------------ 398| 0| printf("Post-txtp-inter[%d->%d][%d->%d]: r=%d\n", 399| 0| tx, t_dim->min, idx, *txtp, ts->msac.rng); 400| 1.44M| } 401| 4.94M| } 402| | 403| | // find end-of-block (eob) 404| 17.7M| int eob; 405| 17.7M| const int slw = imin(t_dim->lw, TX_32X32), slh = imin(t_dim->lh, TX_32X32); 406| 17.7M| const int tx2dszctx = slw + slh; 407| 17.7M| const enum TxClass tx_class = dav1d_tx_type_class[*txtp]; 408| 17.7M| const int is_1d = tx_class != TX_CLASS_2D; 409| 17.7M| switch (tx2dszctx) { ------------------ | Branch (409:13): [True: 18.2M, False: 18.4E] ------------------ 410| 0|#define case_sz(sz, bin, ns, is_1d) \ 411| 0| case sz: { \ 412| 0| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ 413| 0| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ 414| 0| break; \ 415| 0| } 416| 8.89M| case_sz(0, 16, 8, [is_1d]); ------------------ | | 411| 8.89M| case sz: { \ | | ------------------ | | | Branch (411:5): [True: 8.89M, False: 8.89M] | | ------------------ | | 412| 8.89M| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ | | 413| 8.89M| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ | | ------------------ | | | | 48| 8.89M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 | | ------------------ | | 414| 8.89M| break; \ | | 415| 8.89M| } ------------------ 417| 1.28M| case_sz(1, 32, 8, [is_1d]); ------------------ | | 411| 1.28M| case sz: { \ | | ------------------ | | | Branch (411:5): [True: 1.28M, False: 16.4M] | | ------------------ | | 412| 1.28M| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ | | 413| 1.28M| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ | | ------------------ | | | | 48| 1.28M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 | | ------------------ | | 414| 1.28M| break; \ | | 415| 1.28M| } ------------------ 418| 2.49M| case_sz(2, 64, 8, [is_1d]); ------------------ | | 411| 2.49M| case sz: { \ | | ------------------ | | | Branch (411:5): [True: 2.49M, False: 15.2M] | | ------------------ | | 412| 2.49M| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ | | 413| 2.49M| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ | | ------------------ | | | | 48| 2.49M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 | | ------------------ | | 414| 2.49M| break; \ | | 415| 2.49M| } ------------------ 419| 1.32M| case_sz(3, 128, 8, [is_1d]); ------------------ | | 411| 1.32M| case sz: { \ | | ------------------ | | | Branch (411:5): [True: 1.32M, False: 16.4M] | | ------------------ | | 412| 1.32M| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ | | 413| 1.32M| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ | | ------------------ | | | | 48| 1.32M|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 | | ------------------ | | 414| 1.32M| break; \ | | 415| 1.32M| } ------------------ 420| 1.59M| case_sz(4, 256, 16, [is_1d]); ------------------ | | 411| 1.59M| case sz: { \ | | ------------------ | | | Branch (411:5): [True: 1.59M, False: 16.1M] | | ------------------ | | 412| 1.59M| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ | | 413| 1.59M| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ | | ------------------ | | | | 57| 1.59M|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) | | ------------------ | | 414| 1.59M| break; \ | | 415| 1.59M| } ------------------ 421| 588k| case_sz(5, 512, 16, ); ------------------ | | 411| 588k| case sz: { \ | | ------------------ | | | Branch (411:5): [True: 588k, False: 17.1M] | | ------------------ | | 412| 588k| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ | | 413| 588k| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ | | ------------------ | | | | 57| 588k|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) | | ------------------ | | 414| 588k| break; \ | | 415| 588k| } ------------------ 422| 2.01M| case_sz(6, 1024, 16, ); ------------------ | | 411| 2.01M| case sz: { \ | | ------------------ | | | Branch (411:5): [True: 2.01M, False: 15.7M] | | ------------------ | | 412| 2.01M| uint16_t *const eob_bin_cdf = ts->cdf.coef.eob_bin_##bin[chroma]is_1d; \ | | 413| 2.01M| eob = dav1d_msac_decode_symbol_adapt##ns(&ts->msac, eob_bin_cdf, 4 + sz); \ | | ------------------ | | | | 57| 2.01M|#define dav1d_msac_decode_symbol_adapt16(ctx, cdf, symb) ((ctx)->symbol_adapt16(ctx, cdf, symb)) | | ------------------ | | 414| 2.01M| break; \ | | 415| 2.01M| } ------------------ 423| 17.7M|#undef case_sz 424| 17.7M| } 425| 18.1M| if (dbg) ------------------ | Branch (425:9): [Folded, False: 18.1M] ------------------ 426| 0| printf("Post-eob_bin_%d[%d][%d][%d]: r=%d\n", 427| 0| 16 << tx2dszctx, chroma, is_1d, eob, ts->msac.rng); 428| 18.1M| if (eob > 1) { ------------------ | Branch (428:9): [True: 13.0M, False: 5.14M] ------------------ 429| 13.0M| const int eob_bin = eob - 2; 430| 13.0M| uint16_t *const eob_hi_bit_cdf = 431| 13.0M| ts->cdf.coef.eob_hi_bit[t_dim->ctx][chroma][eob_bin]; 432| 13.0M| const int eob_hi_bit = dav1d_msac_decode_bool_adapt(&ts->msac, eob_hi_bit_cdf); ------------------ | | 52| 13.0M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 433| 13.0M| if (dbg) ------------------ | Branch (433:13): [Folded, False: 13.0M] ------------------ 434| 0| printf("Post-eob_hi_bit[%d][%d][%d][%d]: r=%d\n", 435| 0| t_dim->ctx, chroma, eob_bin, eob_hi_bit, ts->msac.rng); 436| 13.0M| eob = ((eob_hi_bit | 2) << eob_bin) | dav1d_msac_decode_bools(&ts->msac, eob_bin); 437| 13.0M| if (dbg) ------------------ | Branch (437:13): [Folded, False: 13.0M] ------------------ 438| 0| printf("Post-eob[%d]: r=%d\n", eob, ts->msac.rng); 439| 13.0M| } 440| 18.1M| assert(eob >= 0); ------------------ | Branch (440:5): [True: 18.1M, False: 18.4E] ------------------ 441| | 442| | // base tokens 443| 18.1M| uint16_t (*const eob_cdf)[4] = ts->cdf.coef.eob_base_tok[t_dim->ctx][chroma]; 444| 18.1M| uint16_t (*const hi_cdf)[4] = ts->cdf.coef.br_tok[imin(t_dim->ctx, 3)][chroma]; 445| 18.1M| unsigned rc, dc_tok; 446| | 447| 18.1M| if (eob) { ------------------ | Branch (447:9): [True: 13.6M, False: 4.46M] ------------------ 448| 13.6M| uint16_t (*const lo_cdf)[4] = ts->cdf.coef.base_tok[t_dim->ctx][chroma]; 449| 13.6M| uint8_t *const levels = t->scratch.levels; // bits 0-5: tok, 6-7: lo_tok 450| | 451| | /* eob */ 452| 13.6M| unsigned ctx = 1 + (eob > 2 << tx2dszctx) + (eob > 4 << tx2dszctx); 453| 13.6M| int eob_tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, eob_cdf[ctx], 2); ------------------ | | 47| 13.6M|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 454| 13.6M| int tok = eob_tok + 1; 455| 13.6M| int level_tok = tok * 0x41; 456| 13.6M| unsigned mag; 457| | 458| 13.6M|#define DECODE_COEFS_CLASS(tx_class) \ 459| 13.6M| unsigned x, y; \ 460| 13.6M| uint8_t *level; \ 461| 13.6M| if (tx_class == TX_CLASS_2D) \ 462| 13.6M| rc = scan[eob], x = rc >> shift, y = rc & mask; \ 463| 13.6M| else if (tx_class == TX_CLASS_H) \ 464| | /* Transposing reduces the stride and padding requirements */ \ 465| 13.6M| x = eob & mask, y = eob >> shift, rc = eob; \ 466| 13.6M| else /* tx_class == TX_CLASS_V */ \ 467| 13.6M| x = eob & mask, y = eob >> shift, rc = (x << shift2) | y; \ 468| 13.6M| if (dbg) \ 469| 13.6M| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ 470| 13.6M| t_dim->ctx, chroma, ctx, eob, rc, tok, ts->msac.rng); \ 471| 13.6M| if (eob_tok == 2) { \ 472| 13.6M| ctx = (tx_class == TX_CLASS_2D ? (x | y) > 1 : y != 0) ? 14 : 7; \ 473| 13.6M| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ 474| 13.6M| level_tok = tok + (3 << 6); \ 475| 13.6M| if (dbg) \ 476| 13.6M| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ 477| 13.6M| imin(t_dim->ctx, 3), chroma, ctx, eob, rc, tok, \ 478| 13.6M| ts->msac.rng); \ 479| 13.6M| } \ 480| 13.6M| cf[rc] = tok << 11; \ 481| 13.6M| if (tx_class == TX_CLASS_2D) \ 482| 13.6M| level = levels + rc; \ 483| 13.6M| else \ 484| 13.6M| level = levels + x * stride + y; \ 485| 13.6M| *level = (uint8_t) level_tok; \ 486| 13.6M| for (int i = eob - 1; i > 0; i--) { /* ac */ \ 487| 13.6M| unsigned rc_i; \ 488| 13.6M| if (tx_class == TX_CLASS_2D) \ 489| 13.6M| rc_i = scan[i], x = rc_i >> shift, y = rc_i & mask; \ 490| 13.6M| else if (tx_class == TX_CLASS_H) \ 491| 13.6M| x = i & mask, y = i >> shift, rc_i = i; \ 492| 13.6M| else /* tx_class == TX_CLASS_V */ \ 493| 13.6M| x = i & mask, y = i >> shift, rc_i = (x << shift2) | y; \ 494| 13.6M| assert(x < 32 && y < 32); \ 495| 13.6M| if (tx_class == TX_CLASS_2D) \ 496| 13.6M| level = levels + rc_i; \ 497| 13.6M| else \ 498| 13.6M| level = levels + x * stride + y; \ 499| 13.6M| ctx = get_lo_ctx(level, tx_class, &mag, lo_ctx_offsets, x, y, stride); \ 500| 13.6M| if (tx_class == TX_CLASS_2D) \ 501| 13.6M| y |= x; \ 502| 13.6M| tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ 503| 13.6M| if (dbg) \ 504| 13.6M| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ 505| 13.6M| t_dim->ctx, chroma, ctx, i, rc_i, tok, ts->msac.rng); \ 506| 13.6M| if (tok == 3) { \ 507| 13.6M| mag &= 63; \ 508| 13.6M| ctx = (y > (tx_class == TX_CLASS_2D) ? 14 : 7) + \ 509| 13.6M| (mag > 12 ? 6 : (mag + 1) >> 1); \ 510| 13.6M| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ 511| 13.6M| if (dbg) \ 512| 13.6M| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ 513| 13.6M| imin(t_dim->ctx, 3), chroma, ctx, i, rc_i, tok, \ 514| 13.6M| ts->msac.rng); \ 515| 13.6M| *level = (uint8_t) (tok + (3 << 6)); \ 516| 13.6M| cf[rc_i] = (tok << 11) | rc; \ 517| 13.6M| rc = rc_i; \ 518| 13.6M| } else { \ 519| | /* 0x1 for tok, 0x7ff as bitmask for rc, 0x41 for level_tok */ \ 520| 13.6M| tok *= 0x17ff41; \ 521| 13.6M| *level = (uint8_t) tok; \ 522| | /* tok ? (tok << 11) | rc : 0 */ \ 523| 13.6M| tok = (tok >> 9) & (rc + ~0x7ffu); \ 524| 13.6M| if (tok) rc = rc_i; \ 525| 13.6M| cf[rc_i] = tok; \ 526| 13.6M| } \ 527| 13.6M| } \ 528| | /* dc */ \ 529| 13.6M| ctx = (tx_class == TX_CLASS_2D) ? 0 : \ 530| 13.6M| get_lo_ctx(levels, tx_class, &mag, lo_ctx_offsets, 0, 0, stride); \ 531| 13.6M| dc_tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ 532| 13.6M| if (dbg) \ 533| 13.6M| printf("Post-dc_lo_tok[%d][%d][%d][%d]: r=%d\n", \ 534| 13.6M| t_dim->ctx, chroma, ctx, dc_tok, ts->msac.rng); \ 535| 13.6M| if (dc_tok == 3) { \ 536| 13.6M| if (tx_class == TX_CLASS_2D) \ 537| 13.6M| mag = levels[0 * stride + 1] + levels[1 * stride + 0] + \ 538| 13.6M| levels[1 * stride + 1]; \ 539| 13.6M| mag &= 63; \ 540| 13.6M| ctx = mag > 12 ? 6 : (mag + 1) >> 1; \ 541| 13.6M| dc_tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ 542| 13.6M| if (dbg) \ 543| 13.6M| printf("Post-dc_hi_tok[%d][%d][0][%d]: r=%d\n", \ 544| 13.6M| imin(t_dim->ctx, 3), chroma, dc_tok, ts->msac.rng); \ 545| 13.6M| } \ 546| 13.6M| break 547| | 548| 13.6M| const uint16_t *scan; 549| 13.6M| switch (tx_class) { 550| 12.8M| case TX_CLASS_2D: { ------------------ | Branch (550:9): [True: 12.8M, False: 827k] ------------------ 551| 12.8M| const unsigned nonsquare_tx = tx >= RTX_4X8; 552| 12.8M| const uint8_t (*const lo_ctx_offsets)[5] = 553| 12.8M| dav1d_lo_ctx_offsets[nonsquare_tx + (tx & nonsquare_tx)]; 554| 12.8M| scan = dav1d_scans[tx]; 555| 12.8M| const ptrdiff_t stride = 4 << slh; 556| 12.8M| const unsigned shift = slh + 2, shift2 = 0; 557| 12.8M| const unsigned mask = (4 << slh) - 1; 558| 12.8M| memset(levels, 0, stride * ((4 << slw) + 2)); 559| 12.8M| DECODE_COEFS_CLASS(TX_CLASS_2D); ------------------ | | 459| 12.8M| unsigned x, y; \ | | 460| 12.8M| uint8_t *level; \ | | 461| 12.8M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (461:13): [True: 12.8M, Folded] | | ------------------ | | 462| 12.8M| rc = scan[eob], x = rc >> shift, y = rc & mask; \ | | 463| 12.8M| else if (tx_class == TX_CLASS_H) \ | | ------------------ | | | Branch (463:18): [Folded, False: 8.26k] | | ------------------ | | 464| 8.26k| /* Transposing reduces the stride and padding requirements */ \ | | 465| 8.26k| x = eob & mask, y = eob >> shift, rc = eob; \ | | 466| 8.26k| else /* tx_class == TX_CLASS_V */ \ | | 467| 8.26k| x = eob & mask, y = eob >> shift, rc = (x << shift2) | y; \ | | 468| 12.8M| if (dbg) \ | | ------------------ | | | Branch (468:13): [Folded, False: 12.8M] | | ------------------ | | 469| 12.8M| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 470| 0| t_dim->ctx, chroma, ctx, eob, rc, tok, ts->msac.rng); \ | | 471| 12.8M| if (eob_tok == 2) { \ | | ------------------ | | | Branch (471:13): [True: 447k, False: 12.4M] | | ------------------ | | 472| 18.4E| ctx = (tx_class == TX_CLASS_2D ? (x | y) > 1 : y != 0) ? 14 : 7; \ | | ------------------ | | | Branch (472:19): [True: 442k, False: 4.48k] | | | Branch (472:20): [True: 447k, Folded] | | ------------------ | | 473| 447k| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 447k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 474| 447k| level_tok = tok + (3 << 6); \ | | 475| 447k| if (dbg) \ | | ------------------ | | | Branch (475:17): [Folded, False: 447k] | | ------------------ | | 476| 447k| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 477| 0| imin(t_dim->ctx, 3), chroma, ctx, eob, rc, tok, \ | | 478| 0| ts->msac.rng); \ | | 479| 447k| } \ | | 480| 12.8M| cf[rc] = tok << 11; \ | | 481| 12.8M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (481:13): [True: 12.8M, Folded] | | ------------------ | | 482| 12.8M| level = levels + rc; \ | | 483| 12.8M| else \ | | 484| 12.8M| level = levels + x * stride + y; \ | | 485| 12.8M| *level = (uint8_t) level_tok; \ | | 486| 219M| for (int i = eob - 1; i > 0; i--) { /* ac */ \ | | ------------------ | | | Branch (486:31): [True: 206M, False: 13.3M] | | ------------------ | | 487| 206M| unsigned rc_i; \ | | 488| 206M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (488:17): [True: 206M, Folded] | | ------------------ | | 489| 206M| rc_i = scan[i], x = rc_i >> shift, y = rc_i & mask; \ | | 490| 206M| else if (tx_class == TX_CLASS_H) \ | | ------------------ | | | Branch (490:22): [Folded, False: 103k] | | ------------------ | | 491| 103k| x = i & mask, y = i >> shift, rc_i = i; \ | | 492| 103k| else /* tx_class == TX_CLASS_V */ \ | | 493| 103k| x = i & mask, y = i >> shift, rc_i = (x << shift2) | y; \ | | 494| 206M| assert(x < 32 && y < 32); \ | | 495| 206M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (495:17): [True: 206M, Folded] | | ------------------ | | 496| 206M| level = levels + rc_i; \ | | 497| 206M| else \ | | 498| 18.4E| level = levels + x * stride + y; \ | | 499| 206M| ctx = get_lo_ctx(level, tx_class, &mag, lo_ctx_offsets, x, y, stride); \ | | 500| 206M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (500:17): [True: 204M, Folded] | | ------------------ | | 501| 206M| y |= x; \ | | 502| 206M| tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ | | ------------------ | | | | 47| 206M|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 | | ------------------ | | 503| 206M| if (dbg) \ | | ------------------ | | | Branch (503:17): [Folded, False: 206M] | | ------------------ | | 504| 206M| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 505| 0| t_dim->ctx, chroma, ctx, i, rc_i, tok, ts->msac.rng); \ | | 506| 206M| if (tok == 3) { \ | | ------------------ | | | Branch (506:17): [True: 19.2M, False: 187M] | | ------------------ | | 507| 19.2M| mag &= 63; \ | | 508| 19.2M| ctx = (y > (tx_class == TX_CLASS_2D) ? 14 : 7) + \ | | ------------------ | | | Branch (508:24): [True: 13.1M, False: 6.09M] | | ------------------ | | 509| 19.2M| (mag > 12 ? 6 : (mag + 1) >> 1); \ | | ------------------ | | | Branch (509:24): [True: 3.39M, False: 15.8M] | | ------------------ | | 510| 19.2M| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 19.2M|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 511| 19.2M| if (dbg) \ | | ------------------ | | | Branch (511:21): [Folded, False: 19.2M] | | ------------------ | | 512| 19.2M| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 513| 0| imin(t_dim->ctx, 3), chroma, ctx, i, rc_i, tok, \ | | 514| 0| ts->msac.rng); \ | | 515| 19.2M| *level = (uint8_t) (tok + (3 << 6)); \ | | 516| 19.2M| cf[rc_i] = (tok << 11) | rc; \ | | 517| 19.2M| rc = rc_i; \ | | 518| 187M| } else { \ | | 519| 187M| /* 0x1 for tok, 0x7ff as bitmask for rc, 0x41 for level_tok */ \ | | 520| 187M| tok *= 0x17ff41; \ | | 521| 187M| *level = (uint8_t) tok; \ | | 522| 187M| /* tok ? (tok << 11) | rc : 0 */ \ | | 523| 187M| tok = (tok >> 9) & (rc + ~0x7ffu); \ | | 524| 187M| if (tok) rc = rc_i; \ | | ------------------ | | | Branch (524:21): [True: 67.6M, False: 119M] | | ------------------ | | 525| 187M| cf[rc_i] = tok; \ | | 526| 187M| } \ | | 527| 206M| } \ | | 528| 12.8M| /* dc */ \ | | 529| 13.3M| ctx = (tx_class == TX_CLASS_2D) ? 0 : \ | | ------------------ | | | Branch (529:15): [True: 12.8M, Folded] | | ------------------ | | 530| 13.3M| get_lo_ctx(levels, tx_class, &mag, lo_ctx_offsets, 0, 0, stride); \ | | 531| 13.3M| dc_tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ | | ------------------ | | | | 47| 13.3M|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 | | ------------------ | | 532| 13.3M| if (dbg) \ | | ------------------ | | | Branch (532:13): [Folded, False: 13.3M] | | ------------------ | | 533| 13.3M| printf("Post-dc_lo_tok[%d][%d][%d][%d]: r=%d\n", \ | | 534| 0| t_dim->ctx, chroma, ctx, dc_tok, ts->msac.rng); \ | | 535| 13.3M| if (dc_tok == 3) { \ | | ------------------ | | | Branch (535:13): [True: 5.31M, False: 8.00M] | | ------------------ | | 536| 5.31M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (536:17): [True: 5.31M, Folded] | | ------------------ | | 537| 5.31M| mag = levels[0 * stride + 1] + levels[1 * stride + 0] + \ | | 538| 5.31M| levels[1 * stride + 1]; \ | | 539| 5.31M| mag &= 63; \ | | 540| 5.31M| ctx = mag > 12 ? 6 : (mag + 1) >> 1; \ | | ------------------ | | | Branch (540:19): [True: 806k, False: 4.50M] | | ------------------ | | 541| 5.31M| dc_tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 5.31M|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 542| 5.31M| if (dbg) \ | | ------------------ | | | Branch (542:17): [Folded, False: 5.31M] | | ------------------ | | 543| 5.31M| printf("Post-dc_hi_tok[%d][%d][0][%d]: r=%d\n", \ | | 544| 0| imin(t_dim->ctx, 3), chroma, dc_tok, ts->msac.rng); \ | | 545| 5.31M| } \ | | 546| 13.3M| break ------------------ | Branch (559:13): [True: 206M, False: 18.4E] | Branch (559:13): [True: 206M, False: 49.3k] ------------------ 560| 12.8M| } 561| 541k| case TX_CLASS_H: { ------------------ | Branch (561:9): [True: 541k, False: 13.1M] ------------------ 562| 541k| const uint8_t (*const lo_ctx_offsets)[5] = NULL; 563| 541k| const ptrdiff_t stride = 16; 564| 541k| const unsigned shift = slh + 2, shift2 = 0; 565| 541k| const unsigned mask = (4 << slh) - 1; 566| 541k| memset(levels, 0, stride * ((4 << slh) + 2)); 567| 541k| DECODE_COEFS_CLASS(TX_CLASS_H); ------------------ | | 459| 541k| unsigned x, y; \ | | 460| 541k| uint8_t *level; \ | | 461| 541k| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (461:13): [Folded, False: 541k] | | ------------------ | | 462| 541k| rc = scan[eob], x = rc >> shift, y = rc & mask; \ | | 463| 541k| else if (tx_class == TX_CLASS_H) \ | | ------------------ | | | Branch (463:18): [True: 541k, Folded] | | ------------------ | | 464| 541k| /* Transposing reduces the stride and padding requirements */ \ | | 465| 541k| x = eob & mask, y = eob >> shift, rc = eob; \ | | 466| 541k| else /* tx_class == TX_CLASS_V */ \ | | 467| 541k| x = eob & mask, y = eob >> shift, rc = (x << shift2) | y; \ | | 468| 541k| if (dbg) \ | | ------------------ | | | Branch (468:13): [Folded, False: 541k] | | ------------------ | | 469| 541k| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 470| 0| t_dim->ctx, chroma, ctx, eob, rc, tok, ts->msac.rng); \ | | 471| 541k| if (eob_tok == 2) { \ | | ------------------ | | | Branch (471:13): [True: 11.3k, False: 530k] | | ------------------ | | 472| 11.3k| ctx = (tx_class == TX_CLASS_2D ? (x | y) > 1 : y != 0) ? 14 : 7; \ | | ------------------ | | | Branch (472:19): [True: 8.56k, False: 2.77k] | | | Branch (472:20): [Folded, False: 11.3k] | | ------------------ | | 473| 11.3k| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 11.3k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 474| 11.3k| level_tok = tok + (3 << 6); \ | | 475| 11.3k| if (dbg) \ | | ------------------ | | | Branch (475:17): [Folded, False: 11.3k] | | ------------------ | | 476| 11.3k| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 477| 0| imin(t_dim->ctx, 3), chroma, ctx, eob, rc, tok, \ | | 478| 0| ts->msac.rng); \ | | 479| 11.3k| } \ | | 480| 541k| cf[rc] = tok << 11; \ | | 481| 541k| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (481:13): [Folded, False: 541k] | | ------------------ | | 482| 541k| level = levels + rc; \ | | 483| 541k| else \ | | 484| 541k| level = levels + x * stride + y; \ | | 485| 541k| *level = (uint8_t) level_tok; \ | | 486| 10.6M| for (int i = eob - 1; i > 0; i--) { /* ac */ \ | | ------------------ | | | Branch (486:31): [True: 10.1M, False: 540k] | | ------------------ | | 487| 10.1M| unsigned rc_i; \ | | 488| 10.1M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (488:17): [Folded, False: 10.1M] | | ------------------ | | 489| 10.1M| rc_i = scan[i], x = rc_i >> shift, y = rc_i & mask; \ | | 490| 10.1M| else if (tx_class == TX_CLASS_H) \ | | ------------------ | | | Branch (490:22): [True: 10.1M, Folded] | | ------------------ | | 491| 10.1M| x = i & mask, y = i >> shift, rc_i = i; \ | | 492| 10.1M| else /* tx_class == TX_CLASS_V */ \ | | 493| 10.1M| x = i & mask, y = i >> shift, rc_i = (x << shift2) | y; \ | | 494| 10.1M| assert(x < 32 && y < 32); \ | | 495| 10.1M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (495:17): [Folded, False: 10.1M] | | ------------------ | | 496| 10.1M| level = levels + rc_i; \ | | 497| 10.1M| else \ | | 498| 10.1M| level = levels + x * stride + y; \ | | 499| 10.1M| ctx = get_lo_ctx(level, tx_class, &mag, lo_ctx_offsets, x, y, stride); \ | | 500| 10.1M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (500:17): [Folded, False: 10.1M] | | ------------------ | | 501| 10.1M| y |= x; \ | | 502| 10.1M| tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ | | ------------------ | | | | 47| 10.1M|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 | | ------------------ | | 503| 10.1M| if (dbg) \ | | ------------------ | | | Branch (503:17): [Folded, False: 10.1M] | | ------------------ | | 504| 10.1M| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 505| 0| t_dim->ctx, chroma, ctx, i, rc_i, tok, ts->msac.rng); \ | | 506| 10.1M| if (tok == 3) { \ | | ------------------ | | | Branch (506:17): [True: 513k, False: 9.64M] | | ------------------ | | 507| 513k| mag &= 63; \ | | 508| 513k| ctx = (y > (tx_class == TX_CLASS_2D) ? 14 : 7) + \ | | ------------------ | | | Branch (508:24): [True: 314k, False: 198k] | | ------------------ | | 509| 513k| (mag > 12 ? 6 : (mag + 1) >> 1); \ | | ------------------ | | | Branch (509:24): [True: 46.8k, False: 466k] | | ------------------ | | 510| 513k| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 513k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 511| 513k| if (dbg) \ | | ------------------ | | | Branch (511:21): [Folded, False: 513k] | | ------------------ | | 512| 513k| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 513| 0| imin(t_dim->ctx, 3), chroma, ctx, i, rc_i, tok, \ | | 514| 0| ts->msac.rng); \ | | 515| 513k| *level = (uint8_t) (tok + (3 << 6)); \ | | 516| 513k| cf[rc_i] = (tok << 11) | rc; \ | | 517| 513k| rc = rc_i; \ | | 518| 9.64M| } else { \ | | 519| 9.64M| /* 0x1 for tok, 0x7ff as bitmask for rc, 0x41 for level_tok */ \ | | 520| 9.64M| tok *= 0x17ff41; \ | | 521| 9.64M| *level = (uint8_t) tok; \ | | 522| 9.64M| /* tok ? (tok << 11) | rc : 0 */ \ | | 523| 9.64M| tok = (tok >> 9) & (rc + ~0x7ffu); \ | | 524| 9.64M| if (tok) rc = rc_i; \ | | ------------------ | | | Branch (524:21): [True: 2.60M, False: 7.03M] | | ------------------ | | 525| 9.64M| cf[rc_i] = tok; \ | | 526| 9.64M| } \ | | 527| 10.1M| } \ | | 528| 541k| /* dc */ \ | | 529| 541k| ctx = (tx_class == TX_CLASS_2D) ? 0 : \ | | ------------------ | | | Branch (529:15): [Folded, False: 540k] | | ------------------ | | 530| 540k| get_lo_ctx(levels, tx_class, &mag, lo_ctx_offsets, 0, 0, stride); \ | | 531| 540k| dc_tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ | | ------------------ | | | | 47| 540k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 | | ------------------ | | 532| 540k| if (dbg) \ | | ------------------ | | | Branch (532:13): [Folded, False: 540k] | | ------------------ | | 533| 540k| printf("Post-dc_lo_tok[%d][%d][%d][%d]: r=%d\n", \ | | 534| 0| t_dim->ctx, chroma, ctx, dc_tok, ts->msac.rng); \ | | 535| 540k| if (dc_tok == 3) { \ | | ------------------ | | | Branch (535:13): [True: 61.6k, False: 478k] | | ------------------ | | 536| 61.6k| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (536:17): [Folded, False: 61.6k] | | ------------------ | | 537| 61.6k| mag = levels[0 * stride + 1] + levels[1 * stride + 0] + \ | | 538| 0| levels[1 * stride + 1]; \ | | 539| 61.6k| mag &= 63; \ | | 540| 61.6k| ctx = mag > 12 ? 6 : (mag + 1) >> 1; \ | | ------------------ | | | Branch (540:19): [True: 7.57k, False: 54.1k] | | ------------------ | | 541| 61.6k| dc_tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 61.6k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 542| 61.6k| if (dbg) \ | | ------------------ | | | Branch (542:17): [Folded, False: 61.6k] | | ------------------ | | 543| 61.6k| printf("Post-dc_hi_tok[%d][%d][0][%d]: r=%d\n", \ | | 544| 0| imin(t_dim->ctx, 3), chroma, dc_tok, ts->msac.rng); \ | | 545| 61.6k| } \ | | 546| 540k| break ------------------ | Branch (567:13): [True: 10.1M, False: 774] | Branch (567:13): [True: 10.1M, False: 634] ------------------ 568| 541k| } 569| 292k| case TX_CLASS_V: { ------------------ | Branch (569:9): [True: 292k, False: 13.3M] ------------------ 570| 292k| const uint8_t (*const lo_ctx_offsets)[5] = NULL; 571| 292k| const ptrdiff_t stride = 16; 572| 292k| const unsigned shift = slw + 2, shift2 = slh + 2; 573| 292k| const unsigned mask = (4 << slw) - 1; 574| 292k| memset(levels, 0, stride * ((4 << slw) + 2)); 575| 292k| DECODE_COEFS_CLASS(TX_CLASS_V); ------------------ | | 459| 292k| unsigned x, y; \ | | 460| 292k| uint8_t *level; \ | | 461| 292k| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (461:13): [Folded, False: 292k] | | ------------------ | | 462| 292k| rc = scan[eob], x = rc >> shift, y = rc & mask; \ | | 463| 292k| else if (tx_class == TX_CLASS_H) \ | | ------------------ | | | Branch (463:18): [Folded, False: 292k] | | ------------------ | | 464| 292k| /* Transposing reduces the stride and padding requirements */ \ | | 465| 292k| x = eob & mask, y = eob >> shift, rc = eob; \ | | 466| 292k| else /* tx_class == TX_CLASS_V */ \ | | 467| 292k| x = eob & mask, y = eob >> shift, rc = (x << shift2) | y; \ | | 468| 292k| if (dbg) \ | | ------------------ | | | Branch (468:13): [Folded, False: 292k] | | ------------------ | | 469| 292k| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 470| 0| t_dim->ctx, chroma, ctx, eob, rc, tok, ts->msac.rng); \ | | 471| 292k| if (eob_tok == 2) { \ | | ------------------ | | | Branch (471:13): [True: 3.43k, False: 288k] | | ------------------ | | 472| 3.43k| ctx = (tx_class == TX_CLASS_2D ? (x | y) > 1 : y != 0) ? 14 : 7; \ | | ------------------ | | | Branch (472:19): [True: 3.11k, False: 314] | | | Branch (472:20): [Folded, False: 3.43k] | | ------------------ | | 473| 3.43k| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 3.43k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 474| 3.43k| level_tok = tok + (3 << 6); \ | | 475| 3.43k| if (dbg) \ | | ------------------ | | | Branch (475:17): [Folded, False: 3.43k] | | ------------------ | | 476| 3.43k| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 477| 0| imin(t_dim->ctx, 3), chroma, ctx, eob, rc, tok, \ | | 478| 0| ts->msac.rng); \ | | 479| 3.43k| } \ | | 480| 292k| cf[rc] = tok << 11; \ | | 481| 292k| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (481:13): [Folded, False: 292k] | | ------------------ | | 482| 292k| level = levels + rc; \ | | 483| 292k| else \ | | 484| 292k| level = levels + x * stride + y; \ | | 485| 292k| *level = (uint8_t) level_tok; \ | | 486| 6.21M| for (int i = eob - 1; i > 0; i--) { /* ac */ \ | | ------------------ | | | Branch (486:31): [True: 5.92M, False: 292k] | | ------------------ | | 487| 5.92M| unsigned rc_i; \ | | 488| 5.92M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (488:17): [Folded, False: 5.92M] | | ------------------ | | 489| 5.92M| rc_i = scan[i], x = rc_i >> shift, y = rc_i & mask; \ | | 490| 5.92M| else if (tx_class == TX_CLASS_H) \ | | ------------------ | | | Branch (490:22): [Folded, False: 5.92M] | | ------------------ | | 491| 5.92M| x = i & mask, y = i >> shift, rc_i = i; \ | | 492| 5.92M| else /* tx_class == TX_CLASS_V */ \ | | 493| 5.92M| x = i & mask, y = i >> shift, rc_i = (x << shift2) | y; \ | | 494| 5.92M| assert(x < 32 && y < 32); \ | | 495| 5.92M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (495:17): [Folded, False: 5.92M] | | ------------------ | | 496| 5.92M| level = levels + rc_i; \ | | 497| 5.92M| else \ | | 498| 5.92M| level = levels + x * stride + y; \ | | 499| 5.92M| ctx = get_lo_ctx(level, tx_class, &mag, lo_ctx_offsets, x, y, stride); \ | | 500| 5.92M| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (500:17): [Folded, False: 5.92M] | | ------------------ | | 501| 5.92M| y |= x; \ | | 502| 5.92M| tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ | | ------------------ | | | | 47| 5.92M|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 | | ------------------ | | 503| 5.92M| if (dbg) \ | | ------------------ | | | Branch (503:17): [Folded, False: 5.92M] | | ------------------ | | 504| 5.92M| printf("Post-lo_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 505| 0| t_dim->ctx, chroma, ctx, i, rc_i, tok, ts->msac.rng); \ | | 506| 5.92M| if (tok == 3) { \ | | ------------------ | | | Branch (506:17): [True: 241k, False: 5.68M] | | ------------------ | | 507| 241k| mag &= 63; \ | | 508| 241k| ctx = (y > (tx_class == TX_CLASS_2D) ? 14 : 7) + \ | | ------------------ | | | Branch (508:24): [True: 144k, False: 96.6k] | | ------------------ | | 509| 241k| (mag > 12 ? 6 : (mag + 1) >> 1); \ | | ------------------ | | | Branch (509:24): [True: 22.9k, False: 218k] | | ------------------ | | 510| 241k| tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 241k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 511| 241k| if (dbg) \ | | ------------------ | | | Branch (511:21): [Folded, False: 241k] | | ------------------ | | 512| 241k| printf("Post-hi_tok[%d][%d][%d][%d=%d=%d]: r=%d\n", \ | | 513| 0| imin(t_dim->ctx, 3), chroma, ctx, i, rc_i, tok, \ | | 514| 0| ts->msac.rng); \ | | 515| 241k| *level = (uint8_t) (tok + (3 << 6)); \ | | 516| 241k| cf[rc_i] = (tok << 11) | rc; \ | | 517| 241k| rc = rc_i; \ | | 518| 5.68M| } else { \ | | 519| 5.68M| /* 0x1 for tok, 0x7ff as bitmask for rc, 0x41 for level_tok */ \ | | 520| 5.68M| tok *= 0x17ff41; \ | | 521| 5.68M| *level = (uint8_t) tok; \ | | 522| 5.68M| /* tok ? (tok << 11) | rc : 0 */ \ | | 523| 5.68M| tok = (tok >> 9) & (rc + ~0x7ffu); \ | | 524| 5.68M| if (tok) rc = rc_i; \ | | ------------------ | | | Branch (524:21): [True: 1.42M, False: 4.26M] | | ------------------ | | 525| 5.68M| cf[rc_i] = tok; \ | | 526| 5.68M| } \ | | 527| 5.92M| } \ | | 528| 292k| /* dc */ \ | | 529| 292k| ctx = (tx_class == TX_CLASS_2D) ? 0 : \ | | ------------------ | | | Branch (529:15): [Folded, False: 292k] | | ------------------ | | 530| 292k| get_lo_ctx(levels, tx_class, &mag, lo_ctx_offsets, 0, 0, stride); \ | | 531| 292k| dc_tok = dav1d_msac_decode_symbol_adapt4(&ts->msac, lo_cdf[ctx], 3); \ | | ------------------ | | | | 47| 292k|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 | | ------------------ | | 532| 292k| if (dbg) \ | | ------------------ | | | Branch (532:13): [Folded, False: 292k] | | ------------------ | | 533| 292k| printf("Post-dc_lo_tok[%d][%d][%d][%d]: r=%d\n", \ | | 534| 0| t_dim->ctx, chroma, ctx, dc_tok, ts->msac.rng); \ | | 535| 292k| if (dc_tok == 3) { \ | | ------------------ | | | Branch (535:13): [True: 26.5k, False: 266k] | | ------------------ | | 536| 26.5k| if (tx_class == TX_CLASS_2D) \ | | ------------------ | | | Branch (536:17): [Folded, False: 26.5k] | | ------------------ | | 537| 26.5k| mag = levels[0 * stride + 1] + levels[1 * stride + 0] + \ | | 538| 0| levels[1 * stride + 1]; \ | | 539| 26.5k| mag &= 63; \ | | 540| 26.5k| ctx = mag > 12 ? 6 : (mag + 1) >> 1; \ | | ------------------ | | | Branch (540:19): [True: 3.90k, False: 22.6k] | | ------------------ | | 541| 26.5k| dc_tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[ctx]); \ | | ------------------ | | | | 49| 26.5k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 | | ------------------ | | 542| 26.5k| if (dbg) \ | | ------------------ | | | Branch (542:17): [Folded, False: 26.5k] | | ------------------ | | 543| 26.5k| printf("Post-dc_hi_tok[%d][%d][0][%d]: r=%d\n", \ | | 544| 0| imin(t_dim->ctx, 3), chroma, dc_tok, ts->msac.rng); \ | | 545| 26.5k| } \ | | 546| 292k| break ------------------ | Branch (575:13): [True: 5.92M, False: 18.4E] | Branch (575:13): [True: 5.92M, False: 113] ------------------ 576| 292k| } 577| 0|#undef DECODE_COEFS_CLASS 578| 0| default: assert(0); ------------------ | Branch (578:9): [True: 0, False: 13.6M] | Branch (578:18): [Folded, False: 0] ------------------ 579| 13.6M| } 580| 13.6M| } else { // dc-only 581| 4.46M| int tok_br = dav1d_msac_decode_symbol_adapt4(&ts->msac, eob_cdf[0], 2); ------------------ | | 47| 4.46M|#define dav1d_msac_decode_symbol_adapt4 dav1d_msac_decode_symbol_adapt4_sse2 ------------------ 582| 4.46M| dc_tok = 1 + tok_br; 583| 4.46M| if (dbg) ------------------ | Branch (583:13): [Folded, False: 4.46M] ------------------ 584| 0| printf("Post-dc_lo_tok[%d][%d][%d][%d]: r=%d\n", 585| 0| t_dim->ctx, chroma, 0, dc_tok, ts->msac.rng); 586| 4.46M| if (tok_br == 2) { ------------------ | Branch (586:13): [True: 279k, False: 4.18M] ------------------ 587| 279k| dc_tok = dav1d_msac_decode_hi_tok(&ts->msac, hi_cdf[0]); ------------------ | | 49| 279k|#define dav1d_msac_decode_hi_tok dav1d_msac_decode_hi_tok_sse2 ------------------ 588| 279k| if (dbg) ------------------ | Branch (588:17): [Folded, False: 279k] ------------------ 589| 0| printf("Post-dc_hi_tok[%d][%d][0][%d]: r=%d\n", 590| 0| imin(t_dim->ctx, 3), chroma, dc_tok, ts->msac.rng); 591| 279k| } 592| 4.46M| rc = 0; 593| 4.46M| } 594| | 595| | // residual and sign 596| 18.1M| const uint16_t *const dq_tbl = ts->dq[b->seg_id][plane]; 597| 18.1M| const uint8_t *const qm_tbl = *txtp < IDTX ? f->qm[tx][plane] : NULL; ------------------ | Branch (597:35): [True: 9.09M, False: 9.03M] ------------------ 598| 18.1M| const int dq_shift = imax(0, t_dim->ctx - 2); 599| 18.1M| const int cf_max = ~(~127U << (BITDEPTH == 8 ? 8 : f->cur.p.bpc)); ------------------ | Branch (599:36): [True: 4.86M, Folded] ------------------ 600| 18.1M| unsigned cul_level, dc_sign_level; 601| | 602| 18.1M| if (!dc_tok) { ------------------ | Branch (602:9): [True: 3.20M, False: 14.9M] ------------------ 603| 3.20M| cul_level = 0; 604| 3.20M| dc_sign_level = 1 << 6; 605| 3.20M| if (qm_tbl) goto ac_qm; ------------------ | Branch (605:13): [True: 193k, False: 3.00M] ------------------ 606| 3.00M| goto ac_noqm; 607| 3.20M| } 608| | 609| 14.9M| const int dc_sign_ctx = get_dc_sign_ctx(tx, a, l); 610| 14.9M| uint16_t *const dc_sign_cdf = ts->cdf.coef.dc_sign[chroma][dc_sign_ctx]; 611| 14.9M| const int dc_sign = dav1d_msac_decode_bool_adapt(&ts->msac, dc_sign_cdf); ------------------ | | 52| 14.9M|#define dav1d_msac_decode_bool_adapt dav1d_msac_decode_bool_adapt_sse2 ------------------ 612| 14.9M| if (dbg) ------------------ | Branch (612:9): [Folded, False: 14.9M] ------------------ 613| 0| printf("Post-dc_sign[%d][%d][%d]: r=%d\n", 614| 0| chroma, dc_sign_ctx, dc_sign, ts->msac.rng); 615| | 616| 14.9M| int dc_dq = dq_tbl[0]; 617| 14.9M| dc_sign_level = (dc_sign - 1) & (2 << 6); 618| | 619| 14.9M| if (qm_tbl) { ------------------ | Branch (619:9): [True: 1.51M, False: 13.4M] ------------------ 620| 1.51M| dc_dq = (dc_dq * qm_tbl[0] + 16) >> 5; 621| | 622| 1.51M| if (dc_tok == 15) { ------------------ | Branch (622:13): [True: 25.9k, False: 1.48M] ------------------ 623| 25.9k| dc_tok = read_golomb(&ts->msac) + 15; 624| 25.9k| if (dbg) ------------------ | Branch (624:17): [Folded, False: 25.9k] ------------------ 625| 0| printf("Post-dc_residual[%d->%d]: r=%d\n", 626| 0| dc_tok - 15, dc_tok, ts->msac.rng); 627| | 628| 25.9k| dc_tok &= 0xfffff; 629| 25.9k| dc_dq = (dc_dq * dc_tok) & 0xffffff; 630| 1.48M| } else { 631| 1.48M| dc_dq *= dc_tok; 632| 1.48M| assert(dc_dq <= 0xffffff); ------------------ | Branch (632:13): [True: 1.48M, False: 18.4E] ------------------ 633| 1.48M| } 634| 1.51M| cul_level = dc_tok; 635| 1.51M| dc_dq >>= dq_shift; 636| 1.51M| dc_dq = umin(dc_dq, cf_max + dc_sign); 637| 1.51M| cf[0] = (coef) (dc_sign ? -dc_dq : dc_dq); ------------------ | Branch (637:25): [True: 727k, False: 787k] ------------------ 638| | 639| 1.51M| if (rc) ac_qm: { ------------------ | Branch (639:13): [True: 619k, False: 895k] ------------------ 640| 1.43M| const unsigned ac_dq = dq_tbl[1]; 641| 10.3M| do { 642| 10.3M| const int sign = dav1d_msac_decode_bool_equi(&ts->msac); ------------------ | | 53| 10.3M|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ 643| 10.3M| if (dbg) ------------------ | Branch (643:21): [Folded, False: 10.3M] ------------------ 644| 0| printf("Post-sign[%d=%d]: r=%d\n", rc, sign, ts->msac.rng); 645| 10.3M| const unsigned rc_tok = cf[rc]; 646| 10.3M| unsigned tok, dq = (ac_dq * qm_tbl[rc] + 16) >> 5; 647| 10.3M| int dq_sat; 648| | 649| 10.3M| if (rc_tok >= (15 << 11)) { ------------------ | Branch (649:21): [True: 530k, False: 9.84M] ------------------ 650| 530k| tok = read_golomb(&ts->msac) + 15; 651| 530k| if (dbg) ------------------ | Branch (651:25): [Folded, False: 530k] ------------------ 652| 0| printf("Post-residual[%d=%d->%d]: r=%d\n", 653| 0| rc, tok - 15, tok, ts->msac.rng); 654| | 655| 530k| tok &= 0xfffff; 656| 530k| dq = (dq * tok) & 0xffffff; 657| 9.84M| } else { 658| 9.84M| tok = rc_tok >> 11; 659| 9.84M| dq *= tok; 660| 9.84M| assert(dq <= 0xffffff); ------------------ | Branch (660:21): [True: 9.84M, False: 2] ------------------ 661| 9.84M| } 662| 10.3M| cul_level += tok; 663| 10.3M| dq >>= dq_shift; 664| 10.3M| dq_sat = umin(dq, cf_max + sign); 665| 10.3M| cf[rc] = (coef) (sign ? -dq_sat : dq_sat); ------------------ | Branch (665:34): [True: 5.23M, False: 5.13M] ------------------ 666| | 667| 10.3M| rc = rc_tok & 0x3ff; 668| 10.3M| } while (rc); ------------------ | Branch (668:22): [True: 9.55M, False: 812k] ------------------ 669| 1.43M| } 670| 13.4M| } else { 671| | // non-qmatrix is the common case and allows for additional optimizations 672| 13.4M| if (dc_tok == 15) { ------------------ | Branch (672:13): [True: 669k, False: 12.7M] ------------------ 673| 669k| dc_tok = read_golomb(&ts->msac) + 15; 674| 669k| if (dbg) ------------------ | Branch (674:17): [Folded, False: 669k] ------------------ 675| 0| printf("Post-dc_residual[%d->%d]: r=%d\n", 676| 0| dc_tok - 15, dc_tok, ts->msac.rng); 677| | 678| 669k| dc_tok &= 0xfffff; 679| 669k| dc_dq = ((dc_dq * dc_tok) & 0xffffff) >> dq_shift; 680| 669k| dc_dq = umin(dc_dq, cf_max + dc_sign); 681| 12.7M| } else { 682| 12.7M| dc_dq = ((dc_dq * dc_tok) >> dq_shift); 683| 12.7M| assert(dc_dq <= cf_max); ------------------ | Branch (683:13): [True: 12.8M, False: 18.4E] ------------------ 684| 12.7M| } 685| 13.4M| cul_level = dc_tok; 686| 13.4M| cf[0] = (coef) (dc_sign ? -dc_dq : dc_dq); ------------------ | Branch (686:25): [True: 6.30M, False: 7.17M] ------------------ 687| | 688| 22.7M| if (rc) ac_noqm: { ------------------ | Branch (688:13): [True: 9.88M, False: 3.59M] ------------------ 689| 22.7M| const unsigned ac_dq = dq_tbl[1]; 690| 98.7M| do { 691| 98.7M| const int sign = dav1d_msac_decode_bool_equi(&ts->msac); ------------------ | | 53| 98.7M|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ 692| 98.7M| if (dbg) ------------------ | Branch (692:21): [Folded, False: 98.7M] ------------------ 693| 0| printf("Post-sign[%d=%d]: r=%d\n", rc, sign, ts->msac.rng); 694| 98.7M| const unsigned rc_tok = cf[rc]; 695| 98.7M| unsigned tok; 696| 98.7M| int dq; 697| | 698| | // residual 699| 98.7M| if (rc_tok >= (15 << 11)) { ------------------ | Branch (699:21): [True: 1.85M, False: 96.8M] ------------------ 700| 1.85M| tok = read_golomb(&ts->msac) + 15; 701| 1.85M| if (dbg) ------------------ | Branch (701:25): [Folded, False: 1.85M] ------------------ 702| 0| printf("Post-residual[%d=%d->%d]: r=%d\n", 703| 0| rc, tok - 15, tok, ts->msac.rng); 704| | 705| | // coefficient parsing, see 5.11.39 706| 1.85M| tok &= 0xfffff; 707| | 708| | // dequant, see 7.12.3 709| 1.85M| dq = ((ac_dq * tok) & 0xffffff) >> dq_shift; 710| 1.85M| dq = umin(dq, cf_max + sign); 711| 96.8M| } else { 712| | // cannot exceed cf_max, so we can avoid the clipping 713| 96.8M| tok = rc_tok >> 11; 714| 96.8M| dq = ((ac_dq * tok) >> dq_shift); 715| 96.8M| assert(dq <= cf_max); ------------------ | Branch (715:21): [True: 96.7M, False: 101k] ------------------ 716| 96.8M| } 717| 98.6M| cul_level += tok; 718| 98.6M| cf[rc] = (coef) (sign ? -dq : dq); ------------------ | Branch (718:34): [True: 49.8M, False: 48.7M] ------------------ 719| | 720| 98.6M| rc = rc_tok & 0x3ff; // next non-zero rc, zero if eob 721| 98.6M| } while (rc); ------------------ | Branch (721:22): [True: 85.8M, False: 12.7M] ------------------ 722| 22.7M| } 723| 13.4M| } 724| | 725| | // context 726| 18.0M| *res_ctx = umin(cul_level, 63) | dc_sign_level; 727| | 728| 18.0M| return eob; 729| 14.9M|} recon_tmpl.c:get_skip_ctx: 65| 49.9M|{ 66| 49.9M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 67| | 68| 49.9M| if (chroma) { ------------------ | Branch (68:9): [True: 32.0M, False: 17.9M] ------------------ 69| 32.0M| const int ss_ver = layout == DAV1D_PIXEL_LAYOUT_I420; 70| 32.0M| const int ss_hor = layout != DAV1D_PIXEL_LAYOUT_I444; 71| 32.0M| const int not_one_blk = b_dim[2] - (!!b_dim[2] && ss_hor) > t_dim->lw || ------------------ | Branch (71:33): [True: 21.7M, False: 10.2M] | Branch (71:45): [True: 30.9M, False: 1.10M] | Branch (71:59): [True: 5.54M, False: 25.4M] ------------------ 72| 10.2M| b_dim[3] - (!!b_dim[3] && ss_ver) > t_dim->lh; ------------------ | Branch (72:33): [True: 689k, False: 9.58M] | Branch (72:45): [True: 8.60M, False: 1.67M] | Branch (72:59): [True: 4.32M, False: 4.27M] ------------------ 73| 32.0M| unsigned ca, cl; 74| | 75| 32.0M|#define MERGE_CTX(dir, type, no_val) \ 76| 32.0M| c##dir = *(const type *) dir != no_val; \ 77| 32.0M| break 78| | 79| 32.0M| switch (t_dim->lw) { 80| | /* For some reason the MSVC CRT _wassert() function is not flagged as 81| | * __declspec(noreturn), so when using those headers the compiler will 82| | * expect execution to continue after an assertion has been triggered 83| | * and will therefore complain about the use of uninitialized variables 84| | * when compiled in debug mode if we put the default case at the end. */ 85| 0| default: assert(0); /* fall-through */ ------------------ | Branch (85:9): [True: 0, False: 32.0M] | Branch (85:18): [Folded, False: 0] ------------------ 86| 21.4M| case TX_4X4: MERGE_CTX(a, uint8_t, 0x40); ------------------ | | 76| 21.4M| c##dir = *(const type *) dir != no_val; \ | | 77| 21.4M| break ------------------ | Branch (86:9): [True: 21.4M, False: 10.6M] ------------------ 87| 3.41M| case TX_8X8: MERGE_CTX(a, uint16_t, 0x4040); ------------------ | | 76| 3.41M| c##dir = *(const type *) dir != no_val; \ | | 77| 3.41M| break ------------------ | Branch (87:9): [True: 3.41M, False: 28.6M] ------------------ 88| 2.49M| case TX_16X16: MERGE_CTX(a, uint32_t, 0x40404040U); ------------------ | | 76| 2.49M| c##dir = *(const type *) dir != no_val; \ | | 77| 2.49M| break ------------------ | Branch (88:9): [True: 2.49M, False: 29.5M] ------------------ 89| 4.77M| case TX_32X32: MERGE_CTX(a, uint64_t, 0x4040404040404040ULL); ------------------ | | 76| 4.77M| c##dir = *(const type *) dir != no_val; \ | | 77| 4.77M| break ------------------ | Branch (89:9): [True: 4.77M, False: 27.2M] ------------------ 90| 32.0M| } 91| 32.0M| switch (t_dim->lh) { 92| 0| default: assert(0); /* fall-through */ ------------------ | Branch (92:9): [True: 0, False: 32.0M] | Branch (92:18): [Folded, False: 0] ------------------ 93| 22.5M| case TX_4X4: MERGE_CTX(l, uint8_t, 0x40); ------------------ | | 76| 22.5M| c##dir = *(const type *) dir != no_val; \ | | 77| 22.5M| break ------------------ | Branch (93:9): [True: 22.5M, False: 9.56M] ------------------ 94| 3.07M| case TX_8X8: MERGE_CTX(l, uint16_t, 0x4040); ------------------ | | 76| 3.07M| c##dir = *(const type *) dir != no_val; \ | | 77| 3.07M| break ------------------ | Branch (94:9): [True: 3.07M, False: 28.9M] ------------------ 95| 2.17M| case TX_16X16: MERGE_CTX(l, uint32_t, 0x40404040U); ------------------ | | 76| 2.17M| c##dir = *(const type *) dir != no_val; \ | | 77| 2.17M| break ------------------ | Branch (95:9): [True: 2.17M, False: 29.8M] ------------------ 96| 4.38M| case TX_32X32: MERGE_CTX(l, uint64_t, 0x4040404040404040ULL); ------------------ | | 76| 4.38M| c##dir = *(const type *) dir != no_val; \ | | 77| 4.38M| break ------------------ | Branch (96:9): [True: 4.38M, False: 27.6M] ------------------ 97| 32.0M| } 98| 32.0M|#undef MERGE_CTX 99| | 100| 32.0M| return 7 + not_one_blk * 3 + ca + cl; 101| 32.0M| } else if (b_dim[2] == t_dim->lw && b_dim[3] == t_dim->lh) { ------------------ | Branch (101:16): [True: 6.20M, False: 11.7M] | Branch (101:41): [True: 5.98M, False: 224k] ------------------ 102| 5.98M| return 0; 103| 11.9M| } else { 104| 11.9M| unsigned la, ll; 105| | 106| 11.9M|#define MERGE_CTX(dir, type, tx) \ 107| 11.9M| if (tx == TX_64X64) { \ 108| 11.9M| uint64_t tmp = *(const uint64_t *) dir; \ 109| 11.9M| tmp |= *(const uint64_t *) &dir[8]; \ 110| 11.9M| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ 111| 11.9M| } else \ 112| 11.9M| l##dir = *(const type *) dir; \ 113| 11.9M| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ 114| 11.9M| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ 115| 11.9M| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ 116| 11.9M| break 117| | 118| 11.9M| switch (t_dim->lw) { 119| 0| default: assert(0); /* fall-through */ ------------------ | Branch (119:9): [True: 0, False: 11.9M] | Branch (119:18): [Folded, False: 0] ------------------ 120| 10.4M| case TX_4X4: MERGE_CTX(a, uint8_t, TX_4X4); ------------------ | | 107| 10.4M| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 10.4M] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 10.4M| l##dir = *(const type *) dir; \ | | 113| 10.4M| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 10.4M] | | ------------------ | | 114| 10.4M| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [Folded, False: 10.4M] | | ------------------ | | 115| 10.4M| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [Folded, False: 10.4M] | | ------------------ | | 116| 10.4M| break ------------------ | Branch (120:9): [True: 10.4M, False: 1.50M] ------------------ 121| 688k| case TX_8X8: MERGE_CTX(a, uint16_t, TX_8X8); ------------------ | | 107| 688k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 688k] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 688k| l##dir = *(const type *) dir; \ | | 113| 688k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 688k] | | ------------------ | | 114| 688k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [Folded, False: 688k] | | ------------------ | | 115| 688k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 688k, Folded] | | ------------------ | | 116| 688k| break ------------------ | Branch (121:9): [True: 688k, False: 11.2M] ------------------ 122| 372k| case TX_16X16: MERGE_CTX(a, uint32_t, TX_16X16); ------------------ | | 107| 372k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 372k] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 372k| l##dir = *(const type *) dir; \ | | 113| 372k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 372k] | | ------------------ | | 114| 372k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [True: 372k, Folded] | | ------------------ | | 115| 372k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 372k, Folded] | | ------------------ | | 116| 372k| break ------------------ | Branch (122:9): [True: 372k, False: 11.5M] ------------------ 123| 41.1k| case TX_32X32: MERGE_CTX(a, uint32_t, TX_32X32); ------------------ | | 107| 41.1k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 41.1k] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 41.1k| l##dir = *(const type *) dir; \ | | 113| 41.1k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [True: 41.1k, Folded] | | ------------------ | | 114| 41.1k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [True: 41.1k, Folded] | | ------------------ | | 115| 41.1k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 41.1k, Folded] | | ------------------ | | 116| 41.1k| break ------------------ | Branch (123:9): [True: 41.1k, False: 11.8M] ------------------ 124| 709k| case TX_64X64: MERGE_CTX(a, uint32_t, TX_64X64); ------------------ | | 107| 709k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [True: 709k, Folded] | | ------------------ | | 108| 709k| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 709k| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 709k| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 709k| } else \ | | 112| 18.4E| l##dir = *(const type *) dir; \ | | 113| 709k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 709k] | | ------------------ | | 114| 709k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [True: 709k, Folded] | | ------------------ | | 115| 709k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 709k, Folded] | | ------------------ | | 116| 709k| break ------------------ | Branch (124:9): [True: 709k, False: 11.2M] ------------------ 125| 11.9M| } 126| 12.2M| switch (t_dim->lh) { 127| 0| default: assert(0); /* fall-through */ ------------------ | Branch (127:9): [True: 0, False: 12.2M] | Branch (127:18): [Folded, False: 0] ------------------ 128| 10.4M| case TX_4X4: MERGE_CTX(l, uint8_t, TX_4X4); ------------------ | | 107| 10.4M| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 10.4M] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 10.4M| l##dir = *(const type *) dir; \ | | 113| 10.4M| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 10.4M] | | ------------------ | | 114| 10.4M| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [Folded, False: 10.4M] | | ------------------ | | 115| 10.4M| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [Folded, False: 10.4M] | | ------------------ | | 116| 10.4M| break ------------------ | Branch (128:9): [True: 10.4M, False: 1.79M] ------------------ 129| 679k| case TX_8X8: MERGE_CTX(l, uint16_t, TX_8X8); ------------------ | | 107| 679k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 679k] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 679k| l##dir = *(const type *) dir; \ | | 113| 679k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 679k] | | ------------------ | | 114| 679k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [Folded, False: 679k] | | ------------------ | | 115| 679k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 679k, Folded] | | ------------------ | | 116| 679k| break ------------------ | Branch (129:9): [True: 679k, False: 11.5M] ------------------ 130| 361k| case TX_16X16: MERGE_CTX(l, uint32_t, TX_16X16); ------------------ | | 107| 361k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 361k] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 361k| l##dir = *(const type *) dir; \ | | 113| 361k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 361k] | | ------------------ | | 114| 361k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [True: 361k, Folded] | | ------------------ | | 115| 361k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 361k, Folded] | | ------------------ | | 116| 361k| break ------------------ | Branch (130:9): [True: 361k, False: 11.8M] ------------------ 131| 41.1k| case TX_32X32: MERGE_CTX(l, uint32_t, TX_32X32); ------------------ | | 107| 41.1k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [Folded, False: 41.1k] | | ------------------ | | 108| 0| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 0| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 0| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 0| } else \ | | 112| 41.1k| l##dir = *(const type *) dir; \ | | 113| 41.1k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [True: 41.1k, Folded] | | ------------------ | | 114| 41.1k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [True: 41.1k, Folded] | | ------------------ | | 115| 41.1k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 41.1k, Folded] | | ------------------ | | 116| 41.1k| break ------------------ | Branch (131:9): [True: 41.1k, False: 12.1M] ------------------ 132| 709k| case TX_64X64: MERGE_CTX(l, uint32_t, TX_64X64); ------------------ | | 107| 709k| if (tx == TX_64X64) { \ | | ------------------ | | | Branch (107:13): [True: 709k, Folded] | | ------------------ | | 108| 709k| uint64_t tmp = *(const uint64_t *) dir; \ | | 109| 709k| tmp |= *(const uint64_t *) &dir[8]; \ | | 110| 709k| l##dir = (unsigned) (tmp >> 32) | (unsigned) tmp; \ | | 111| 709k| } else \ | | 112| 709k| l##dir = *(const type *) dir; \ | | 113| 709k| if (tx == TX_32X32) l##dir |= *(const type *) &dir[sizeof(type)]; \ | | ------------------ | | | Branch (113:13): [Folded, False: 709k] | | ------------------ | | 114| 709k| if (tx >= TX_16X16) l##dir |= l##dir >> 16; \ | | ------------------ | | | Branch (114:13): [True: 709k, Folded] | | ------------------ | | 115| 709k| if (tx >= TX_8X8) l##dir |= l##dir >> 8; \ | | ------------------ | | | Branch (115:13): [True: 709k, Folded] | | ------------------ | | 116| 709k| break ------------------ | Branch (132:9): [True: 709k, False: 11.5M] ------------------ 133| 12.2M| } 134| 12.2M|#undef MERGE_CTX 135| | 136| 12.2M| return dav1d_skip_ctx[umin(la & 0x3F, 4)][umin(ll & 0x3F, 4)]; 137| 12.2M| } 138| 49.9M|} recon_tmpl.c:get_lo_ctx: 304| 219M|{ 305| 219M| unsigned mag = levels[0 * stride + 1] + levels[1 * stride + 0]; 306| 219M| unsigned offset; 307| 219M| if (tx_class == TX_CLASS_2D) { ------------------ | Branch (307:9): [True: 207M, False: 12.2M] ------------------ 308| 207M| mag += levels[1 * stride + 1]; 309| 207M| *hi_mag = mag; 310| 207M| mag += levels[0 * stride + 2] + levels[2 * stride + 0]; 311| 207M| offset = ctx_offsets[umin(y, 4)][umin(x, 4)]; 312| 207M| } else { 313| 12.2M| mag += levels[0 * stride + 2]; 314| 12.2M| *hi_mag = mag; 315| 12.2M| mag += levels[0 * stride + 3] + levels[0 * stride + 4]; 316| 12.2M| offset = 26 + (y > 1 ? 10 : y * 5); ------------------ | Branch (316:24): [True: 8.00M, False: 4.29M] ------------------ 317| 12.2M| } 318| 219M| return offset + (mag > 512 ? 4 : (mag + 64) >> 7); ------------------ | Branch (318:22): [True: 12.7M, False: 207M] ------------------ 319| 219M|} recon_tmpl.c:get_dc_sign_ctx: 143| 14.9M|{ 144| 14.9M| uint64_t mask = 0xC0C0C0C0C0C0C0C0ULL, mul = 0x0101010101010101ULL; 145| 14.9M| int s; 146| | 147| 14.9M|#if ARCH_X86_64 && defined(__GNUC__) 148| | /* Coerce compilers into producing better code. For some reason 149| | * every x86-64 compiler is awful at handling 64-bit constants. */ 150| 14.9M| __asm__("" : "+r"(mask), "+r"(mul)); 151| 14.9M|#endif 152| | 153| 14.9M| switch(tx) { 154| 0| default: assert(0); /* fall-through */ ------------------ | Branch (154:5): [True: 0, False: 14.9M] | Branch (154:14): [Folded, False: 0] ------------------ 155| 7.49M| case TX_4X4: { ------------------ | Branch (155:5): [True: 7.49M, False: 7.47M] ------------------ 156| 7.49M| int t = *(const uint8_t *) a >> 6; 157| 7.49M| t += *(const uint8_t *) l >> 6; 158| 7.49M| s = t - 1 - 1; 159| 7.49M| break; 160| 0| } 161| 1.23M| case TX_8X8: { ------------------ | Branch (161:5): [True: 1.23M, False: 13.7M] ------------------ 162| 1.23M| uint32_t t = *(const uint16_t *) a & (uint32_t) mask; 163| 1.23M| t += *(const uint16_t *) l & (uint32_t) mask; 164| 1.23M| t *= 0x04040404U; 165| 1.23M| s = (int) (t >> 24) - 2 - 2; 166| 1.23M| break; 167| 0| } 168| 821k| case TX_16X16: { ------------------ | Branch (168:5): [True: 821k, False: 14.1M] ------------------ 169| 821k| uint32_t t = (*(const uint32_t *) a & (uint32_t) mask) >> 6; 170| 821k| t += (*(const uint32_t *) l & (uint32_t) mask) >> 6; 171| 821k| t *= (uint32_t) mul; 172| 821k| s = (int) (t >> 24) - 4 - 4; 173| 821k| break; 174| 0| } 175| 1.03M| case TX_32X32: { ------------------ | Branch (175:5): [True: 1.03M, False: 13.9M] ------------------ 176| 1.03M| uint64_t t = (*(const uint64_t *) a & mask) >> 6; 177| 1.03M| t += (*(const uint64_t *) l & mask) >> 6; 178| 1.03M| t *= mul; 179| 1.03M| s = (int) (t >> 56) - 8 - 8; 180| 1.03M| break; 181| 0| } 182| 596k| case TX_64X64: { ------------------ | Branch (182:5): [True: 596k, False: 14.3M] ------------------ 183| 596k| uint64_t t = (*(const uint64_t *) &a[0] & mask) >> 6; 184| 596k| t += (*(const uint64_t *) &a[8] & mask) >> 6; 185| 596k| t += (*(const uint64_t *) &l[0] & mask) >> 6; 186| 596k| t += (*(const uint64_t *) &l[8] & mask) >> 6; 187| 596k| t *= mul; 188| 596k| s = (int) (t >> 56) - 16 - 16; 189| 596k| break; 190| 0| } 191| 380k| case RTX_4X8: { ------------------ | Branch (191:5): [True: 380k, False: 14.5M] ------------------ 192| 380k| uint32_t t = *(const uint8_t *) a & (uint32_t) mask; 193| 380k| t += *(const uint16_t *) l & (uint32_t) mask; 194| 380k| t *= 0x04040404U; 195| 380k| s = (int) (t >> 24) - 1 - 2; 196| 380k| break; 197| 0| } 198| 645k| case RTX_8X4: { ------------------ | Branch (198:5): [True: 645k, False: 14.3M] ------------------ 199| 645k| uint32_t t = *(const uint16_t *) a & (uint32_t) mask; 200| 645k| t += *(const uint8_t *) l & (uint32_t) mask; 201| 645k| t *= 0x04040404U; 202| 645k| s = (int) (t >> 24) - 2 - 1; 203| 645k| break; 204| 0| } 205| 331k| case RTX_8X16: { ------------------ | Branch (205:5): [True: 331k, False: 14.6M] ------------------ 206| 331k| uint32_t t = *(const uint16_t *) a & (uint32_t) mask; 207| 331k| t += *(const uint32_t *) l & (uint32_t) mask; 208| 331k| t = (t >> 6) * (uint32_t) mul; 209| 331k| s = (int) (t >> 24) - 2 - 4; 210| 331k| break; 211| 0| } 212| 672k| case RTX_16X8: { ------------------ | Branch (212:5): [True: 672k, False: 14.3M] ------------------ 213| 672k| uint32_t t = *(const uint32_t *) a & (uint32_t) mask; 214| 672k| t += *(const uint16_t *) l & (uint32_t) mask; 215| 672k| t = (t >> 6) * (uint32_t) mul; 216| 672k| s = (int) (t >> 24) - 4 - 2; 217| 672k| break; 218| 0| } 219| 161k| case RTX_16X32: { ------------------ | Branch (219:5): [True: 161k, False: 14.8M] ------------------ 220| 161k| uint64_t t = *(const uint32_t *) a & (uint32_t) mask; 221| 161k| t += *(const uint64_t *) l & mask; 222| 161k| t = (t >> 6) * mul; 223| 161k| s = (int) (t >> 56) - 4 - 8; 224| 161k| break; 225| 0| } 226| 224k| case RTX_32X16: { ------------------ | Branch (226:5): [True: 224k, False: 14.7M] ------------------ 227| 224k| uint64_t t = *(const uint64_t *) a & mask; 228| 224k| t += *(const uint32_t *) l & (uint32_t) mask; 229| 224k| t = (t >> 6) * mul; 230| 224k| s = (int) (t >> 56) - 8 - 4; 231| 224k| break; 232| 0| } 233| 30.2k| case RTX_32X64: { ------------------ | Branch (233:5): [True: 30.2k, False: 14.9M] ------------------ 234| 30.2k| uint64_t t = (*(const uint64_t *) &a[0] & mask) >> 6; 235| 30.2k| t += (*(const uint64_t *) &l[0] & mask) >> 6; 236| 30.2k| t += (*(const uint64_t *) &l[8] & mask) >> 6; 237| 30.2k| t *= mul; 238| 30.2k| s = (int) (t >> 56) - 8 - 16; 239| 30.2k| break; 240| 0| } 241| 48.7k| case RTX_64X32: { ------------------ | Branch (241:5): [True: 48.7k, False: 14.9M] ------------------ 242| 48.7k| uint64_t t = (*(const uint64_t *) &a[0] & mask) >> 6; 243| 48.7k| t += (*(const uint64_t *) &a[8] & mask) >> 6; 244| 48.7k| t += (*(const uint64_t *) &l[0] & mask) >> 6; 245| 48.7k| t *= mul; 246| 48.7k| s = (int) (t >> 56) - 16 - 8; 247| 48.7k| break; 248| 0| } 249| 237k| case RTX_4X16: { ------------------ | Branch (249:5): [True: 237k, False: 14.7M] ------------------ 250| 237k| uint32_t t = *(const uint8_t *) a & (uint32_t) mask; 251| 237k| t += *(const uint32_t *) l & (uint32_t) mask; 252| 237k| t = (t >> 6) * (uint32_t) mul; 253| 237k| s = (int) (t >> 24) - 1 - 4; 254| 237k| break; 255| 0| } 256| 511k| case RTX_16X4: { ------------------ | Branch (256:5): [True: 511k, False: 14.4M] ------------------ 257| 511k| uint32_t t = *(const uint32_t *) a & (uint32_t) mask; 258| 511k| t += *(const uint8_t *) l & (uint32_t) mask; 259| 511k| t = (t >> 6) * (uint32_t) mul; 260| 511k| s = (int) (t >> 24) - 4 - 1; 261| 511k| break; 262| 0| } 263| 168k| case RTX_8X32: { ------------------ | Branch (263:5): [True: 168k, False: 14.8M] ------------------ 264| 168k| uint64_t t = *(const uint16_t *) a & (uint32_t) mask; 265| 168k| t += *(const uint64_t *) l & mask; 266| 168k| t = (t >> 6) * mul; 267| 168k| s = (int) (t >> 56) - 2 - 8; 268| 168k| break; 269| 0| } 270| 301k| case RTX_32X8: { ------------------ | Branch (270:5): [True: 301k, False: 14.6M] ------------------ 271| 301k| uint64_t t = *(const uint64_t *) a & mask; 272| 301k| t += *(const uint16_t *) l & (uint32_t) mask; 273| 301k| t = (t >> 6) * mul; 274| 301k| s = (int) (t >> 56) - 8 - 2; 275| 301k| break; 276| 0| } 277| 53.1k| case RTX_16X64: { ------------------ | Branch (277:5): [True: 53.1k, False: 14.9M] ------------------ 278| 53.1k| uint64_t t = *(const uint32_t *) a & (uint32_t) mask; 279| 53.1k| t += *(const uint64_t *) &l[0] & mask; 280| 53.1k| t = (t >> 6) + ((*(const uint64_t *) &l[8] & mask) >> 6); 281| 53.1k| t *= mul; 282| 53.1k| s = (int) (t >> 56) - 4 - 16; 283| 53.1k| break; 284| 0| } 285| 72.0k| case RTX_64X16: { ------------------ | Branch (285:5): [True: 72.0k, False: 14.9M] ------------------ 286| 72.0k| uint64_t t = *(const uint64_t *) &a[0] & mask; 287| 72.0k| t += *(const uint32_t *) l & (uint32_t) mask; 288| 72.0k| t = (t >> 6) + ((*(const uint64_t *) &a[8] & mask) >> 6); 289| 72.0k| t *= mul; 290| 72.0k| s = (int) (t >> 56) - 16 - 4; 291| 72.0k| break; 292| 0| } 293| 14.9M| } 294| | 295| 14.9M| return (s != 0) + (s > 0); 296| 14.9M|} recon_tmpl.c:read_golomb: 49| 3.08M|static inline unsigned read_golomb(MsacContext *const msac) { 50| 3.08M| int len = 0; 51| 3.08M| unsigned val = 1; 52| | 53| 7.01M| while (!dav1d_msac_decode_bool_equi(msac) && len < 32) len++; ------------------ | | 53| 7.01M|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (53:12): [True: 3.93M, False: 3.08M] | Branch (53:50): [True: 3.93M, False: 18.4E] ------------------ 54| 7.01M| while (len--) val = (val << 1) + dav1d_msac_decode_bool_equi(msac); ------------------ | | 53| 3.93M|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (54:12): [True: 3.93M, False: 3.08M] ------------------ 55| | 56| 3.08M| return val - 1; 57| 3.08M|} recon_tmpl.c:mc: 944| 5.12M|{ 945| 5.12M| assert((dst8 != NULL) ^ (dst16 != NULL)); ------------------ | Branch (945:5): [True: 5.12M, False: 181] ------------------ 946| 5.12M| const Dav1dFrameContext *const f = t->f; 947| 5.12M| const int ss_ver = !!pl && f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (947:24): [True: 2.62M, False: 2.49M] | Branch (947:32): [True: 2.13M, False: 487k] ------------------ 948| 5.12M| const int ss_hor = !!pl && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (948:24): [True: 2.62M, False: 2.49M] | Branch (948:32): [True: 2.14M, False: 484k] ------------------ 949| 5.12M| const int h_mul = 4 >> ss_hor, v_mul = 4 >> ss_ver; 950| 5.12M| const int mvx = mv.x, mvy = mv.y; 951| 5.12M| const int mx = mvx & (15 >> !ss_hor), my = mvy & (15 >> !ss_ver); 952| 5.12M| ptrdiff_t ref_stride = refp->p.stride[!!pl]; 953| 5.12M| const pixel *ref; 954| | 955| 5.12M| if (refp->p.p.w == f->cur.p.w && refp->p.p.h == f->cur.p.h) { ------------------ | Branch (955:9): [True: 4.59M, False: 532k] | Branch (955:38): [True: 4.45M, False: 137k] ------------------ 956| 4.45M| const int dx = bx * h_mul + (mvx >> (3 + ss_hor)); 957| 4.45M| const int dy = by * v_mul + (mvy >> (3 + ss_ver)); 958| 4.45M| int w, h; 959| | 960| 4.45M| if (refp->p.data[0] != f->cur.data[0]) { // i.e. not for intrabc ------------------ | Branch (960:13): [True: 4.40M, False: 53.9k] ------------------ 961| 4.40M| w = (f->cur.p.w + ss_hor) >> ss_hor; 962| 4.40M| h = (f->cur.p.h + ss_ver) >> ss_ver; 963| 4.40M| } else { 964| 53.9k| w = f->bw * 4 >> ss_hor; 965| 53.9k| h = f->bh * 4 >> ss_ver; 966| 53.9k| } 967| 4.45M| if (dx < !!mx * 3 || dy < !!my * 3 || ------------------ | Branch (967:13): [True: 97.7k, False: 4.35M] | Branch (967:30): [True: 55.8k, False: 4.30M] ------------------ 968| 4.30M| dx + bw4 * h_mul + !!mx * 4 > w || ------------------ | Branch (968:13): [True: 917k, False: 3.38M] ------------------ 969| 3.38M| dy + bh4 * v_mul + !!my * 4 > h) ------------------ | Branch (969:13): [True: 185k, False: 3.19M] ------------------ 970| 1.26M| { 971| 1.26M| pixel *const emu_edge_buf = bitfn(t->scratch.emu_edge); ------------------ | | 51| 1.26M|#define bitfn(x) x##_8bpc ------------------ 972| 1.26M| f->dsp->mc.emu_edge(bw4 * h_mul + !!mx * 7, bh4 * v_mul + !!my * 7, 973| 1.26M| w, h, dx - !!mx * 3, dy - !!my * 3, 974| 1.26M| emu_edge_buf, 192 * sizeof(pixel), 975| 1.26M| refp->p.data[pl], ref_stride); 976| 1.26M| ref = &emu_edge_buf[192 * !!my * 3 + !!mx * 3]; 977| 1.26M| ref_stride = 192 * sizeof(pixel); 978| 3.19M| } else { 979| 3.19M| ref = ((pixel *) refp->p.data[pl]) + PXSTRIDE(ref_stride) * dy + dx; ------------------ | | 53| 3.19M|#define PXSTRIDE(x) (x) ------------------ 980| 3.19M| } 981| | 982| 4.45M| if (dst8 != NULL) { ------------------ | Branch (982:13): [True: 3.89M, False: 564k] ------------------ 983| 3.89M| f->dsp->mc.mc[filter_2d](dst8, dst_stride, ref, ref_stride, bw4 * h_mul, 984| 3.89M| bh4 * v_mul, mx << !ss_hor, my << !ss_ver 985| 3.89M| HIGHBD_CALL_SUFFIX); 986| 3.89M| } else { 987| 564k| f->dsp->mc.mct[filter_2d](dst16, ref, ref_stride, bw4 * h_mul, 988| 564k| bh4 * v_mul, mx << !ss_hor, my << !ss_ver 989| 564k| HIGHBD_CALL_SUFFIX); 990| 564k| } 991| 4.45M| } else { 992| 670k| assert(refp != &f->sr_cur); ------------------ | Branch (992:9): [True: 673k, False: 18.4E] ------------------ 993| | 994| 673k| const int orig_pos_y = (by * v_mul << 4) + mvy * (1 << !ss_ver); 995| 673k| const int orig_pos_x = (bx * h_mul << 4) + mvx * (1 << !ss_hor); 996| 673k|#define scale_mv(res, val, scale) do { \ 997| 673k| const int64_t tmp = (int64_t)(val) * scale + (scale - 0x4000) * 8; \ 998| 673k| res = apply_sign64((int) ((llabs(tmp) + 128) >> 8), tmp) + 32; \ 999| 673k| } while (0) 1000| 673k| int pos_y, pos_x; 1001| 673k| scale_mv(pos_x, orig_pos_x, f->svc[refidx][0].scale); ------------------ | | 996| 673k|#define scale_mv(res, val, scale) do { \ | | 997| 673k| const int64_t tmp = (int64_t)(val) * scale + (scale - 0x4000) * 8; \ | | 998| 673k| res = apply_sign64((int) ((llabs(tmp) + 128) >> 8), tmp) + 32; \ | | 999| 673k| } while (0) | | ------------------ | | | Branch (999:18): [Folded, False: 673k] | | ------------------ ------------------ 1002| 673k| scale_mv(pos_y, orig_pos_y, f->svc[refidx][1].scale); ------------------ | | 996| 673k|#define scale_mv(res, val, scale) do { \ | | 997| 673k| const int64_t tmp = (int64_t)(val) * scale + (scale - 0x4000) * 8; \ | | 998| 673k| res = apply_sign64((int) ((llabs(tmp) + 128) >> 8), tmp) + 32; \ | | 999| 673k| } while (0) | | ------------------ | | | Branch (999:18): [Folded, False: 673k] | | ------------------ ------------------ 1003| 673k|#undef scale_mv 1004| 673k| const int left = pos_x >> 10; 1005| 673k| const int top = pos_y >> 10; 1006| 673k| const int right = 1007| 673k| ((pos_x + (bw4 * h_mul - 1) * f->svc[refidx][0].step) >> 10) + 1; 1008| 673k| const int bottom = 1009| 673k| ((pos_y + (bh4 * v_mul - 1) * f->svc[refidx][1].step) >> 10) + 1; 1010| | 1011| 673k| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 673k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 673k] | | ------------------ | | 35| 673k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 673k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1012| 0| printf("Off %dx%d [%d,%d,%d], size %dx%d [%d,%d]\n", 1013| 0| left, top, orig_pos_x, f->svc[refidx][0].scale, refidx, 1014| 0| right-left, bottom-top, 1015| 0| f->svc[refidx][0].step, f->svc[refidx][1].step); 1016| | 1017| 673k| const int w = (refp->p.p.w + ss_hor) >> ss_hor; 1018| 673k| const int h = (refp->p.p.h + ss_ver) >> ss_ver; 1019| 673k| if (left < 3 || top < 3 || right + 4 > w || bottom + 4 > h) { ------------------ | Branch (1019:13): [True: 132k, False: 541k] | Branch (1019:25): [True: 35.9k, False: 505k] | Branch (1019:36): [True: 110k, False: 394k] | Branch (1019:53): [True: 51.9k, False: 343k] ------------------ 1020| 327k| pixel *const emu_edge_buf = bitfn(t->scratch.emu_edge); ------------------ | | 51| 327k|#define bitfn(x) x##_8bpc ------------------ 1021| 327k| f->dsp->mc.emu_edge(right - left + 7, bottom - top + 7, 1022| 327k| w, h, left - 3, top - 3, 1023| 327k| emu_edge_buf, 320 * sizeof(pixel), 1024| 327k| refp->p.data[pl], ref_stride); 1025| 327k| ref = &emu_edge_buf[320 * 3 + 3]; 1026| 327k| ref_stride = 320 * sizeof(pixel); 1027| 327k| if (DEBUG_BLOCK_INFO) printf("Emu\n"); ------------------ | | 34| 327k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 327k] | | ------------------ | | 35| 327k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 327k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1028| 345k| } else { 1029| 345k| ref = ((pixel *) refp->p.data[pl]) + PXSTRIDE(ref_stride) * top + left; ------------------ | | 53| 345k|#define PXSTRIDE(x) (x) ------------------ 1030| 345k| } 1031| | 1032| 673k| if (dst8 != NULL) { ------------------ | Branch (1032:13): [True: 569k, False: 104k] ------------------ 1033| 569k| f->dsp->mc.mc_scaled[filter_2d](dst8, dst_stride, ref, ref_stride, 1034| 569k| bw4 * h_mul, bh4 * v_mul, 1035| 569k| pos_x & 0x3ff, pos_y & 0x3ff, 1036| 569k| f->svc[refidx][0].step, 1037| 569k| f->svc[refidx][1].step 1038| 569k| HIGHBD_CALL_SUFFIX); 1039| 569k| } else { 1040| 104k| f->dsp->mc.mct_scaled[filter_2d](dst16, ref, ref_stride, 1041| 104k| bw4 * h_mul, bh4 * v_mul, 1042| 104k| pos_x & 0x3ff, pos_y & 0x3ff, 1043| 104k| f->svc[refidx][0].step, 1044| 104k| f->svc[refidx][1].step 1045| 104k| HIGHBD_CALL_SUFFIX); 1046| 104k| } 1047| 673k| } 1048| | 1049| 5.12M| return 0; 1050| 5.12M|} recon_tmpl.c:warp_affine: 1120| 1.03M|{ 1121| 1.03M| assert((dst8 != NULL) ^ (dst16 != NULL)); ------------------ | Branch (1121:5): [True: 1.03M, False: 18.4E] ------------------ 1122| 1.03M| const Dav1dFrameContext *const f = t->f; 1123| 1.03M| const Dav1dDSPContext *const dsp = f->dsp; 1124| 1.03M| const int ss_ver = !!pl && f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (1124:24): [True: 162k, False: 874k] | Branch (1124:32): [True: 134k, False: 28.1k] ------------------ 1125| 1.03M| const int ss_hor = !!pl && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (1125:24): [True: 162k, False: 874k] | Branch (1125:32): [True: 134k, False: 28.0k] ------------------ 1126| 1.03M| const int h_mul = 4 >> ss_hor, v_mul = 4 >> ss_ver; 1127| 1.03M| assert(!((b_dim[0] * h_mul) & 7) && !((b_dim[1] * v_mul) & 7)); ------------------ | Branch (1127:5): [True: 1.03M, False: 18.4E] | Branch (1127:5): [True: 1.03M, False: 7] ------------------ 1128| 1.03M| const int32_t *const mat = wmp->matrix; 1129| 1.03M| const int width = (refp->p.p.w + ss_hor) >> ss_hor; 1130| 1.03M| const int height = (refp->p.p.h + ss_ver) >> ss_ver; 1131| | 1132| 9.01M| for (int y = 0; y < b_dim[1] * v_mul; y += 8) { ------------------ | Branch (1132:21): [True: 7.97M, False: 1.03M] ------------------ 1133| 7.97M| const int src_y = t->by * 4 + ((y + 4) << ss_ver); 1134| 7.97M| const int64_t mat3_y = (int64_t) mat[3] * src_y + mat[0]; 1135| 7.97M| const int64_t mat5_y = (int64_t) mat[5] * src_y + mat[1]; 1136| 80.6M| for (int x = 0; x < b_dim[0] * h_mul; x += 8) { ------------------ | Branch (1136:25): [True: 72.7M, False: 7.97M] ------------------ 1137| | // calculate transformation relative to center of 8x8 block in 1138| | // luma pixel units 1139| 72.7M| const int src_x = t->bx * 4 + ((x + 4) << ss_hor); 1140| 72.7M| const int64_t mvx = ((int64_t) mat[2] * src_x + mat3_y) >> ss_hor; 1141| 72.7M| const int64_t mvy = ((int64_t) mat[4] * src_x + mat5_y) >> ss_ver; 1142| | 1143| 72.7M| const int dx = (int) (mvx >> 16) - 4; 1144| 72.7M| const int mx = (((int) mvx & 0xffff) - wmp->u.p.alpha * 4 - 1145| 72.7M| wmp->u.p.beta * 7) & ~0x3f; 1146| 72.7M| const int dy = (int) (mvy >> 16) - 4; 1147| 72.7M| const int my = (((int) mvy & 0xffff) - wmp->u.p.gamma * 4 - 1148| 72.7M| wmp->u.p.delta * 4) & ~0x3f; 1149| | 1150| 72.7M| const pixel *ref_ptr; 1151| 72.7M| ptrdiff_t ref_stride = refp->p.stride[!!pl]; 1152| | 1153| 72.7M| if (dx < 3 || dx + 8 + 4 > width || dy < 3 || dy + 8 + 4 > height) { ------------------ | Branch (1153:17): [True: 4.33M, False: 68.3M] | Branch (1153:27): [True: 17.9M, False: 50.4M] | Branch (1153:49): [True: 82.3k, False: 50.3M] | Branch (1153:59): [True: 167k, False: 50.2M] ------------------ 1154| 24.2M| pixel *const emu_edge_buf = bitfn(t->scratch.emu_edge); ------------------ | | 51| 24.2M|#define bitfn(x) x##_8bpc ------------------ 1155| 24.2M| f->dsp->mc.emu_edge(15, 15, width, height, dx - 3, dy - 3, 1156| 24.2M| emu_edge_buf, 32 * sizeof(pixel), 1157| 24.2M| refp->p.data[pl], ref_stride); 1158| 24.2M| ref_ptr = &emu_edge_buf[32 * 3 + 3]; 1159| 24.2M| ref_stride = 32 * sizeof(pixel); 1160| 48.4M| } else { 1161| 48.4M| ref_ptr = ((pixel *) refp->p.data[pl]) + PXSTRIDE(ref_stride) * dy + dx; ------------------ | | 53| 48.4M|#define PXSTRIDE(x) (x) ------------------ 1162| 48.4M| } 1163| 72.7M| if (dst16 != NULL) ------------------ | Branch (1163:17): [True: 106k, False: 72.6M] ------------------ 1164| 106k| dsp->mc.warp8x8t(&dst16[x], dstride, ref_ptr, ref_stride, 1165| 106k| wmp->u.abcd, mx, my HIGHBD_CALL_SUFFIX); 1166| 72.6M| else 1167| 72.6M| dsp->mc.warp8x8(&dst8[x], dstride, ref_ptr, ref_stride, 1168| 72.6M| wmp->u.abcd, mx, my HIGHBD_CALL_SUFFIX); 1169| 72.7M| } 1170| 7.97M| if (dst8) dst8 += 8 * PXSTRIDE(dstride); ------------------ | | 53| 7.94M|#define PXSTRIDE(x) (x) ------------------ | Branch (1170:13): [True: 7.94M, False: 38.3k] ------------------ 1171| 38.3k| else dst16 += 8 * dstride; 1172| 7.97M| } 1173| 1.03M| return 0; 1174| 1.03M|} recon_tmpl.c:obmc: 1056| 575k|{ 1057| 575k| assert(!(t->bx & 1) && !(t->by & 1)); ------------------ | Branch (1057:5): [True: 575k, False: 18.4E] | Branch (1057:5): [True: 575k, False: 18.4E] ------------------ 1058| 575k| const Dav1dFrameContext *const f = t->f; 1059| 575k| /*const*/ refmvs_block **r = &t->rt.r[(t->by & 31) + 5]; 1060| 575k| pixel *const lap = bitfn(t->scratch.lap); ------------------ | | 51| 575k|#define bitfn(x) x##_8bpc ------------------ 1061| 575k| const int ss_ver = !!pl && f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (1061:24): [True: 364k, False: 210k] | Branch (1061:32): [True: 288k, False: 76.5k] ------------------ 1062| 575k| const int ss_hor = !!pl && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (1062:24): [True: 364k, False: 210k] | Branch (1062:32): [True: 289k, False: 75.9k] ------------------ 1063| 575k| const int h_mul = 4 >> ss_hor, v_mul = 4 >> ss_ver; 1064| 575k| int res; 1065| | 1066| 575k| if (t->by > t->ts->tiling.row_start && ------------------ | Branch (1066:9): [True: 558k, False: 17.2k] ------------------ 1067| 558k| (!pl || b_dim[0] * h_mul + b_dim[1] * v_mul >= 16)) ------------------ | Branch (1067:10): [True: 204k, False: 354k] | Branch (1067:17): [True: 151k, False: 202k] ------------------ 1068| 356k| { 1069| 759k| for (int i = 0, x = 0; x < w4 && i < imin(b_dim[2], 4); ) { ------------------ | Branch (1069:32): [True: 404k, False: 355k] | Branch (1069:42): [True: 403k, False: 1.21k] ------------------ 1070| | // only odd blocks are considered for overlap handling, hence +1 1071| 403k| const refmvs_block *const a_r = &r[-1][t->bx + x + 1]; 1072| 403k| const uint8_t *const a_b_dim = dav1d_block_dimensions[a_r->bs]; 1073| 403k| const int step4 = iclip(a_b_dim[0], 2, 16); 1074| | 1075| 403k| if (a_r->ref.ref[0] > 0) { ------------------ | Branch (1075:17): [True: 379k, False: 23.8k] ------------------ 1076| 379k| const int ow4 = imin(step4, b_dim[0]); 1077| 379k| const int oh4 = imin(b_dim[1], 16) >> 1; 1078| 379k| res = mc(t, lap, NULL, ow4 * h_mul * sizeof(pixel), ow4, (oh4 * 3 + 3) >> 2, 1079| 379k| t->bx + x, t->by, pl, a_r->mv.mv[0], 1080| 379k| &f->refp[a_r->ref.ref[0] - 1], a_r->ref.ref[0] - 1, 1081| 379k| dav1d_filter_2d[t->a->filter[1][bx4 + x + 1]][t->a->filter[0][bx4 + x + 1]]); 1082| 379k| if (res) return res; ------------------ | Branch (1082:21): [True: 0, False: 379k] ------------------ 1083| 379k| f->dsp->mc.blend_h(&dst[x * h_mul], dst_stride, lap, 1084| 379k| h_mul * ow4, v_mul * oh4); 1085| 379k| i++; 1086| 379k| } 1087| 403k| x += step4; 1088| 403k| } 1089| 356k| } 1090| | 1091| 575k| if (t->bx > t->ts->tiling.col_start) ------------------ | Branch (1091:9): [True: 553k, False: 22.4k] ------------------ 1092| 1.15M| for (int i = 0, y = 0; y < h4 && i < imin(b_dim[3], 4); ) { ------------------ | Branch (1092:32): [True: 604k, False: 552k] | Branch (1092:42): [True: 603k, False: 1.33k] ------------------ 1093| | // only odd blocks are considered for overlap handling, hence +1 1094| 603k| const refmvs_block *const l_r = &r[y + 1][t->bx - 1]; 1095| 603k| const uint8_t *const l_b_dim = dav1d_block_dimensions[l_r->bs]; 1096| 603k| const int step4 = iclip(l_b_dim[1], 2, 16); 1097| | 1098| 603k| if (l_r->ref.ref[0] > 0) { ------------------ | Branch (1098:17): [True: 559k, False: 43.4k] ------------------ 1099| 559k| const int ow4 = imin(b_dim[0], 16) >> 1; 1100| 559k| const int oh4 = imin(step4, b_dim[1]); 1101| 559k| res = mc(t, lap, NULL, h_mul * ow4 * sizeof(pixel), ow4, oh4, 1102| 559k| t->bx, t->by + y, pl, l_r->mv.mv[0], 1103| 559k| &f->refp[l_r->ref.ref[0] - 1], l_r->ref.ref[0] - 1, 1104| 559k| dav1d_filter_2d[t->l.filter[1][by4 + y + 1]][t->l.filter[0][by4 + y + 1]]); 1105| 559k| if (res) return res; ------------------ | Branch (1105:21): [True: 0, False: 559k] ------------------ 1106| 559k| f->dsp->mc.blend_v(&dst[y * v_mul * PXSTRIDE(dst_stride)], ------------------ | | 53| 559k|#define PXSTRIDE(x) (x) ------------------ 1107| 559k| dst_stride, lap, h_mul * ow4, v_mul * oh4); 1108| 559k| i++; 1109| 559k| } 1110| 603k| y += step4; 1111| 603k| } 1112| 575k| return 0; 1113| 575k|} dav1d_read_coef_blocks_16bpc: 826| 6.05M|{ 827| 6.05M| const Dav1dFrameContext *const f = t->f; 828| 6.05M| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 829| 6.05M| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 830| 6.05M| const int bx4 = t->bx & 31, by4 = t->by & 31; 831| 6.05M| const int cbx4 = bx4 >> ss_hor, cby4 = by4 >> ss_ver; 832| 6.05M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 833| 6.05M| const int bw4 = b_dim[0], bh4 = b_dim[1]; 834| 6.05M| const int cbw4 = (bw4 + ss_hor) >> ss_hor, cbh4 = (bh4 + ss_ver) >> ss_ver; 835| 6.05M| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 && ------------------ | Branch (835:28): [True: 4.41M, False: 1.64M] ------------------ 836| 4.41M| (bw4 > ss_hor || t->bx & 1) && ------------------ | Branch (836:29): [True: 4.08M, False: 326k] | Branch (836:45): [True: 163k, False: 163k] ------------------ 837| 4.24M| (bh4 > ss_ver || t->by & 1); ------------------ | Branch (837:29): [True: 3.89M, False: 352k] | Branch (837:45): [True: 176k, False: 176k] ------------------ 838| | 839| 6.05M| if (b->skip) { ------------------ | Branch (839:9): [True: 2.24M, False: 3.80M] ------------------ 840| 2.24M| BlockContext *const a = t->a; 841| 2.24M| dav1d_memset_pow2[b_dim[2]](&a->lcoef[bx4], 0x40); 842| 2.24M| dav1d_memset_pow2[b_dim[3]](&t->l.lcoef[by4], 0x40); 843| 2.24M| if (has_chroma) { ------------------ | Branch (843:13): [True: 1.16M, False: 1.08M] ------------------ 844| 1.16M| dav1d_memset_pow2_fn memset_cw = dav1d_memset_pow2[ulog2(cbw4)]; 845| 1.16M| dav1d_memset_pow2_fn memset_ch = dav1d_memset_pow2[ulog2(cbh4)]; 846| 1.16M| memset_cw(&a->ccoef[0][cbx4], 0x40); 847| 1.16M| memset_cw(&a->ccoef[1][cbx4], 0x40); 848| 1.16M| memset_ch(&t->l.ccoef[0][cby4], 0x40); 849| 1.16M| memset_ch(&t->l.ccoef[1][cby4], 0x40); 850| 1.16M| } 851| 2.24M| return; 852| 2.24M| } 853| | 854| 3.80M| Dav1dTileState *const ts = t->ts; 855| 3.80M| const int w4 = imin(bw4, f->bw - t->bx), h4 = imin(bh4, f->bh - t->by); 856| 3.80M| const int cw4 = (w4 + ss_hor) >> ss_hor, ch4 = (h4 + ss_ver) >> ss_ver; 857| 3.80M| assert(t->frame_thread.pass == 1); ------------------ | Branch (857:5): [True: 3.80M, False: 2.81k] ------------------ 858| 3.80M| assert(!b->skip); ------------------ | Branch (858:5): [True: 3.80M, False: 18.4E] ------------------ 859| 3.80M| const TxfmInfo *const uv_t_dim = &dav1d_txfm_dimensions[b->uvtx]; 860| 3.80M| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[b->intra ? b->tx : b->max_ytx]; ------------------ | Branch (860:58): [True: 2.63M, False: 1.16M] ------------------ 861| 3.80M| const uint16_t tx_split[2] = { b->tx_split0, b->tx_split1 }; 862| | 863| 7.75M| for (int init_y = 0; init_y < h4; init_y += 16) { ------------------ | Branch (863:26): [True: 3.94M, False: 3.80M] ------------------ 864| 3.94M| const int sub_h4 = imin(h4, 16 + init_y); 865| 8.17M| for (int init_x = 0; init_x < w4; init_x += 16) { ------------------ | Branch (865:30): [True: 4.23M, False: 3.94M] ------------------ 866| 4.23M| const int sub_w4 = imin(w4, init_x + 16); 867| 4.23M| int y_off = !!init_y, y, x; 868| 9.15M| for (y = init_y, t->by += init_y; y < sub_h4; ------------------ | Branch (868:47): [True: 4.92M, False: 4.23M] ------------------ 869| 4.92M| y += t_dim->h, t->by += t_dim->h, y_off++) 870| 4.92M| { 871| 4.92M| int x_off = !!init_x; 872| 19.0M| for (x = init_x, t->bx += init_x; x < sub_w4; ------------------ | Branch (872:51): [True: 14.0M, False: 4.92M] ------------------ 873| 14.0M| x += t_dim->w, t->bx += t_dim->w, x_off++) 874| 14.0M| { 875| 14.0M| if (!b->intra) { ------------------ | Branch (875:25): [True: 5.17M, False: 8.89M] ------------------ 876| 5.17M| read_coef_tree(t, bs, b, b->max_ytx, 0, tx_split, 877| 5.17M| x_off, y_off, NULL); 878| 8.89M| } else { 879| 8.89M| uint8_t cf_ctx = 0x40; 880| 8.89M| enum TxfmType txtp; 881| 8.89M| const int eob = 882| 8.89M| decode_coefs(t, &t->a->lcoef[bx4 + x], 883| 8.89M| &t->l.lcoef[by4 + y], b->tx, bs, b, 1, 884| 8.89M| 0, ts->frame_thread[1].cf, &txtp, &cf_ctx); 885| 8.89M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 8.89M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 8.89M] | | ------------------ | | 35| 8.89M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 8.89M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 886| 0| printf("Post-y-cf-blk[tx=%d,txtp=%d,eob=%d]: r=%d\n", 887| 0| b->tx, txtp, eob, ts->msac.rng); 888| 8.89M| *ts->frame_thread[1].cbi++ = eob * (1 << 5) + txtp; 889| 8.89M| ts->frame_thread[1].cf += imin(t_dim->w, 8) * imin(t_dim->h, 8) * 16; 890| 8.89M| dav1d_memset_likely_pow2(&t->a->lcoef[bx4 + x], cf_ctx, imin(t_dim->w, f->bw - t->bx)); 891| 8.89M| dav1d_memset_likely_pow2(&t->l.lcoef[by4 + y], cf_ctx, imin(t_dim->h, f->bh - t->by)); 892| 8.89M| } 893| 14.0M| } 894| 4.92M| t->bx -= x; 895| 4.92M| } 896| 4.23M| t->by -= y; 897| | 898| 4.23M| if (!has_chroma) continue; ------------------ | Branch (898:17): [True: 927k, False: 3.30M] ------------------ 899| | 900| 3.30M| const int sub_ch4 = imin(ch4, (init_y + 16) >> ss_ver); 901| 3.30M| const int sub_cw4 = imin(cw4, (init_x + 16) >> ss_hor); 902| 9.88M| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (902:30): [True: 6.57M, False: 3.30M] ------------------ 903| 14.7M| for (y = init_y >> ss_ver, t->by += init_y; y < sub_ch4; ------------------ | Branch (903:61): [True: 8.15M, False: 6.57M] ------------------ 904| 8.15M| y += uv_t_dim->h, t->by += uv_t_dim->h << ss_ver) 905| 8.15M| { 906| 33.8M| for (x = init_x >> ss_hor, t->bx += init_x; x < sub_cw4; ------------------ | Branch (906:65): [True: 25.6M, False: 8.15M] ------------------ 907| 25.6M| x += uv_t_dim->w, t->bx += uv_t_dim->w << ss_hor) 908| 25.6M| { 909| 25.6M| uint8_t cf_ctx = 0x40; 910| 25.6M| enum TxfmType txtp; 911| 25.6M| if (!b->intra) ------------------ | Branch (911:29): [True: 9.67M, False: 15.9M] ------------------ 912| 9.67M| txtp = t->scratch.txtp_map[(by4 + (y << ss_ver)) * 32 + 913| 9.67M| bx4 + (x << ss_hor)]; 914| 25.6M| const int eob = 915| 25.6M| decode_coefs(t, &t->a->ccoef[pl][cbx4 + x], 916| 25.6M| &t->l.ccoef[pl][cby4 + y], b->uvtx, bs, 917| 25.6M| b, b->intra, 1 + pl, ts->frame_thread[1].cf, 918| 25.6M| &txtp, &cf_ctx); 919| 25.6M| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 25.6M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 25.6M] | | ------------------ | | 35| 25.6M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 25.6M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 920| 0| printf("Post-uv-cf-blk[pl=%d,tx=%d," 921| 0| "txtp=%d,eob=%d]: r=%d\n", 922| 0| pl, b->uvtx, txtp, eob, ts->msac.rng); 923| 25.6M| *ts->frame_thread[1].cbi++ = eob * (1 << 5) + txtp; 924| 25.6M| ts->frame_thread[1].cf += uv_t_dim->w * uv_t_dim->h * 16; 925| 25.6M| int ctw = imin(uv_t_dim->w, (f->bw - t->bx + ss_hor) >> ss_hor); 926| 25.6M| int cth = imin(uv_t_dim->h, (f->bh - t->by + ss_ver) >> ss_ver); 927| 25.6M| dav1d_memset_likely_pow2(&t->a->ccoef[pl][cbx4 + x], cf_ctx, ctw); 928| 25.6M| dav1d_memset_likely_pow2(&t->l.ccoef[pl][cby4 + y], cf_ctx, cth); 929| 25.6M| } 930| 8.15M| t->bx -= x << ss_hor; 931| 8.15M| } 932| 6.57M| t->by -= y << ss_ver; 933| 6.57M| } 934| 3.30M| } 935| 3.94M| } 936| 3.80M|} dav1d_recon_b_intra_16bpc: 1179| 887k|{ 1180| 887k| Dav1dTileState *const ts = t->ts; 1181| 887k| const Dav1dFrameContext *const f = t->f; 1182| 887k| const Dav1dDSPContext *const dsp = f->dsp; 1183| 887k| const int bx4 = t->bx & 31, by4 = t->by & 31; 1184| 887k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 1185| 887k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 1186| 887k| const int cbx4 = bx4 >> ss_hor, cby4 = by4 >> ss_ver; 1187| 887k| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 1188| 887k| const int bw4 = b_dim[0], bh4 = b_dim[1]; 1189| 887k| const int w4 = imin(bw4, f->bw - t->bx), h4 = imin(bh4, f->bh - t->by); 1190| 887k| const int cw4 = (w4 + ss_hor) >> ss_hor, ch4 = (h4 + ss_ver) >> ss_ver; 1191| 887k| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 && ------------------ | Branch (1191:28): [True: 539k, False: 347k] ------------------ 1192| 539k| (bw4 > ss_hor || t->bx & 1) && ------------------ | Branch (1192:29): [True: 468k, False: 71.7k] | Branch (1192:45): [True: 36.3k, False: 35.4k] ------------------ 1193| 504k| (bh4 > ss_ver || t->by & 1); ------------------ | Branch (1193:29): [True: 432k, False: 71.7k] | Branch (1193:45): [True: 36.4k, False: 35.3k] ------------------ 1194| 887k| const TxfmInfo *const t_dim = &dav1d_txfm_dimensions[b->tx]; 1195| 887k| const TxfmInfo *const uv_t_dim = &dav1d_txfm_dimensions[b->uvtx]; 1196| | 1197| | // coefficient coding 1198| 887k| pixel *const edge = bitfn(t->scratch.edge) + 128; ------------------ | | 77| 887k|#define bitfn(x) x##_16bpc ------------------ 1199| 887k| const int cbw4 = (bw4 + ss_hor) >> ss_hor, cbh4 = (bh4 + ss_ver) >> ss_ver; 1200| | 1201| 887k| const int intra_edge_filter_flag = f->seq_hdr->intra_edge_filter << 10; 1202| | 1203| 1.92M| for (int init_y = 0; init_y < h4; init_y += 16) { ------------------ | Branch (1203:26): [True: 1.03M, False: 889k] ------------------ 1204| 1.03M| const int sub_h4 = imin(h4, 16 + init_y); 1205| 1.03M| const int sub_ch4 = imin(ch4, (init_y + 16) >> ss_ver); 1206| 2.33M| for (int init_x = 0; init_x < w4; init_x += 16) { ------------------ | Branch (1206:30): [True: 1.30M, False: 1.03M] ------------------ 1207| 1.30M| if (b->pal_sz[0]) { ------------------ | Branch (1207:17): [True: 8.21k, False: 1.29M] ------------------ 1208| 8.21k| pixel *dst = ((pixel *) f->cur.data[0]) + 1209| 8.21k| 4 * (t->by * PXSTRIDE(f->cur.stride[0]) + t->bx); 1210| 8.21k| const uint8_t *pal_idx; 1211| 8.21k| if (t->frame_thread.pass) { ------------------ | Branch (1211:21): [True: 8.21k, False: 0] ------------------ 1212| 8.21k| const int p = t->frame_thread.pass & 1; 1213| 8.21k| assert(ts->frame_thread[p].pal_idx); ------------------ | Branch (1213:21): [True: 8.21k, False: 0] ------------------ 1214| 8.21k| pal_idx = ts->frame_thread[p].pal_idx; 1215| 8.21k| ts->frame_thread[p].pal_idx += bw4 * bh4 * 8; 1216| 8.21k| } else { 1217| 0| pal_idx = t->scratch.pal_idx_y; 1218| 0| } 1219| 8.21k| const pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (1219:42): [True: 8.21k, False: 0] ------------------ 1220| 8.21k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 1221| 8.21k| ((t->bx >> 1) + (t->by & 1))][0] : 1222| 8.21k| bytefn(t->scratch.pal)[0]; ------------------ | | 87| 0|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 8.21k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 1223| 8.21k| f->dsp->ipred.pal_pred(dst, f->cur.stride[0], pal, 1224| 8.21k| pal_idx, bw4 * 4, bh4 * 4); 1225| 8.21k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 8.21k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 8.21k] | | ------------------ | | 35| 8.21k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 8.21k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1226| 0| hex_dump(dst, PXSTRIDE(f->cur.stride[0]), 1227| 0| bw4 * 4, bh4 * 4, "y-pal-pred"); 1228| 8.21k| } 1229| | 1230| 1.30M| const int intra_flags = (sm_flag(t->a, bx4) | 1231| 1.30M| sm_flag(&t->l, by4) | 1232| 1.30M| intra_edge_filter_flag); 1233| 1.30M| const int sb_has_tr = init_x + 16 < w4 ? 1 : init_y ? 0 : ------------------ | Branch (1233:35): [True: 276k, False: 1.03M] | Branch (1233:58): [True: 143k, False: 887k] ------------------ 1234| 1.03M| intra_edge_flags & EDGE_I444_TOP_HAS_RIGHT; 1235| 1.30M| const int sb_has_bl = init_x ? 0 : init_y + 16 < h4 ? 1 : ------------------ | Branch (1235:35): [True: 276k, False: 1.03M] | Branch (1235:48): [True: 143k, False: 887k] ------------------ 1236| 1.03M| intra_edge_flags & EDGE_I444_LEFT_HAS_BOTTOM; 1237| 1.30M| int y, x; 1238| 1.30M| const int sub_w4 = imin(w4, init_x + 16); 1239| 3.00M| for (y = init_y, t->by += init_y; y < sub_h4; ------------------ | Branch (1239:47): [True: 1.69M, False: 1.30M] ------------------ 1240| 1.69M| y += t_dim->h, t->by += t_dim->h) 1241| 1.69M| { 1242| 1.69M| pixel *dst = ((pixel *) f->cur.data[0]) + 1243| 1.69M| 4 * (t->by * PXSTRIDE(f->cur.stride[0]) + 1244| 1.69M| t->bx + init_x); 1245| 8.02M| for (x = init_x, t->bx += init_x; x < sub_w4; ------------------ | Branch (1245:51): [True: 6.32M, False: 1.69M] ------------------ 1246| 6.32M| x += t_dim->w, t->bx += t_dim->w) 1247| 6.32M| { 1248| 6.32M| if (b->pal_sz[0]) goto skip_y_pred; ------------------ | Branch (1248:25): [True: 62.9k, False: 6.25M] ------------------ 1249| | 1250| 6.25M| int angle = b->y_angle; 1251| 6.25M| const enum EdgeFlags edge_flags = 1252| 6.25M| (((y > init_y || !sb_has_tr) && (x + t_dim->w >= sub_w4)) ? ------------------ | Branch (1252:28): [True: 4.56M, False: 1.69M] | Branch (1252:42): [True: 471k, False: 1.22M] | Branch (1252:57): [True: 737k, False: 4.30M] ------------------ 1253| 5.52M| 0 : EDGE_I444_TOP_HAS_RIGHT) | 1254| 6.25M| ((x > init_x || (!sb_has_bl && y + t_dim->h >= sub_h4)) ? ------------------ | Branch (1254:27): [True: 4.58M, False: 1.67M] | Branch (1254:42): [True: 1.06M, False: 611k] | Branch (1254:56): [True: 781k, False: 285k] ------------------ 1255| 5.36M| 0 : EDGE_I444_LEFT_HAS_BOTTOM); 1256| 6.25M| const pixel *top_sb_edge = NULL; 1257| 6.25M| if (!(t->by & (f->sb_step - 1))) { ------------------ | Branch (1257:25): [True: 750k, False: 5.50M] ------------------ 1258| 750k| top_sb_edge = f->ipred_edge[0]; 1259| 750k| const int sby = t->by >> f->sb_shift; 1260| 750k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1261| 750k| } 1262| 6.25M| const enum IntraPredMode m = 1263| 6.25M| bytefn(dav1d_prepare_intra_edges)(t->bx, ------------------ | | 87| 6.25M|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 6.25M|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 1264| 6.25M| t->bx > ts->tiling.col_start, 1265| 6.25M| t->by, 1266| 6.25M| t->by > ts->tiling.row_start, 1267| 6.25M| ts->tiling.col_end, 1268| 6.25M| ts->tiling.row_end, 1269| 6.25M| edge_flags, dst, 1270| 6.25M| f->cur.stride[0], top_sb_edge, 1271| 6.25M| b->y_mode, &angle, 1272| 6.25M| t_dim->w, t_dim->h, 1273| 6.25M| f->seq_hdr->intra_edge_filter, 1274| 6.25M| edge HIGHBD_CALL_SUFFIX); ------------------ | | 73| 6.25M|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1275| 6.25M| dsp->ipred.intra_pred[m](dst, f->cur.stride[0], edge, 1276| 6.25M| t_dim->w * 4, t_dim->h * 4, 1277| 6.25M| angle | intra_flags, 1278| 6.25M| 4 * f->bw - 4 * t->bx, 1279| 6.25M| 4 * f->bh - 4 * t->by 1280| 6.25M| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 6.25M|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1281| | 1282| 6.25M| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 6.25M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 6.25M] | | ------------------ | | 35| 6.25M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 6.25M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1283| 0| hex_dump(edge - t_dim->h * 4, t_dim->h * 4, 1284| 0| t_dim->h * 4, 2, "l"); 1285| 0| hex_dump(edge, 0, 1, 1, "tl"); 1286| 0| hex_dump(edge + 1, t_dim->w * 4, 1287| 0| t_dim->w * 4, 2, "t"); 1288| 0| hex_dump(dst, f->cur.stride[0], 1289| 0| t_dim->w * 4, t_dim->h * 4, "y-intra-pred"); 1290| 0| } 1291| | 1292| 6.32M| skip_y_pred: {} 1293| 6.32M| if (!b->skip) { ------------------ | Branch (1293:25): [True: 853k, False: 5.47M] ------------------ 1294| 853k| coef *cf; 1295| 853k| int eob; 1296| 853k| enum TxfmType txtp; 1297| 853k| if (t->frame_thread.pass) { ------------------ | Branch (1297:29): [True: 853k, False: 18.4E] ------------------ 1298| 853k| const int p = t->frame_thread.pass & 1; 1299| 853k| const int cbi = *ts->frame_thread[p].cbi++; 1300| 853k| cf = ts->frame_thread[p].cf; 1301| 853k| ts->frame_thread[p].cf += imin(t_dim->w, 8) * imin(t_dim->h, 8) * 16; 1302| 853k| eob = cbi >> 5; 1303| 853k| txtp = cbi & 0x1f; 1304| 18.4E| } else { 1305| 18.4E| uint8_t cf_ctx; 1306| 18.4E| cf = bitfn(t->cf); ------------------ | | 77| 18.4E|#define bitfn(x) x##_16bpc ------------------ 1307| 18.4E| eob = decode_coefs(t, &t->a->lcoef[bx4 + x], 1308| 18.4E| &t->l.lcoef[by4 + y], b->tx, bs, 1309| 18.4E| b, 1, 0, cf, &txtp, &cf_ctx); 1310| 18.4E| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 18.4E|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 18.4E] | | ------------------ | | 35| 18.4E| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 18.4E| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1311| 0| printf("Post-y-cf-blk[tx=%d,txtp=%d,eob=%d]: r=%d\n", 1312| 0| b->tx, txtp, eob, ts->msac.rng); 1313| 18.4E| dav1d_memset_likely_pow2(&t->a->lcoef[bx4 + x], cf_ctx, imin(t_dim->w, f->bw - t->bx)); 1314| 18.4E| dav1d_memset_likely_pow2(&t->l.lcoef[by4 + y], cf_ctx, imin(t_dim->h, f->bh - t->by)); 1315| 18.4E| } 1316| 853k| if (eob >= 0) { ------------------ | Branch (1316:29): [True: 653k, False: 199k] ------------------ 1317| 653k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 653k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 653k] | | ------------------ | | 35| 653k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 653k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1318| 0| coef_dump(cf, imin(t_dim->h, 8) * 4, 1319| 0| imin(t_dim->w, 8) * 4, 3, "dq"); 1320| 653k| dsp->itx.itxfm_add[b->tx] 1321| 653k| [txtp](dst, 1322| 653k| f->cur.stride[0], 1323| 653k| cf, eob HIGHBD_CALL_SUFFIX); ------------------ | | 73| 653k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1324| 653k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 653k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 653k] | | ------------------ | | 35| 653k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 653k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1325| 0| hex_dump(dst, f->cur.stride[0], 1326| 0| t_dim->w * 4, t_dim->h * 4, "recon"); 1327| 653k| } 1328| 5.47M| } else if (!t->frame_thread.pass) { ------------------ | Branch (1328:32): [True: 0, False: 5.47M] ------------------ 1329| 0| dav1d_memset_pow2[t_dim->lw](&t->a->lcoef[bx4 + x], 0x40); 1330| 0| dav1d_memset_pow2[t_dim->lh](&t->l.lcoef[by4 + y], 0x40); 1331| 0| } 1332| 6.32M| dst += 4 * t_dim->w; 1333| 6.32M| } 1334| 1.69M| t->bx -= x; 1335| 1.69M| } 1336| 1.30M| t->by -= y; 1337| | 1338| 1.30M| if (!has_chroma) continue; ------------------ | Branch (1338:17): [True: 641k, False: 666k] ------------------ 1339| | 1340| 666k| const ptrdiff_t stride = f->cur.stride[1]; 1341| | 1342| 666k| if (b->uv_mode == CFL_PRED) { ------------------ | Branch (1342:17): [True: 37.5k, False: 628k] ------------------ 1343| 37.5k| assert(!init_x && !init_y); ------------------ | Branch (1343:17): [True: 37.5k, False: 0] | Branch (1343:17): [True: 37.5k, False: 0] ------------------ 1344| | 1345| 37.5k| int16_t *const ac = t->scratch.ac; 1346| 37.5k| pixel *y_src = ((pixel *) f->cur.data[0]) + 4 * (t->bx & ~ss_hor) + 1347| 37.5k| 4 * (t->by & ~ss_ver) * PXSTRIDE(f->cur.stride[0]); 1348| 37.5k| const ptrdiff_t uv_off = 4 * ((t->bx >> ss_hor) + 1349| 37.5k| (t->by >> ss_ver) * PXSTRIDE(stride)); 1350| 37.5k| pixel *const uv_dst[2] = { ((pixel *) f->cur.data[1]) + uv_off, 1351| 37.5k| ((pixel *) f->cur.data[2]) + uv_off }; 1352| | 1353| 37.5k| const int furthest_r = 1354| 37.5k| ((cw4 << ss_hor) + t_dim->w - 1) & ~(t_dim->w - 1); 1355| 37.5k| const int furthest_b = 1356| 37.5k| ((ch4 << ss_ver) + t_dim->h - 1) & ~(t_dim->h - 1); 1357| 37.5k| dsp->ipred.cfl_ac[f->cur.p.layout - 1](ac, y_src, f->cur.stride[0], 1358| 37.5k| cbw4 - (furthest_r >> ss_hor), 1359| 37.5k| cbh4 - (furthest_b >> ss_ver), 1360| 37.5k| cbw4 * 4, cbh4 * 4); 1361| 112k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1361:34): [True: 75.0k, False: 37.5k] ------------------ 1362| 75.0k| if (!b->cfl_alpha[pl]) continue; ------------------ | Branch (1362:25): [True: 16.7k, False: 58.3k] ------------------ 1363| 58.3k| int angle = 0; 1364| 58.3k| const pixel *top_sb_edge = NULL; 1365| 58.3k| if (!((t->by & ~ss_ver) & (f->sb_step - 1))) { ------------------ | Branch (1365:25): [True: 7.27k, False: 51.0k] ------------------ 1366| 7.27k| top_sb_edge = f->ipred_edge[pl + 1]; 1367| 7.27k| const int sby = t->by >> f->sb_shift; 1368| 7.27k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1369| 7.27k| } 1370| 58.3k| const int xpos = t->bx >> ss_hor, ypos = t->by >> ss_ver; 1371| 58.3k| const int xstart = ts->tiling.col_start >> ss_hor; 1372| 58.3k| const int ystart = ts->tiling.row_start >> ss_ver; 1373| 58.3k| const enum IntraPredMode m = 1374| 58.3k| bytefn(dav1d_prepare_intra_edges)(xpos, xpos > xstart, ------------------ | | 87| 58.3k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 58.3k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 1375| 58.3k| ypos, ypos > ystart, 1376| 58.3k| ts->tiling.col_end >> ss_hor, 1377| 58.3k| ts->tiling.row_end >> ss_ver, 1378| 58.3k| 0, uv_dst[pl], stride, 1379| 58.3k| top_sb_edge, DC_PRED, &angle, 1380| 58.3k| uv_t_dim->w, uv_t_dim->h, 0, 1381| 58.3k| edge HIGHBD_CALL_SUFFIX); ------------------ | | 73| 58.3k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1382| 58.3k| dsp->ipred.cfl_pred[m](uv_dst[pl], stride, edge, 1383| 58.3k| uv_t_dim->w * 4, 1384| 58.3k| uv_t_dim->h * 4, 1385| 58.3k| ac, b->cfl_alpha[pl] 1386| 58.3k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 58.3k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1387| 58.3k| } 1388| 37.5k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 37.5k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 37.5k] | | ------------------ | | 35| 37.5k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 37.5k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1389| 0| ac_dump(ac, 4*cbw4, 4*cbh4, "ac"); 1390| 0| hex_dump(uv_dst[0], stride, cbw4 * 4, cbh4 * 4, "u-cfl-pred"); 1391| 0| hex_dump(uv_dst[1], stride, cbw4 * 4, cbh4 * 4, "v-cfl-pred"); 1392| 0| } 1393| 628k| } else if (b->pal_sz[1]) { ------------------ | Branch (1393:24): [True: 1.96k, False: 626k] ------------------ 1394| 1.96k| const ptrdiff_t uv_dstoff = 4 * ((t->bx >> ss_hor) + 1395| 1.96k| (t->by >> ss_ver) * PXSTRIDE(f->cur.stride[1])); 1396| 1.96k| const pixel (*pal)[8]; 1397| 1.96k| const uint8_t *pal_idx; 1398| 1.96k| if (t->frame_thread.pass) { ------------------ | Branch (1398:21): [True: 1.96k, False: 0] ------------------ 1399| 1.96k| const int p = t->frame_thread.pass & 1; 1400| 1.96k| assert(ts->frame_thread[p].pal_idx); ------------------ | Branch (1400:21): [True: 1.96k, False: 0] ------------------ 1401| 1.96k| pal = f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 1402| 1.96k| ((t->bx >> 1) + (t->by & 1))]; 1403| 1.96k| pal_idx = ts->frame_thread[p].pal_idx; 1404| 1.96k| ts->frame_thread[p].pal_idx += cbw4 * cbh4 * 8; 1405| 1.96k| } else { 1406| 0| pal = bytefn(t->scratch.pal); ------------------ | | 87| 0|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 0|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 1407| 0| pal_idx = t->scratch.pal_idx_uv; 1408| 0| } 1409| | 1410| 1.96k| f->dsp->ipred.pal_pred(((pixel *) f->cur.data[1]) + uv_dstoff, 1411| 1.96k| f->cur.stride[1], pal[1], 1412| 1.96k| pal_idx, cbw4 * 4, cbh4 * 4); 1413| 1.96k| f->dsp->ipred.pal_pred(((pixel *) f->cur.data[2]) + uv_dstoff, 1414| 1.96k| f->cur.stride[1], pal[2], 1415| 1.96k| pal_idx, cbw4 * 4, cbh4 * 4); 1416| 1.96k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 1.96k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.96k] | | ------------------ | | 35| 1.96k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.96k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1417| 0| hex_dump(((pixel *) f->cur.data[1]) + uv_dstoff, 1418| 0| PXSTRIDE(f->cur.stride[1]), 1419| 0| cbw4 * 4, cbh4 * 4, "u-pal-pred"); 1420| 0| hex_dump(((pixel *) f->cur.data[2]) + uv_dstoff, 1421| 0| PXSTRIDE(f->cur.stride[1]), 1422| 0| cbw4 * 4, cbh4 * 4, "v-pal-pred"); 1423| 0| } 1424| 1.96k| } 1425| | 1426| 666k| const int sm_uv_fl = sm_uv_flag(t->a, cbx4) | 1427| 666k| sm_uv_flag(&t->l, cby4); 1428| 666k| const int uv_sb_has_tr = 1429| 666k| ((init_x + 16) >> ss_hor) < cw4 ? 1 : init_y ? 0 : ------------------ | Branch (1429:17): [True: 127k, False: 538k] | Branch (1429:55): [True: 68.2k, False: 469k] ------------------ 1430| 538k| intra_edge_flags & (EDGE_I420_TOP_HAS_RIGHT >> (f->cur.p.layout - 1)); 1431| 666k| const int uv_sb_has_bl = 1432| 666k| init_x ? 0 : ((init_y + 16) >> ss_ver) < ch4 ? 1 : ------------------ | Branch (1432:17): [True: 127k, False: 538k] | Branch (1432:30): [True: 68.2k, False: 469k] ------------------ 1433| 538k| intra_edge_flags & (EDGE_I420_LEFT_HAS_BOTTOM >> (f->cur.p.layout - 1)); 1434| 666k| const int sub_cw4 = imin(cw4, (init_x + 16) >> ss_hor); 1435| 1.99M| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1435:30): [True: 1.33M, False: 667k] ------------------ 1436| 3.53M| for (y = init_y >> ss_ver, t->by += init_y; y < sub_ch4; ------------------ | Branch (1436:61): [True: 2.20M, False: 1.33M] ------------------ 1437| 2.20M| y += uv_t_dim->h, t->by += uv_t_dim->h << ss_ver) 1438| 2.20M| { 1439| 2.20M| pixel *dst = ((pixel *) f->cur.data[1 + pl]) + 1440| 2.20M| 4 * ((t->by >> ss_ver) * PXSTRIDE(stride) + 1441| 2.20M| ((t->bx + init_x) >> ss_hor)); 1442| 12.4M| for (x = init_x >> ss_hor, t->bx += init_x; x < sub_cw4; ------------------ | Branch (1442:65): [True: 10.2M, False: 2.20M] ------------------ 1443| 10.2M| x += uv_t_dim->w, t->bx += uv_t_dim->w << ss_hor) 1444| 10.2M| { 1445| 10.2M| if ((b->uv_mode == CFL_PRED && b->cfl_alpha[pl]) || ------------------ | Branch (1445:30): [True: 75.0k, False: 10.1M] | Branch (1445:56): [True: 58.3k, False: 16.7k] ------------------ 1446| 10.1M| b->pal_sz[1]) ------------------ | Branch (1446:29): [True: 9.32k, False: 10.1M] ------------------ 1447| 68.3k| { 1448| 68.3k| goto skip_uv_pred; 1449| 68.3k| } 1450| | 1451| 10.1M| int angle = b->uv_angle; 1452| | // this probably looks weird because we're using 1453| | // luma flags in a chroma loop, but that's because 1454| | // prepare_intra_edges() expects luma flags as input 1455| 10.1M| const enum EdgeFlags edge_flags = 1456| 10.1M| (((y > (init_y >> ss_ver) || !uv_sb_has_tr) && ------------------ | Branch (1456:32): [True: 8.31M, False: 1.86M] | Branch (1456:58): [True: 457k, False: 1.40M] ------------------ 1457| 8.77M| (x + uv_t_dim->w >= sub_cw4)) ? ------------------ | Branch (1457:31): [True: 1.19M, False: 7.58M] ------------------ 1458| 8.98M| 0 : EDGE_I444_TOP_HAS_RIGHT) | 1459| 10.1M| ((x > (init_x >> ss_hor) || ------------------ | Branch (1459:31): [True: 8.03M, False: 2.14M] ------------------ 1460| 2.14M| (!uv_sb_has_bl && y + uv_t_dim->h >= sub_ch4)) ? ------------------ | Branch (1460:32): [True: 1.11M, False: 1.02M] | Branch (1460:49): [True: 662k, False: 456k] ------------------ 1461| 8.70M| 0 : EDGE_I444_LEFT_HAS_BOTTOM); 1462| 10.1M| const pixel *top_sb_edge = NULL; 1463| 10.1M| if (!((t->by & ~ss_ver) & (f->sb_step - 1))) { ------------------ | Branch (1463:29): [True: 934k, False: 9.24M] ------------------ 1464| 934k| top_sb_edge = f->ipred_edge[1 + pl]; 1465| 934k| const int sby = t->by >> f->sb_shift; 1466| 934k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1467| 934k| } 1468| 10.1M| const enum IntraPredMode uv_mode = 1469| 10.1M| b->uv_mode == CFL_PRED ? DC_PRED : b->uv_mode; ------------------ | Branch (1469:30): [True: 16.7k, False: 10.1M] ------------------ 1470| 10.1M| const int xpos = t->bx >> ss_hor, ypos = t->by >> ss_ver; 1471| 10.1M| const int xstart = ts->tiling.col_start >> ss_hor; 1472| 10.1M| const int ystart = ts->tiling.row_start >> ss_ver; 1473| 10.1M| const enum IntraPredMode m = 1474| 10.1M| bytefn(dav1d_prepare_intra_edges)(xpos, xpos > xstart, ------------------ | | 87| 10.1M|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 10.1M|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 1475| 10.1M| ypos, ypos > ystart, 1476| 10.1M| ts->tiling.col_end >> ss_hor, 1477| 10.1M| ts->tiling.row_end >> ss_ver, 1478| 10.1M| edge_flags, dst, stride, 1479| 10.1M| top_sb_edge, uv_mode, 1480| 10.1M| &angle, uv_t_dim->w, 1481| 10.1M| uv_t_dim->h, 1482| 10.1M| f->seq_hdr->intra_edge_filter, 1483| 10.1M| edge HIGHBD_CALL_SUFFIX); ------------------ | | 73| 10.1M|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1484| 10.1M| angle |= intra_edge_filter_flag; 1485| 10.1M| dsp->ipred.intra_pred[m](dst, stride, edge, 1486| 10.1M| uv_t_dim->w * 4, 1487| 10.1M| uv_t_dim->h * 4, 1488| 10.1M| angle | sm_uv_fl, 1489| 10.1M| (4 * f->bw + ss_hor - 1490| 10.1M| 4 * (t->bx & ~ss_hor)) >> ss_hor, 1491| 10.1M| (4 * f->bh + ss_ver - 1492| 10.1M| 4 * (t->by & ~ss_ver)) >> ss_ver 1493| 10.1M| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 10.1M|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1494| 10.1M| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 10.1M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 10.1M] | | ------------------ | | 35| 10.1M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 10.1M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1495| 0| hex_dump(edge - uv_t_dim->h * 4, uv_t_dim->h * 4, 1496| 0| uv_t_dim->h * 4, 2, "l"); 1497| 0| hex_dump(edge, 0, 1, 1, "tl"); 1498| 0| hex_dump(edge + 1, uv_t_dim->w * 4, 1499| 0| uv_t_dim->w * 4, 2, "t"); 1500| 0| hex_dump(dst, stride, uv_t_dim->w * 4, 1501| 0| uv_t_dim->h * 4, pl ? "v-intra-pred" : "u-intra-pred"); ------------------ | Branch (1501:55): [True: 0, False: 0] ------------------ 1502| 0| } 1503| | 1504| 10.2M| skip_uv_pred: {} 1505| 10.2M| if (!b->skip) { ------------------ | Branch (1505:29): [True: 1.02M, False: 9.21M] ------------------ 1506| 1.02M| enum TxfmType txtp; 1507| 1.02M| int eob; 1508| 1.02M| coef *cf; 1509| 1.02M| if (t->frame_thread.pass) { ------------------ | Branch (1509:33): [True: 1.02M, False: 18.4E] ------------------ 1510| 1.02M| const int p = t->frame_thread.pass & 1; 1511| 1.02M| const int cbi = *ts->frame_thread[p].cbi++; 1512| 1.02M| cf = ts->frame_thread[p].cf; 1513| 1.02M| ts->frame_thread[p].cf += uv_t_dim->w * uv_t_dim->h * 16; 1514| 1.02M| eob = cbi >> 5; 1515| 1.02M| txtp = cbi & 0x1f; 1516| 18.4E| } else { 1517| 18.4E| uint8_t cf_ctx; 1518| 18.4E| cf = bitfn(t->cf); ------------------ | | 77| 18.4E|#define bitfn(x) x##_16bpc ------------------ 1519| 18.4E| eob = decode_coefs(t, &t->a->ccoef[pl][cbx4 + x], 1520| 18.4E| &t->l.ccoef[pl][cby4 + y], 1521| 18.4E| b->uvtx, bs, b, 1, 1 + pl, cf, 1522| 18.4E| &txtp, &cf_ctx); 1523| 18.4E| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 18.4E|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 18.4E] | | ------------------ | | 35| 18.4E| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 18.4E| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1524| 0| printf("Post-uv-cf-blk[pl=%d,tx=%d," 1525| 0| "txtp=%d,eob=%d]: r=%d [x=%d,cbx4=%d]\n", 1526| 0| pl, b->uvtx, txtp, eob, ts->msac.rng, x, cbx4); 1527| 18.4E| int ctw = imin(uv_t_dim->w, (f->bw - t->bx + ss_hor) >> ss_hor); 1528| 18.4E| int cth = imin(uv_t_dim->h, (f->bh - t->by + ss_ver) >> ss_ver); 1529| 18.4E| dav1d_memset_likely_pow2(&t->a->ccoef[pl][cbx4 + x], cf_ctx, ctw); 1530| 18.4E| dav1d_memset_likely_pow2(&t->l.ccoef[pl][cby4 + y], cf_ctx, cth); 1531| 18.4E| } 1532| 1.02M| if (eob >= 0) { ------------------ | Branch (1532:33): [True: 230k, False: 798k] ------------------ 1533| 230k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 230k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 230k] | | ------------------ | | 35| 230k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 230k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1534| 0| coef_dump(cf, uv_t_dim->h * 4, 1535| 0| uv_t_dim->w * 4, 3, "dq"); 1536| 230k| dsp->itx.itxfm_add[b->uvtx] 1537| 230k| [txtp](dst, stride, 1538| 230k| cf, eob HIGHBD_CALL_SUFFIX); ------------------ | | 73| 230k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1539| 230k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 230k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 230k] | | ------------------ | | 35| 230k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 230k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1540| 0| hex_dump(dst, stride, uv_t_dim->w * 4, 1541| 0| uv_t_dim->h * 4, "recon"); 1542| 230k| } 1543| 9.21M| } else if (!t->frame_thread.pass) { ------------------ | Branch (1543:36): [True: 0, False: 9.21M] ------------------ 1544| 0| dav1d_memset_pow2[uv_t_dim->lw](&t->a->ccoef[pl][cbx4 + x], 0x40); 1545| 0| dav1d_memset_pow2[uv_t_dim->lh](&t->l.ccoef[pl][cby4 + y], 0x40); 1546| 0| } 1547| 10.2M| dst += uv_t_dim->w * 4; 1548| 10.2M| } 1549| 2.20M| t->bx -= x << ss_hor; 1550| 2.20M| } 1551| 1.33M| t->by -= y << ss_ver; 1552| 1.33M| } 1553| 666k| } 1554| 1.03M| } 1555| 887k|} dav1d_recon_b_inter_16bpc: 1559| 1.04M|{ 1560| 1.04M| Dav1dTileState *const ts = t->ts; 1561| 1.04M| const Dav1dFrameContext *const f = t->f; 1562| 1.04M| const Dav1dDSPContext *const dsp = f->dsp; 1563| 1.04M| const int bx4 = t->bx & 31, by4 = t->by & 31; 1564| 1.04M| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 1565| 1.04M| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 1566| 1.04M| const int cbx4 = bx4 >> ss_hor, cby4 = by4 >> ss_ver; 1567| 1.04M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 1568| 1.04M| const int bw4 = b_dim[0], bh4 = b_dim[1]; 1569| 1.04M| const int w4 = imin(bw4, f->bw - t->bx), h4 = imin(bh4, f->bh - t->by); 1570| 1.04M| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400 && ------------------ | Branch (1570:28): [True: 285k, False: 754k] ------------------ 1571| 285k| (bw4 > ss_hor || t->bx & 1) && ------------------ | Branch (1571:29): [True: 256k, False: 28.4k] | Branch (1571:45): [True: 13.7k, False: 14.7k] ------------------ 1572| 270k| (bh4 > ss_ver || t->by & 1); ------------------ | Branch (1572:29): [True: 240k, False: 29.6k] | Branch (1572:45): [True: 14.2k, False: 15.3k] ------------------ 1573| 1.04M| const int chr_layout_idx = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I400 ? 0 : ------------------ | Branch (1573:32): [True: 754k, False: 285k] ------------------ 1574| 1.04M| DAV1D_PIXEL_LAYOUT_I444 - f->cur.p.layout; 1575| 1.04M| int res; 1576| | 1577| | // prediction 1578| 1.04M| const int cbh4 = (bh4 + ss_ver) >> ss_ver, cbw4 = (bw4 + ss_hor) >> ss_hor; 1579| 1.04M| pixel *dst = ((pixel *) f->cur.data[0]) + 1580| 1.04M| 4 * (t->by * PXSTRIDE(f->cur.stride[0]) + t->bx); 1581| 1.04M| const ptrdiff_t uvdstoff = 1582| 1.04M| 4 * ((t->bx >> ss_hor) + (t->by >> ss_ver) * PXSTRIDE(f->cur.stride[1])); 1583| 1.04M| if (IS_KEY_OR_INTRA(f->frame_hdr)) { ------------------ | | 43| 1.04M| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 1.04M| ((frame_header)->frame_type & 1) | | ------------------ | | | Branch (43:5): [True: 6.97k, False: 1.03M] | | ------------------ ------------------ 1584| | // intrabc 1585| 6.97k| assert(!f->frame_hdr->super_res.enabled); ------------------ | Branch (1585:9): [True: 6.97k, False: 0] ------------------ 1586| 6.97k| res = mc(t, dst, NULL, f->cur.stride[0], bw4, bh4, t->bx, t->by, 0, 1587| 6.97k| b->mv[0], &f->sr_cur, 0 /* unused */, FILTER_2D_BILINEAR); 1588| 6.97k| if (res) return res; ------------------ | Branch (1588:13): [True: 0, False: 6.97k] ------------------ 1589| 16.0k| if (has_chroma) for (int pl = 1; pl < 3; pl++) { ------------------ | Branch (1589:13): [True: 5.35k, False: 1.62k] | Branch (1589:42): [True: 10.7k, False: 5.35k] ------------------ 1590| 10.7k| res = mc(t, ((pixel *)f->cur.data[pl]) + uvdstoff, NULL, f->cur.stride[1], 1591| 10.7k| bw4 << (bw4 == ss_hor), bh4 << (bh4 == ss_ver), 1592| 10.7k| t->bx & ~ss_hor, t->by & ~ss_ver, pl, b->mv[0], 1593| 10.7k| &f->sr_cur, 0 /* unused */, FILTER_2D_BILINEAR); 1594| 10.7k| if (res) return res; ------------------ | Branch (1594:17): [True: 0, False: 10.7k] ------------------ 1595| 10.7k| } 1596| 1.03M| } else if (b->comp_type == COMP_INTER_NONE) { ------------------ | Branch (1596:16): [True: 981k, False: 51.8k] ------------------ 1597| 981k| const Dav1dThreadPicture *const refp = &f->refp[b->ref[0]]; 1598| 981k| const enum Filter2d filter_2d = b->filter2d; 1599| | 1600| 981k| if (imin(bw4, bh4) > 1 && ------------------ | Branch (1600:13): [True: 884k, False: 96.9k] ------------------ 1601| 884k| ((b->inter_mode == GLOBALMV && f->gmv_warp_allowed[b->ref[0]]) || ------------------ | Branch (1601:15): [True: 758k, False: 125k] | Branch (1601:44): [True: 123k, False: 634k] ------------------ 1602| 760k| (b->motion_mode == MM_WARP && t->warpmv.type > DAV1D_WM_TYPE_TRANSLATION))) ------------------ | Branch (1602:15): [True: 37.2k, False: 723k] | Branch (1602:44): [True: 32.9k, False: 4.29k] ------------------ 1603| 156k| { 1604| 156k| res = warp_affine(t, dst, NULL, f->cur.stride[0], b_dim, 0, refp, 1605| 156k| b->motion_mode == MM_WARP ? &t->warpmv : ------------------ | Branch (1605:31): [True: 32.9k, False: 123k] ------------------ 1606| 156k| &f->frame_hdr->gmv[b->ref[0]]); 1607| 156k| if (res) return res; ------------------ | Branch (1607:17): [True: 0, False: 156k] ------------------ 1608| 824k| } else { 1609| 824k| res = mc(t, dst, NULL, f->cur.stride[0], 1610| 824k| bw4, bh4, t->bx, t->by, 0, b->mv[0], refp, b->ref[0], filter_2d); 1611| 824k| if (res) return res; ------------------ | Branch (1611:17): [True: 0, False: 824k] ------------------ 1612| 824k| if (b->motion_mode == MM_OBMC) { ------------------ | Branch (1612:17): [True: 50.0k, False: 774k] ------------------ 1613| 50.0k| res = obmc(t, dst, f->cur.stride[0], b_dim, 0, bx4, by4, w4, h4); 1614| 50.0k| if (res) return res; ------------------ | Branch (1614:21): [True: 0, False: 50.0k] ------------------ 1615| 50.0k| } 1616| 824k| } 1617| 981k| if (b->interintra_type) { ------------------ | Branch (1617:13): [True: 11.0k, False: 970k] ------------------ 1618| 11.0k| pixel *const tl_edge = bitfn(t->scratch.edge) + 32; ------------------ | | 77| 11.0k|#define bitfn(x) x##_16bpc ------------------ 1619| 11.0k| enum IntraPredMode m = b->interintra_mode == II_SMOOTH_PRED ? ------------------ | Branch (1619:36): [True: 2.00k, False: 9.01k] ------------------ 1620| 9.01k| SMOOTH_PRED : b->interintra_mode; 1621| 11.0k| pixel *const tmp = bitfn(t->scratch.interintra); ------------------ | | 77| 11.0k|#define bitfn(x) x##_16bpc ------------------ 1622| 11.0k| int angle = 0; 1623| 11.0k| const pixel *top_sb_edge = NULL; 1624| 11.0k| if (!(t->by & (f->sb_step - 1))) { ------------------ | Branch (1624:17): [True: 1.53k, False: 9.48k] ------------------ 1625| 1.53k| top_sb_edge = f->ipred_edge[0]; 1626| 1.53k| const int sby = t->by >> f->sb_shift; 1627| 1.53k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1628| 1.53k| } 1629| 11.0k| m = bytefn(dav1d_prepare_intra_edges)(t->bx, t->bx > ts->tiling.col_start, ------------------ | | 87| 11.0k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 11.0k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 1630| 11.0k| t->by, t->by > ts->tiling.row_start, 1631| 11.0k| ts->tiling.col_end, ts->tiling.row_end, 1632| 11.0k| 0, dst, f->cur.stride[0], top_sb_edge, 1633| 11.0k| m, &angle, bw4, bh4, 0, tl_edge 1634| 11.0k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 11.0k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1635| 11.0k| dsp->ipred.intra_pred[m](tmp, 4 * bw4 * sizeof(pixel), 1636| 11.0k| tl_edge, bw4 * 4, bh4 * 4, 0, 0, 0 1637| 11.0k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 11.0k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1638| 11.0k| dsp->mc.blend(dst, f->cur.stride[0], tmp, 1639| 11.0k| bw4 * 4, bh4 * 4, II_MASK(0, bs, b)); ------------------ | | 83| 11.0k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 84| 11.0k| (size_t)((b)->interintra_type == INTER_INTRA_BLEND ? \ | | ------------------ | | | Branch (84:14): [True: 8.38k, False: 2.63k] | | ------------------ | | 85| 11.0k| dav1d_masks.offsets[c][(bs)-BS_32x32].ii[(b)->interintra_mode] : \ | | 86| 11.0k| dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[0][(b)->wedge_idx]) * 8)) ------------------ 1640| 11.0k| } 1641| | 1642| 981k| if (!has_chroma) goto skip_inter_chroma_pred; ------------------ | Branch (1642:13): [True: 751k, False: 229k] ------------------ 1643| | 1644| | // sub8x8 derivation 1645| 229k| int is_sub8x8 = bw4 == ss_hor || bh4 == ss_ver; ------------------ | Branch (1645:25): [True: 9.70k, False: 220k] | Branch (1645:42): [True: 11.5k, False: 208k] ------------------ 1646| 229k| refmvs_block *const *r; 1647| 229k| if (is_sub8x8) { ------------------ | Branch (1647:13): [True: 21.5k, False: 208k] ------------------ 1648| 21.5k| assert(ss_hor == 1); ------------------ | Branch (1648:13): [True: 21.5k, False: 0] ------------------ 1649| 21.5k| r = &t->rt.r[(t->by & 31) + 5]; 1650| 21.5k| if (bw4 == 1) is_sub8x8 &= r[0][t->bx - 1].ref.ref[0] > 0; ------------------ | Branch (1650:17): [True: 9.96k, False: 11.5k] ------------------ 1651| 21.5k| if (bh4 == ss_ver) is_sub8x8 &= r[-1][t->bx].ref.ref[0] > 0; ------------------ | Branch (1651:17): [True: 13.7k, False: 7.78k] ------------------ 1652| 21.5k| if (bw4 == 1 && bh4 == ss_ver) ------------------ | Branch (1652:17): [True: 9.96k, False: 11.5k] | Branch (1652:29): [True: 2.18k, False: 7.78k] ------------------ 1653| 2.18k| is_sub8x8 &= r[-1][t->bx - 1].ref.ref[0] > 0; 1654| 21.5k| } 1655| | 1656| | // chroma prediction 1657| 229k| if (is_sub8x8) { ------------------ | Branch (1657:13): [True: 16.8k, False: 213k] ------------------ 1658| 16.8k| assert(ss_hor == 1); ------------------ | Branch (1658:13): [True: 16.8k, False: 0] ------------------ 1659| 16.8k| ptrdiff_t h_off = 0, v_off = 0; 1660| 16.8k| if (bw4 == 1 && bh4 == ss_ver) { ------------------ | Branch (1660:17): [True: 7.81k, False: 9.07k] | Branch (1660:29): [True: 1.54k, False: 6.27k] ------------------ 1661| 4.62k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1661:34): [True: 3.08k, False: 1.54k] ------------------ 1662| 3.08k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, 1663| 3.08k| NULL, f->cur.stride[1], 1664| 3.08k| bw4, bh4, t->bx - 1, t->by - 1, 1 + pl, 1665| 3.08k| r[-1][t->bx - 1].mv.mv[0], 1666| 3.08k| &f->refp[r[-1][t->bx - 1].ref.ref[0] - 1], 1667| 3.08k| r[-1][t->bx - 1].ref.ref[0] - 1, 1668| 3.08k| t->frame_thread.pass != 2 ? t->tl_4x4_filter : ------------------ | Branch (1668:30): [True: 0, False: 3.08k] ------------------ 1669| 3.08k| f->frame_thread.b[((t->by - 1) * f->b4_stride) + t->bx - 1].filter2d); 1670| 3.08k| if (res) return res; ------------------ | Branch (1670:25): [True: 0, False: 3.08k] ------------------ 1671| 3.08k| } 1672| 1.54k| v_off = 2 * PXSTRIDE(f->cur.stride[1]); 1673| 1.54k| h_off = 2; 1674| 1.54k| } 1675| 16.8k| if (bw4 == 1) { ------------------ | Branch (1675:17): [True: 7.81k, False: 9.07k] ------------------ 1676| 7.81k| const enum Filter2d left_filter_2d = 1677| 7.81k| dav1d_filter_2d[t->l.filter[1][by4]][t->l.filter[0][by4]]; 1678| 23.4k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1678:34): [True: 15.6k, False: 7.81k] ------------------ 1679| 15.6k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff + v_off, NULL, 1680| 15.6k| f->cur.stride[1], bw4, bh4, t->bx - 1, 1681| 15.6k| t->by, 1 + pl, r[0][t->bx - 1].mv.mv[0], 1682| 15.6k| &f->refp[r[0][t->bx - 1].ref.ref[0] - 1], 1683| 15.6k| r[0][t->bx - 1].ref.ref[0] - 1, 1684| 15.6k| t->frame_thread.pass != 2 ? left_filter_2d : ------------------ | Branch (1684:30): [True: 0, False: 15.6k] ------------------ 1685| 15.6k| f->frame_thread.b[(t->by * f->b4_stride) + t->bx - 1].filter2d); 1686| 15.6k| if (res) return res; ------------------ | Branch (1686:25): [True: 0, False: 15.6k] ------------------ 1687| 15.6k| } 1688| 7.81k| h_off = 2; 1689| 7.81k| } 1690| 16.8k| if (bh4 == ss_ver) { ------------------ | Branch (1690:17): [True: 10.6k, False: 6.27k] ------------------ 1691| 10.6k| const enum Filter2d top_filter_2d = 1692| 10.6k| dav1d_filter_2d[t->a->filter[1][bx4]][t->a->filter[0][bx4]]; 1693| 31.8k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1693:34): [True: 21.2k, False: 10.6k] ------------------ 1694| 21.2k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff + h_off, NULL, 1695| 21.2k| f->cur.stride[1], bw4, bh4, t->bx, t->by - 1, 1696| 21.2k| 1 + pl, r[-1][t->bx].mv.mv[0], 1697| 21.2k| &f->refp[r[-1][t->bx].ref.ref[0] - 1], 1698| 21.2k| r[-1][t->bx].ref.ref[0] - 1, 1699| 21.2k| t->frame_thread.pass != 2 ? top_filter_2d : ------------------ | Branch (1699:30): [True: 0, False: 21.2k] ------------------ 1700| 21.2k| f->frame_thread.b[((t->by - 1) * f->b4_stride) + t->bx].filter2d); 1701| 21.2k| if (res) return res; ------------------ | Branch (1701:25): [True: 0, False: 21.2k] ------------------ 1702| 21.2k| } 1703| 10.6k| v_off = 2 * PXSTRIDE(f->cur.stride[1]); 1704| 10.6k| } 1705| 50.6k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1705:30): [True: 33.7k, False: 16.8k] ------------------ 1706| 33.7k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff + h_off + v_off, NULL, f->cur.stride[1], 1707| 33.7k| bw4, bh4, t->bx, t->by, 1 + pl, b->mv[0], 1708| 33.7k| refp, b->ref[0], filter_2d); 1709| 33.7k| if (res) return res; ------------------ | Branch (1709:21): [True: 0, False: 33.7k] ------------------ 1710| 33.7k| } 1711| 213k| } else { 1712| 213k| if (imin(cbw4, cbh4) > 1 && ------------------ | Branch (1712:17): [True: 128k, False: 84.2k] ------------------ 1713| 128k| ((b->inter_mode == GLOBALMV && f->gmv_warp_allowed[b->ref[0]]) || ------------------ | Branch (1713:19): [True: 84.6k, False: 44.1k] | Branch (1713:48): [True: 11.2k, False: 73.4k] ------------------ 1714| 117k| (b->motion_mode == MM_WARP && t->warpmv.type > DAV1D_WM_TYPE_TRANSLATION))) ------------------ | Branch (1714:19): [True: 14.2k, False: 103k] | Branch (1714:48): [True: 13.7k, False: 522] ------------------ 1715| 24.9k| { 1716| 74.8k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1716:34): [True: 49.9k, False: 24.9k] ------------------ 1717| 49.9k| res = warp_affine(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, NULL, 1718| 49.9k| f->cur.stride[1], b_dim, 1 + pl, refp, 1719| 49.9k| b->motion_mode == MM_WARP ? &t->warpmv : ------------------ | Branch (1719:39): [True: 27.5k, False: 22.4k] ------------------ 1720| 49.9k| &f->frame_hdr->gmv[b->ref[0]]); 1721| 49.9k| if (res) return res; ------------------ | Branch (1721:25): [True: 0, False: 49.9k] ------------------ 1722| 49.9k| } 1723| 188k| } else { 1724| 564k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1724:34): [True: 376k, False: 188k] ------------------ 1725| 376k| res = mc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, 1726| 376k| NULL, f->cur.stride[1], 1727| 376k| bw4 << (bw4 == ss_hor), bh4 << (bh4 == ss_ver), 1728| 376k| t->bx & ~ss_hor, t->by & ~ss_ver, 1729| 376k| 1 + pl, b->mv[0], refp, b->ref[0], filter_2d); 1730| 376k| if (res) return res; ------------------ | Branch (1730:25): [True: 0, False: 376k] ------------------ 1731| 376k| if (b->motion_mode == MM_OBMC) { ------------------ | Branch (1731:25): [True: 68.3k, False: 308k] ------------------ 1732| 68.3k| res = obmc(t, ((pixel *) f->cur.data[1 + pl]) + uvdstoff, 1733| 68.3k| f->cur.stride[1], b_dim, 1 + pl, bx4, by4, w4, h4); 1734| 68.3k| if (res) return res; ------------------ | Branch (1734:29): [True: 0, False: 68.3k] ------------------ 1735| 68.3k| } 1736| 376k| } 1737| 188k| } 1738| 213k| if (b->interintra_type) { ------------------ | Branch (1738:17): [True: 9.78k, False: 203k] ------------------ 1739| | // FIXME for 8x32 with 4:2:2 subsampling, this probably does 1740| | // the wrong thing since it will select 4x16, not 4x32, as a 1741| | // transform size... 1742| 9.78k| const uint8_t *const ii_mask = II_MASK(chr_layout_idx, bs, b); ------------------ | | 83| 9.78k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 84| 9.78k| (size_t)((b)->interintra_type == INTER_INTRA_BLEND ? \ | | ------------------ | | | Branch (84:14): [True: 7.47k, False: 2.31k] | | ------------------ | | 85| 9.78k| dav1d_masks.offsets[c][(bs)-BS_32x32].ii[(b)->interintra_mode] : \ | | 86| 9.78k| dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[0][(b)->wedge_idx]) * 8)) ------------------ 1743| | 1744| 29.3k| for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1744:34): [True: 19.5k, False: 9.78k] ------------------ 1745| 19.5k| pixel *const tmp = bitfn(t->scratch.interintra); ------------------ | | 77| 19.5k|#define bitfn(x) x##_16bpc ------------------ 1746| 19.5k| pixel *const tl_edge = bitfn(t->scratch.edge) + 32; ------------------ | | 77| 19.5k|#define bitfn(x) x##_16bpc ------------------ 1747| 19.5k| enum IntraPredMode m = 1748| 19.5k| b->interintra_mode == II_SMOOTH_PRED ? ------------------ | Branch (1748:25): [True: 3.42k, False: 16.1k] ------------------ 1749| 16.1k| SMOOTH_PRED : b->interintra_mode; 1750| 19.5k| int angle = 0; 1751| 19.5k| pixel *const uvdst = ((pixel *) f->cur.data[1 + pl]) + uvdstoff; 1752| 19.5k| const pixel *top_sb_edge = NULL; 1753| 19.5k| if (!(t->by & (f->sb_step - 1))) { ------------------ | Branch (1753:25): [True: 2.53k, False: 17.0k] ------------------ 1754| 2.53k| top_sb_edge = f->ipred_edge[pl + 1]; 1755| 2.53k| const int sby = t->by >> f->sb_shift; 1756| 2.53k| top_sb_edge += f->sb128w * 128 * (sby - 1); 1757| 2.53k| } 1758| 19.5k| m = bytefn(dav1d_prepare_intra_edges)(t->bx >> ss_hor, ------------------ | | 87| 19.5k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 19.5k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 1759| 19.5k| (t->bx >> ss_hor) > 1760| 19.5k| (ts->tiling.col_start >> ss_hor), 1761| 19.5k| t->by >> ss_ver, 1762| 19.5k| (t->by >> ss_ver) > 1763| 19.5k| (ts->tiling.row_start >> ss_ver), 1764| 19.5k| ts->tiling.col_end >> ss_hor, 1765| 19.5k| ts->tiling.row_end >> ss_ver, 1766| 19.5k| 0, uvdst, f->cur.stride[1], 1767| 19.5k| top_sb_edge, m, 1768| 19.5k| &angle, cbw4, cbh4, 0, tl_edge 1769| 19.5k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 19.5k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1770| 19.5k| dsp->ipred.intra_pred[m](tmp, cbw4 * 4 * sizeof(pixel), 1771| 19.5k| tl_edge, cbw4 * 4, cbh4 * 4, 0, 0, 0 1772| 19.5k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 19.5k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1773| 19.5k| dsp->mc.blend(uvdst, f->cur.stride[1], tmp, 1774| 19.5k| cbw4 * 4, cbh4 * 4, ii_mask); 1775| 19.5k| } 1776| 9.78k| } 1777| 213k| } 1778| | 1779| 981k| skip_inter_chroma_pred: {} 1780| 981k| t->tl_4x4_filter = filter_2d; 1781| 981k| } else { 1782| 51.8k| const enum Filter2d filter_2d = b->filter2d; 1783| | // Maximum super block size is 128x128 1784| 51.8k| int16_t (*tmp)[128 * 128] = t->scratch.compinter; 1785| 51.8k| int jnt_weight; 1786| 51.8k| uint8_t *const seg_mask = t->scratch.seg_mask; 1787| 51.8k| const uint8_t *mask; 1788| | 1789| 154k| for (int i = 0; i < 2; i++) { ------------------ | Branch (1789:25): [True: 103k, False: 51.8k] ------------------ 1790| 103k| const Dav1dThreadPicture *const refp = &f->refp[b->ref[i]]; 1791| | 1792| 103k| if (b->inter_mode == GLOBALMV_GLOBALMV && f->gmv_warp_allowed[b->ref[i]]) { ------------------ | Branch (1792:17): [True: 18.3k, False: 84.6k] | Branch (1792:55): [True: 7.60k, False: 10.7k] ------------------ 1793| 7.60k| res = warp_affine(t, NULL, tmp[i], bw4 * 4, b_dim, 0, refp, 1794| 7.60k| &f->frame_hdr->gmv[b->ref[i]]); 1795| 7.60k| if (res) return res; ------------------ | Branch (1795:21): [True: 0, False: 7.60k] ------------------ 1796| 95.4k| } else { 1797| 95.4k| res = mc(t, NULL, tmp[i], 0, bw4, bh4, t->bx, t->by, 0, 1798| 95.4k| b->mv[i], refp, b->ref[i], filter_2d); 1799| 95.4k| if (res) return res; ------------------ | Branch (1799:21): [True: 0, False: 95.4k] ------------------ 1800| 95.4k| } 1801| 103k| } 1802| 51.8k| switch (b->comp_type) { ------------------ | Branch (1802:17): [True: 51.5k, False: 320] ------------------ 1803| 30.0k| case COMP_INTER_AVG: ------------------ | Branch (1803:9): [True: 30.0k, False: 21.7k] ------------------ 1804| 30.0k| dsp->mc.avg(dst, f->cur.stride[0], tmp[0], tmp[1], 1805| 30.0k| bw4 * 4, bh4 * 4 HIGHBD_CALL_SUFFIX); ------------------ | | 73| 30.0k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1806| 30.0k| break; 1807| 8.59k| case COMP_INTER_WEIGHTED_AVG: ------------------ | Branch (1807:9): [True: 8.59k, False: 43.2k] ------------------ 1808| 8.59k| jnt_weight = f->jnt_weights[b->ref[0]][b->ref[1]]; 1809| 8.59k| dsp->mc.w_avg(dst, f->cur.stride[0], tmp[0], tmp[1], 1810| 8.59k| bw4 * 4, bh4 * 4, jnt_weight HIGHBD_CALL_SUFFIX); ------------------ | | 73| 8.59k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1811| 8.59k| break; 1812| 9.30k| case COMP_INTER_SEG: ------------------ | Branch (1812:9): [True: 9.30k, False: 42.5k] ------------------ 1813| 9.30k| dsp->mc.w_mask[chr_layout_idx](dst, f->cur.stride[0], 1814| 9.30k| tmp[b->mask_sign], tmp[!b->mask_sign], 1815| 9.30k| bw4 * 4, bh4 * 4, seg_mask, 1816| 9.30k| b->mask_sign HIGHBD_CALL_SUFFIX); ------------------ | | 73| 9.30k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1817| 9.30k| mask = seg_mask; 1818| 9.30k| break; 1819| 3.56k| case COMP_INTER_WEDGE: ------------------ | Branch (1819:9): [True: 3.56k, False: 48.2k] ------------------ 1820| 3.56k| mask = WEDGE_MASK(0, bs, 0, b->wedge_idx); ------------------ | | 89| 3.56k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 90| 3.56k| (size_t)dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[sign][idx] * 8)) ------------------ 1821| 3.56k| dsp->mc.mask(dst, f->cur.stride[0], 1822| 3.56k| tmp[b->mask_sign], tmp[!b->mask_sign], 1823| 3.56k| bw4 * 4, bh4 * 4, mask HIGHBD_CALL_SUFFIX); ------------------ | | 73| 3.56k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1824| 3.56k| if (has_chroma) ------------------ | Branch (1824:17): [True: 1.94k, False: 1.61k] ------------------ 1825| 1.94k| mask = WEDGE_MASK(chr_layout_idx, bs, b->mask_sign, b->wedge_idx); ------------------ | | 89| 1.94k| ((const uint8_t*)((uintptr_t)&dav1d_masks + \ | | 90| 1.94k| (size_t)dav1d_masks.offsets[c][(bs)-BS_32x32].wedge[sign][idx] * 8)) ------------------ 1826| 3.56k| break; 1827| 51.8k| } 1828| | 1829| | // chroma 1830| 58.2k| if (has_chroma) for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1830:13): [True: 19.4k, False: 32.0k] | Branch (1830:42): [True: 38.8k, False: 19.4k] ------------------ 1831| 116k| for (int i = 0; i < 2; i++) { ------------------ | Branch (1831:29): [True: 77.6k, False: 38.8k] ------------------ 1832| 77.6k| const Dav1dThreadPicture *const refp = &f->refp[b->ref[i]]; 1833| 77.6k| if (b->inter_mode == GLOBALMV_GLOBALMV && ------------------ | Branch (1833:21): [True: 13.8k, False: 63.8k] ------------------ 1834| 13.8k| imin(cbw4, cbh4) > 1 && f->gmv_warp_allowed[b->ref[i]]) ------------------ | Branch (1834:21): [True: 7.12k, False: 6.74k] | Branch (1834:45): [True: 3.70k, False: 3.41k] ------------------ 1835| 3.70k| { 1836| 3.70k| res = warp_affine(t, NULL, tmp[i], bw4 * 4 >> ss_hor, 1837| 3.70k| b_dim, 1 + pl, 1838| 3.70k| refp, &f->frame_hdr->gmv[b->ref[i]]); 1839| 3.70k| if (res) return res; ------------------ | Branch (1839:25): [True: 0, False: 3.70k] ------------------ 1840| 73.9k| } else { 1841| 73.9k| res = mc(t, NULL, tmp[i], 0, bw4, bh4, t->bx, t->by, 1842| 73.9k| 1 + pl, b->mv[i], refp, b->ref[i], filter_2d); 1843| 73.9k| if (res) return res; ------------------ | Branch (1843:25): [True: 0, False: 73.9k] ------------------ 1844| 73.9k| } 1845| 77.6k| } 1846| 38.8k| pixel *const uvdst = ((pixel *) f->cur.data[1 + pl]) + uvdstoff; 1847| 38.8k| switch (b->comp_type) { ------------------ | Branch (1847:21): [True: 38.8k, False: 18.4E] ------------------ 1848| 21.0k| case COMP_INTER_AVG: ------------------ | Branch (1848:13): [True: 21.0k, False: 17.8k] ------------------ 1849| 21.0k| dsp->mc.avg(uvdst, f->cur.stride[1], tmp[0], tmp[1], 1850| 21.0k| bw4 * 4 >> ss_hor, bh4 * 4 >> ss_ver 1851| 21.0k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 21.0k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1852| 21.0k| break; 1853| 3.11k| case COMP_INTER_WEIGHTED_AVG: ------------------ | Branch (1853:13): [True: 3.11k, False: 35.7k] ------------------ 1854| 3.11k| dsp->mc.w_avg(uvdst, f->cur.stride[1], tmp[0], tmp[1], 1855| 3.11k| bw4 * 4 >> ss_hor, bh4 * 4 >> ss_ver, jnt_weight 1856| 3.11k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 3.11k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1857| 3.11k| break; 1858| 3.89k| case COMP_INTER_WEDGE: ------------------ | Branch (1858:13): [True: 3.89k, False: 34.9k] ------------------ 1859| 14.7k| case COMP_INTER_SEG: ------------------ | Branch (1859:13): [True: 10.8k, False: 27.9k] ------------------ 1860| 14.7k| dsp->mc.mask(uvdst, f->cur.stride[1], 1861| 14.7k| tmp[b->mask_sign], tmp[!b->mask_sign], 1862| 14.7k| bw4 * 4 >> ss_hor, bh4 * 4 >> ss_ver, mask 1863| 14.7k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 14.7k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1864| 14.7k| break; 1865| 38.8k| } 1866| 38.8k| } 1867| 51.4k| } 1868| | 1869| 1.04M| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 34| 1.04M|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 1.04M] | | ------------------ | | 35| 1.04M| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 1.04M| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) { ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1870| 0| hex_dump(dst, f->cur.stride[0], b_dim[0] * 4, b_dim[1] * 4, "y-pred"); 1871| 0| if (has_chroma) { ------------------ | Branch (1871:13): [True: 0, False: 0] ------------------ 1872| 0| hex_dump(&((pixel *) f->cur.data[1])[uvdstoff], f->cur.stride[1], 1873| 0| cbw4 * 4, cbh4 * 4, "u-pred"); 1874| 0| hex_dump(&((pixel *) f->cur.data[2])[uvdstoff], f->cur.stride[1], 1875| 0| cbw4 * 4, cbh4 * 4, "v-pred"); 1876| 0| } 1877| 0| } 1878| | 1879| 1.04M| const int cw4 = (w4 + ss_hor) >> ss_hor, ch4 = (h4 + ss_ver) >> ss_ver; 1880| | 1881| 1.04M| if (b->skip) { ------------------ | Branch (1881:9): [True: 876k, False: 163k] ------------------ 1882| | // reset coef contexts 1883| 876k| BlockContext *const a = t->a; 1884| 876k| dav1d_memset_pow2[b_dim[2]](&a->lcoef[bx4], 0x40); 1885| 876k| dav1d_memset_pow2[b_dim[3]](&t->l.lcoef[by4], 0x40); 1886| 876k| if (has_chroma) { ------------------ | Branch (1886:13): [True: 160k, False: 715k] ------------------ 1887| 160k| dav1d_memset_pow2_fn memset_cw = dav1d_memset_pow2[ulog2(cbw4)]; 1888| 160k| dav1d_memset_pow2_fn memset_ch = dav1d_memset_pow2[ulog2(cbh4)]; 1889| 160k| memset_cw(&a->ccoef[0][cbx4], 0x40); 1890| 160k| memset_cw(&a->ccoef[1][cbx4], 0x40); 1891| 160k| memset_ch(&t->l.ccoef[0][cby4], 0x40); 1892| 160k| memset_ch(&t->l.ccoef[1][cby4], 0x40); 1893| 160k| } 1894| 876k| return 0; 1895| 876k| } 1896| | 1897| 163k| const TxfmInfo *const uvtx = &dav1d_txfm_dimensions[b->uvtx]; 1898| 163k| const TxfmInfo *const ytx = &dav1d_txfm_dimensions[b->max_ytx]; 1899| 163k| const uint16_t tx_split[2] = { b->tx_split0, b->tx_split1 }; 1900| | 1901| 330k| for (int init_y = 0; init_y < bh4; init_y += 16) { ------------------ | Branch (1901:26): [True: 167k, False: 163k] ------------------ 1902| 340k| for (int init_x = 0; init_x < bw4; init_x += 16) { ------------------ | Branch (1902:30): [True: 173k, False: 167k] ------------------ 1903| | // coefficient coding & inverse transforms 1904| 173k| int y_off = !!init_y, y; 1905| 173k| dst += PXSTRIDE(f->cur.stride[0]) * 4 * init_y; 1906| 353k| for (y = init_y, t->by += init_y; y < imin(h4, init_y + 16); ------------------ | Branch (1906:47): [True: 179k, False: 173k] ------------------ 1907| 179k| y += ytx->h, y_off++) 1908| 179k| { 1909| 179k| int x, x_off = !!init_x; 1910| 369k| for (x = init_x, t->bx += init_x; x < imin(w4, init_x + 16); ------------------ | Branch (1910:51): [True: 189k, False: 179k] ------------------ 1911| 189k| x += ytx->w, x_off++) 1912| 189k| { 1913| 189k| read_coef_tree(t, bs, b, b->max_ytx, 0, tx_split, 1914| 189k| x_off, y_off, &dst[x * 4]); 1915| 189k| t->bx += ytx->w; 1916| 189k| } 1917| 179k| dst += PXSTRIDE(f->cur.stride[0]) * 4 * ytx->h; 1918| 179k| t->bx -= x; 1919| 179k| t->by += ytx->h; 1920| 179k| } 1921| 173k| dst -= PXSTRIDE(f->cur.stride[0]) * 4 * y; 1922| 173k| t->by -= y; 1923| | 1924| | // chroma coefs and inverse transform 1925| 299k| if (has_chroma) for (int pl = 0; pl < 2; pl++) { ------------------ | Branch (1925:17): [True: 99.7k, False: 73.7k] | Branch (1925:46): [True: 199k, False: 99.7k] ------------------ 1926| 199k| pixel *uvdst = ((pixel *) f->cur.data[1 + pl]) + uvdstoff + 1927| 199k| (PXSTRIDE(f->cur.stride[1]) * init_y * 4 >> ss_ver); 1928| 199k| for (y = init_y >> ss_ver, t->by += init_y; 1929| 414k| y < imin(ch4, (init_y + 16) >> ss_ver); y += uvtx->h) ------------------ | Branch (1929:22): [True: 215k, False: 199k] ------------------ 1930| 215k| { 1931| 215k| int x; 1932| 215k| for (x = init_x >> ss_hor, t->bx += init_x; 1933| 451k| x < imin(cw4, (init_x + 16) >> ss_hor); x += uvtx->w) ------------------ | Branch (1933:26): [True: 236k, False: 215k] ------------------ 1934| 236k| { 1935| 236k| coef *cf; 1936| 236k| int eob; 1937| 236k| enum TxfmType txtp; 1938| 236k| if (t->frame_thread.pass) { ------------------ | Branch (1938:29): [True: 236k, False: 18.4E] ------------------ 1939| 236k| const int p = t->frame_thread.pass & 1; 1940| 236k| const int cbi = *ts->frame_thread[p].cbi++; 1941| 236k| cf = ts->frame_thread[p].cf; 1942| 236k| ts->frame_thread[p].cf += uvtx->w * uvtx->h * 16; 1943| 236k| eob = cbi >> 5; 1944| 236k| txtp = cbi & 0x1f; 1945| 18.4E| } else { 1946| 18.4E| uint8_t cf_ctx; 1947| 18.4E| cf = bitfn(t->cf); ------------------ | | 77| 18.4E|#define bitfn(x) x##_16bpc ------------------ 1948| 18.4E| txtp = t->scratch.txtp_map[(by4 + (y << ss_ver)) * 32 + 1949| 18.4E| bx4 + (x << ss_hor)]; 1950| 18.4E| eob = decode_coefs(t, &t->a->ccoef[pl][cbx4 + x], 1951| 18.4E| &t->l.ccoef[pl][cby4 + y], 1952| 18.4E| b->uvtx, bs, b, 0, 1 + pl, 1953| 18.4E| cf, &txtp, &cf_ctx); 1954| 18.4E| if (DEBUG_BLOCK_INFO) ------------------ | | 34| 18.4E|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 18.4E] | | ------------------ | | 35| 18.4E| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 18.4E| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 1955| 0| printf("Post-uv-cf-blk[pl=%d,tx=%d," 1956| 0| "txtp=%d,eob=%d]: r=%d\n", 1957| 0| pl, b->uvtx, txtp, eob, ts->msac.rng); 1958| 18.4E| int ctw = imin(uvtx->w, (f->bw - t->bx + ss_hor) >> ss_hor); 1959| 18.4E| int cth = imin(uvtx->h, (f->bh - t->by + ss_ver) >> ss_ver); 1960| 18.4E| dav1d_memset_likely_pow2(&t->a->ccoef[pl][cbx4 + x], cf_ctx, ctw); 1961| 18.4E| dav1d_memset_likely_pow2(&t->l.ccoef[pl][cby4 + y], cf_ctx, cth); 1962| 18.4E| } 1963| 236k| if (eob >= 0) { ------------------ | Branch (1963:29): [True: 85.5k, False: 151k] ------------------ 1964| 85.5k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 85.5k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 85.5k] | | ------------------ | | 35| 85.5k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 85.5k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1965| 0| coef_dump(cf, uvtx->h * 4, uvtx->w * 4, 3, "dq"); 1966| 85.5k| dsp->itx.itxfm_add[b->uvtx] 1967| 85.5k| [txtp](&uvdst[4 * x], 1968| 85.5k| f->cur.stride[1], 1969| 85.5k| cf, eob HIGHBD_CALL_SUFFIX); ------------------ | | 73| 85.5k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 1970| 85.5k| if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 34| 85.5k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 85.5k] | | ------------------ | | 35| 85.5k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 85.5k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ if (DEBUG_BLOCK_INFO && DEBUG_B_PIXELS) ------------------ | | 37| 0|#define DEBUG_B_PIXELS 0 | | ------------------ | | | Branch (37:24): [Folded, False: 0] | | ------------------ ------------------ 1971| 0| hex_dump(&uvdst[4 * x], f->cur.stride[1], 1972| 0| uvtx->w * 4, uvtx->h * 4, "recon"); 1973| 85.5k| } 1974| 236k| t->bx += uvtx->w << ss_hor; 1975| 236k| } 1976| 215k| uvdst += PXSTRIDE(f->cur.stride[1]) * 4 * uvtx->h; 1977| 215k| t->bx -= x << ss_hor; 1978| 215k| t->by += uvtx->h << ss_ver; 1979| 215k| } 1980| 199k| t->by -= y << ss_ver; 1981| 199k| } 1982| 173k| } 1983| 167k| } 1984| 163k| return 0; 1985| 1.04M|} dav1d_filter_sbrow_deblock_cols_16bpc: 1987| 702k|void bytefn(dav1d_filter_sbrow_deblock_cols)(Dav1dFrameContext *const f, const int sby) { 1988| 702k| if (!(f->c->inloop_filters & DAV1D_INLOOPFILTER_DEBLOCK) || ------------------ | Branch (1988:9): [True: 8, False: 702k] ------------------ 1989| 702k| (!f->frame_hdr->loopfilter.level_y[0] && !f->frame_hdr->loopfilter.level_y[1])) ------------------ | Branch (1989:10): [True: 8.49k, False: 693k] | Branch (1989:50): [True: 0, False: 8.49k] ------------------ 1990| 0| { 1991| 0| return; 1992| 0| } 1993| 702k| const int y = sby * f->sb_step * 4; 1994| 702k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 1995| 702k| pixel *const p[3] = { 1996| 702k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), 1997| 702k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), 1998| 702k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) 1999| 702k| }; 2000| 702k| Av1Filter *mask = f->lf.mask + (sby >> !f->seq_hdr->sb128) * f->sb128w; 2001| 702k| bytefn(dav1d_loopfilter_sbrow_cols)(f, p, mask, sby, ------------------ | | 87| 702k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 702k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2002| 702k| f->lf.start_of_tile_row[sby]); 2003| 702k|} dav1d_filter_sbrow_deblock_rows_16bpc: 2005| 932k|void bytefn(dav1d_filter_sbrow_deblock_rows)(Dav1dFrameContext *const f, const int sby) { 2006| 932k| const int y = sby * f->sb_step * 4; 2007| 932k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2008| 932k| pixel *const p[3] = { 2009| 932k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), 2010| 932k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), 2011| 932k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) 2012| 932k| }; 2013| 932k| Av1Filter *mask = f->lf.mask + (sby >> !f->seq_hdr->sb128) * f->sb128w; 2014| 932k| if (f->c->inloop_filters & DAV1D_INLOOPFILTER_DEBLOCK && ------------------ | Branch (2014:9): [True: 931k, False: 574] ------------------ 2015| 931k| (f->frame_hdr->loopfilter.level_y[0] || f->frame_hdr->loopfilter.level_y[1])) ------------------ | Branch (2015:10): [True: 693k, False: 238k] | Branch (2015:49): [True: 8.48k, False: 230k] ------------------ 2016| 702k| { 2017| 702k| bytefn(dav1d_loopfilter_sbrow_rows)(f, p, mask, sby); ------------------ | | 87| 702k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 702k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2018| 702k| } 2019| 932k| if (f->seq_hdr->cdef || f->lf.restore_planes) { ------------------ | Branch (2019:9): [True: 859k, False: 72.7k] | Branch (2019:29): [True: 11.5k, False: 61.2k] ------------------ 2020| | // Store loop filtered pixels required by CDEF / LR 2021| 871k| bytefn(dav1d_copy_lpf)(f, p, sby); ------------------ | | 87| 871k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 871k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2022| 871k| } 2023| 932k|} dav1d_filter_sbrow_cdef_16bpc: 2025| 859k|void bytefn(dav1d_filter_sbrow_cdef)(Dav1dTaskContext *const tc, const int sby) { 2026| 859k| const Dav1dFrameContext *const f = tc->f; 2027| 859k| if (!(f->c->inloop_filters & DAV1D_INLOOPFILTER_CDEF)) return; ------------------ | Branch (2027:9): [True: 0, False: 859k] ------------------ 2028| 859k| const int sbsz = f->sb_step; 2029| 859k| const int y = sby * sbsz * 4; 2030| 859k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2031| 859k| pixel *const p[3] = { 2032| 859k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), 2033| 859k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), 2034| 859k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) 2035| 859k| }; 2036| 859k| Av1Filter *prev_mask = f->lf.mask + ((sby - 1) >> !f->seq_hdr->sb128) * f->sb128w; 2037| 859k| Av1Filter *mask = f->lf.mask + (sby >> !f->seq_hdr->sb128) * f->sb128w; 2038| 859k| const int start = sby * sbsz; 2039| 859k| if (sby) { ------------------ | Branch (2039:9): [True: 714k, False: 144k] ------------------ 2040| 714k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2041| 714k| pixel *p_up[3] = { 2042| 714k| p[0] - 8 * PXSTRIDE(f->cur.stride[0]), 2043| 714k| p[1] - (8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver), 2044| 714k| p[2] - (8 * PXSTRIDE(f->cur.stride[1]) >> ss_ver), 2045| 714k| }; 2046| 714k| bytefn(dav1d_cdef_brow)(tc, p_up, prev_mask, start - 2, start, 1, sby); ------------------ | | 87| 714k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 714k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2047| 714k| } 2048| 859k| const int n_blks = sbsz - 2 * (sby + 1 < f->sbh); 2049| 859k| const int end = imin(start + n_blks, f->bh); 2050| 859k| bytefn(dav1d_cdef_brow)(tc, p, mask, start, end, 0, sby); ------------------ | | 87| 859k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 859k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2051| 859k|} dav1d_filter_sbrow_resize_16bpc: 2053| 29.8k|void bytefn(dav1d_filter_sbrow_resize)(Dav1dFrameContext *const f, const int sby) { 2054| 29.8k| const int sbsz = f->sb_step; 2055| 29.8k| const int y = sby * sbsz * 4; 2056| 29.8k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2057| 29.8k| const pixel *const p[3] = { 2058| 29.8k| f->lf.p[0] + y * PXSTRIDE(f->cur.stride[0]), 2059| 29.8k| f->lf.p[1] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver), 2060| 29.8k| f->lf.p[2] + (y * PXSTRIDE(f->cur.stride[1]) >> ss_ver) 2061| 29.8k| }; 2062| 29.8k| pixel *const sr_p[3] = { 2063| 29.8k| f->lf.sr_p[0] + y * PXSTRIDE(f->sr_cur.p.stride[0]), 2064| 29.8k| f->lf.sr_p[1] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver), 2065| 29.8k| f->lf.sr_p[2] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver) 2066| 29.8k| }; 2067| 29.8k| const int has_chroma = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400; 2068| 78.8k| for (int pl = 0; pl < 1 + 2 * has_chroma; pl++) { ------------------ | Branch (2068:22): [True: 49.0k, False: 29.8k] ------------------ 2069| 49.0k| const int ss_ver = pl && f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; ------------------ | Branch (2069:28): [True: 19.2k, False: 29.8k] | Branch (2069:34): [True: 9.58k, False: 9.61k] ------------------ 2070| 49.0k| const int h_start = 8 * !!sby >> ss_ver; 2071| 49.0k| const ptrdiff_t dst_stride = f->sr_cur.p.stride[!!pl]; 2072| 49.0k| pixel *dst = sr_p[pl] - h_start * PXSTRIDE(dst_stride); 2073| 49.0k| const ptrdiff_t src_stride = f->cur.stride[!!pl]; 2074| 49.0k| const pixel *src = p[pl] - h_start * PXSTRIDE(src_stride); 2075| 49.0k| const int h_end = 4 * (sbsz - 2 * (sby + 1 < f->sbh)) >> ss_ver; 2076| 49.0k| const int ss_hor = pl && f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; ------------------ | Branch (2076:28): [True: 19.2k, False: 29.8k] | Branch (2076:34): [True: 10.5k, False: 8.61k] ------------------ 2077| 49.0k| const int dst_w = (f->sr_cur.p.p.w + ss_hor) >> ss_hor; 2078| 49.0k| const int src_w = (4 * f->bw + ss_hor) >> ss_hor; 2079| 49.0k| const int img_h = (f->cur.p.h - sbsz * 4 * sby + ss_ver) >> ss_ver; 2080| | 2081| 49.0k| f->dsp->mc.resize(dst, dst_stride, src, src_stride, dst_w, 2082| 49.0k| imin(img_h, h_end) + h_start, src_w, 2083| 49.0k| f->resize_step[!!pl], f->resize_start[!!pl] 2084| 49.0k| HIGHBD_CALL_SUFFIX); ------------------ | | 73| 49.0k|#define HIGHBD_CALL_SUFFIX , f->bitdepth_max ------------------ 2085| 49.0k| } 2086| 29.8k|} dav1d_filter_sbrow_lr_16bpc: 2088| 42.4k|void bytefn(dav1d_filter_sbrow_lr)(Dav1dFrameContext *const f, const int sby) { 2089| 42.4k| if (!(f->c->inloop_filters & DAV1D_INLOOPFILTER_RESTORATION)) return; ------------------ | Branch (2089:9): [True: 0, False: 42.4k] ------------------ 2090| 42.4k| const int y = sby * f->sb_step * 4; 2091| 42.4k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2092| 42.4k| pixel *const sr_p[3] = { 2093| 42.4k| f->lf.sr_p[0] + y * PXSTRIDE(f->sr_cur.p.stride[0]), 2094| 42.4k| f->lf.sr_p[1] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver), 2095| 42.4k| f->lf.sr_p[2] + (y * PXSTRIDE(f->sr_cur.p.stride[1]) >> ss_ver) 2096| 42.4k| }; 2097| 42.4k| bytefn(dav1d_lr_sbrow)(f, sr_p, sby); ------------------ | | 87| 42.4k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 42.4k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2098| 42.4k|} dav1d_backup_ipred_edge_16bpc: 2111| 1.00M|void bytefn(dav1d_backup_ipred_edge)(Dav1dTaskContext *const t) { 2112| 1.00M| const Dav1dFrameContext *const f = t->f; 2113| 1.00M| Dav1dTileState *const ts = t->ts; 2114| 1.00M| const int sby = t->by >> f->sb_shift; 2115| 1.00M| const int sby_off = f->sb128w * 128 * sby; 2116| 1.00M| const int x_off = ts->tiling.col_start; 2117| | 2118| 1.00M| const pixel *const y = 2119| 1.00M| ((const pixel *) f->cur.data[0]) + x_off * 4 + 2120| 1.00M| ((t->by + f->sb_step) * 4 - 1) * PXSTRIDE(f->cur.stride[0]); 2121| 1.00M| pixel_copy(&f->ipred_edge[0][sby_off + x_off * 4], y, ------------------ | | 65| 1.00M|#define pixel_copy(a, b, c) memcpy(a, b, (c) << 1) ------------------ 2122| 1.00M| 4 * (ts->tiling.col_end - x_off)); 2123| | 2124| 1.00M| if (f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I400) { ------------------ | Branch (2124:9): [True: 258k, False: 742k] ------------------ 2125| 258k| const int ss_ver = f->cur.p.layout == DAV1D_PIXEL_LAYOUT_I420; 2126| 258k| const int ss_hor = f->cur.p.layout != DAV1D_PIXEL_LAYOUT_I444; 2127| | 2128| 258k| const ptrdiff_t uv_off = (x_off * 4 >> ss_hor) + 2129| 258k| (((t->by + f->sb_step) * 4 >> ss_ver) - 1) * PXSTRIDE(f->cur.stride[1]); 2130| 776k| for (int pl = 1; pl <= 2; pl++) ------------------ | Branch (2130:26): [True: 517k, False: 258k] ------------------ 2131| 517k| pixel_copy(&f->ipred_edge[pl][sby_off + (x_off * 4 >> ss_hor)], ------------------ | | 65| 517k|#define pixel_copy(a, b, c) memcpy(a, b, (c) << 1) ------------------ 2132| 258k| &((const pixel *) f->cur.data[pl])[uv_off], 2133| 258k| 4 * (ts->tiling.col_end - x_off) >> ss_hor); 2134| 258k| } 2135| 1.00M|} dav1d_copy_pal_block_y_16bpc: 2141| 53.5k|{ 2142| 53.5k| const Dav1dFrameContext *const f = t->f; 2143| 53.5k| pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (2143:24): [True: 53.5k, False: 1] ------------------ 2144| 53.5k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2145| 53.5k| ((t->bx >> 1) + (t->by & 1))][0] : 2146| 53.5k| bytefn(t->scratch.pal)[0]; ------------------ | | 87| 1|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 53.5k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2147| 265k| for (int x = 0; x < bw4; x++) ------------------ | Branch (2147:21): [True: 212k, False: 53.5k] ------------------ 2148| 212k| memcpy(bytefn(t->al_pal)[0][bx4 + x][0], pal, 8 * sizeof(pixel)); ------------------ | | 87| 212k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 212k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2149| 193k| for (int y = 0; y < bh4; y++) ------------------ | Branch (2149:21): [True: 139k, False: 53.5k] ------------------ 2150| 139k| memcpy(bytefn(t->al_pal)[1][by4 + y][0], pal, 8 * sizeof(pixel)); ------------------ | | 87| 139k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 139k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2151| 53.5k|} dav1d_copy_pal_block_uv_16bpc: 2157| 13.8k|{ 2158| 13.8k| const Dav1dFrameContext *const f = t->f; 2159| 13.8k| const pixel (*const pal)[8] = t->frame_thread.pass ? ------------------ | Branch (2159:35): [True: 13.8k, False: 0] ------------------ 2160| 13.8k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2161| 13.8k| ((t->bx >> 1) + (t->by & 1))] : 2162| 13.8k| bytefn(t->scratch.pal); ------------------ | | 87| 0|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 0|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2163| | // see aomedia bug 2183 for why we use luma coordinates here 2164| 41.5k| for (int pl = 1; pl <= 2; pl++) { ------------------ | Branch (2164:22): [True: 27.7k, False: 13.8k] ------------------ 2165| 147k| for (int x = 0; x < bw4; x++) ------------------ | Branch (2165:25): [True: 119k, False: 27.7k] ------------------ 2166| 119k| memcpy(bytefn(t->al_pal)[0][bx4 + x][pl], pal[pl], 8 * sizeof(pixel)); ------------------ | | 87| 119k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 119k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2167| 129k| for (int y = 0; y < bh4; y++) ------------------ | Branch (2167:25): [True: 101k, False: 27.7k] ------------------ 2168| 101k| memcpy(bytefn(t->al_pal)[1][by4 + y][pl], pal[pl], 8 * sizeof(pixel)); ------------------ | | 87| 101k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 101k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2169| 27.7k| } 2170| 13.8k|} dav1d_read_pal_plane_16bpc: 2175| 67.3k|{ 2176| 67.3k| Dav1dTileState *const ts = t->ts; 2177| 67.3k| const Dav1dFrameContext *const f = t->f; 2178| 67.3k| const int pal_sz = b->pal_sz[pl] = dav1d_msac_decode_symbol_adapt8(&ts->msac, ------------------ | | 48| 67.3k|#define dav1d_msac_decode_symbol_adapt8 dav1d_msac_decode_symbol_adapt8_sse2 ------------------ 2179| 67.3k| ts->cdf.m.pal_sz[pl][sz_ctx], 6) + 2; 2180| 67.3k| pixel cache[16], used_cache[8]; 2181| 67.3k| int l_cache = pl ? t->pal_sz_uv[1][by4] : t->l.pal_sz[by4]; ------------------ | Branch (2181:19): [True: 13.8k, False: 53.5k] ------------------ 2182| 67.3k| int n_cache = 0; 2183| | // don't reuse above palette outside SB64 boundaries 2184| 67.3k| int a_cache = by4 & 15 ? pl ? t->pal_sz_uv[0][bx4] : t->a->pal_sz[bx4] : 0; ------------------ | Branch (2184:19): [True: 59.9k, False: 7.43k] | Branch (2184:30): [True: 11.7k, False: 48.1k] ------------------ 2185| 67.3k| const pixel *l = bytefn(t->al_pal)[1][by4][pl]; ------------------ | | 87| 67.3k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 67.3k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2186| 67.3k| const pixel *a = bytefn(t->al_pal)[0][bx4][pl]; ------------------ | | 87| 67.3k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 67.3k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2187| | 2188| | // fill/sort cache 2189| 135k| while (l_cache && a_cache) { ------------------ | Branch (2189:12): [True: 88.6k, False: 47.0k] | Branch (2189:23): [True: 68.3k, False: 20.3k] ------------------ 2190| 68.3k| if (*l < *a) { ------------------ | Branch (2190:13): [True: 32.6k, False: 35.6k] ------------------ 2191| 32.6k| if (!n_cache || cache[n_cache - 1] != *l) ------------------ | Branch (2191:17): [True: 6.43k, False: 26.2k] | Branch (2191:29): [True: 25.3k, False: 929] ------------------ 2192| 31.7k| cache[n_cache++] = *l; 2193| 32.6k| l++; 2194| 32.6k| l_cache--; 2195| 35.6k| } else { 2196| 35.6k| if (*a == *l) { ------------------ | Branch (2196:17): [True: 12.2k, False: 23.4k] ------------------ 2197| 12.2k| l++; 2198| 12.2k| l_cache--; 2199| 12.2k| } 2200| 35.6k| if (!n_cache || cache[n_cache - 1] != *a) ------------------ | Branch (2200:17): [True: 4.99k, False: 30.6k] | Branch (2200:29): [True: 30.1k, False: 430] ------------------ 2201| 35.1k| cache[n_cache++] = *a; 2202| 35.6k| a++; 2203| 35.6k| a_cache--; 2204| 35.6k| } 2205| 68.3k| } 2206| 67.3k| if (l_cache) { ------------------ | Branch (2206:9): [True: 20.3k, False: 46.9k] ------------------ 2207| 85.6k| do { 2208| 85.6k| if (!n_cache || cache[n_cache - 1] != *l) ------------------ | Branch (2208:17): [True: 16.5k, False: 69.0k] | Branch (2208:29): [True: 56.4k, False: 12.6k] ------------------ 2209| 72.9k| cache[n_cache++] = *l; 2210| 85.6k| l++; 2211| 85.6k| } while (--l_cache > 0); ------------------ | Branch (2211:18): [True: 65.2k, False: 20.3k] ------------------ 2212| 46.9k| } else if (a_cache) { ------------------ | Branch (2212:16): [True: 19.1k, False: 27.8k] ------------------ 2213| 86.7k| do { 2214| 86.7k| if (!n_cache || cache[n_cache - 1] != *a) ------------------ | Branch (2214:17): [True: 12.8k, False: 73.9k] | Branch (2214:29): [True: 62.3k, False: 11.5k] ------------------ 2215| 75.1k| cache[n_cache++] = *a; 2216| 86.7k| a++; 2217| 86.7k| } while (--a_cache > 0); ------------------ | Branch (2217:18): [True: 67.6k, False: 19.1k] ------------------ 2218| 19.1k| } 2219| | 2220| | // find reused cache entries 2221| 67.3k| int i = 0; 2222| 255k| for (int n = 0; n < n_cache && i < pal_sz; n++) ------------------ | Branch (2222:21): [True: 197k, False: 58.3k] | Branch (2222:36): [True: 188k, False: 9.03k] ------------------ 2223| 188k| if (dav1d_msac_decode_bool_equi(&ts->msac)) ------------------ | | 53| 188k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (2223:13): [True: 86.1k, False: 102k] ------------------ 2224| 86.1k| used_cache[i++] = cache[n]; 2225| 67.3k| const int n_used_cache = i; 2226| | 2227| | // parse new entries 2228| 67.3k| pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (2228:24): [True: 67.3k, False: 18.4E] ------------------ 2229| 67.3k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2230| 67.3k| ((t->bx >> 1) + (t->by & 1))][pl] : 2231| 18.4E| bytefn(t->scratch.pal)[pl]; ------------------ | | 87| 18.4E|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 67.3k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2232| 67.3k| if (i < pal_sz) { ------------------ | Branch (2232:9): [True: 56.8k, False: 10.5k] ------------------ 2233| 56.8k| const int bpc = BITDEPTH == 8 ? 8 : f->cur.p.bpc; ------------------ | Branch (2233:25): [Folded, False: 56.8k] ------------------ 2234| 56.8k| int prev = pal[i++] = dav1d_msac_decode_bools(&ts->msac, bpc); 2235| | 2236| 56.8k| if (i < pal_sz) { ------------------ | Branch (2236:13): [True: 49.8k, False: 6.98k] ------------------ 2237| 49.8k| int bits = bpc - 3 + dav1d_msac_decode_bools(&ts->msac, 2); 2238| 49.8k| const int max = (1 << bpc) - 1; 2239| | 2240| 126k| do { 2241| 126k| const int delta = dav1d_msac_decode_bools(&ts->msac, bits); 2242| 126k| prev = pal[i++] = imin(prev + delta + !pl, max); 2243| 126k| if (prev + !pl >= max) { ------------------ | Branch (2243:21): [True: 22.1k, False: 103k] ------------------ 2244| 57.0k| for (; i < pal_sz; i++) ------------------ | Branch (2244:28): [True: 34.9k, False: 22.1k] ------------------ 2245| 34.9k| pal[i] = max; 2246| 22.1k| break; 2247| 22.1k| } 2248| 103k| bits = imin(bits, 1 + ulog2(max - prev - !pl)); 2249| 103k| } while (i < pal_sz); ------------------ | Branch (2249:22): [True: 76.2k, False: 27.7k] ------------------ 2250| 49.8k| } 2251| | 2252| | // merge cache+new entries 2253| 56.8k| int n = 0, m = n_used_cache; 2254| 330k| for (i = 0; i < pal_sz; i++) { ------------------ | Branch (2254:21): [True: 273k, False: 56.8k] ------------------ 2255| 273k| if (n < n_used_cache && (m >= pal_sz || used_cache[n] <= pal[m])) { ------------------ | Branch (2255:17): [True: 92.5k, False: 181k] | Branch (2255:38): [True: 20.2k, False: 72.3k] | Branch (2255:53): [True: 35.5k, False: 36.7k] ------------------ 2256| 55.8k| pal[i] = used_cache[n++]; 2257| 217k| } else { 2258| 217k| assert(m < pal_sz); ------------------ | Branch (2258:17): [True: 217k, False: 18.4E] ------------------ 2259| 217k| pal[i] = pal[m++]; 2260| 217k| } 2261| 273k| } 2262| 56.8k| } else { 2263| 10.5k| memcpy(pal, used_cache, n_used_cache * sizeof(*used_cache)); 2264| 10.5k| } 2265| | 2266| 67.4k| if (DEBUG_BLOCK_INFO) { ------------------ | | 34| 67.4k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 67.4k] | | ------------------ | | 35| 67.4k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 67.4k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2267| 0| printf("Post-pal[pl=%d,sz=%d,cache_size=%d,used_cache=%d]: r=%d, cache=", 2268| 0| pl, pal_sz, n_cache, n_used_cache, ts->msac.rng); 2269| 0| for (int n = 0; n < n_cache; n++) ------------------ | Branch (2269:25): [True: 0, False: 0] ------------------ 2270| 0| printf("%c%02x", n ? ' ' : '[', cache[n]); ------------------ | Branch (2270:30): [True: 0, False: 0] ------------------ 2271| 0| printf("%s, pal=", n_cache ? "]" : "[]"); ------------------ | Branch (2271:28): [True: 0, False: 0] ------------------ 2272| 0| for (int n = 0; n < pal_sz; n++) ------------------ | Branch (2272:25): [True: 0, False: 0] ------------------ 2273| 0| printf("%c%02x", n ? ' ' : '[', pal[n]); ------------------ | Branch (2273:30): [True: 0, False: 0] ------------------ 2274| 0| printf("]\n"); 2275| 0| } 2276| 67.4k|} dav1d_read_pal_uv_16bpc: 2280| 13.8k|{ 2281| 13.8k| bytefn(dav1d_read_pal_plane)(t, b, 1, sz_ctx, bx4, by4); ------------------ | | 87| 13.8k|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 13.8k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2282| | 2283| | // V pal coding 2284| 13.8k| Dav1dTileState *const ts = t->ts; 2285| 13.8k| const Dav1dFrameContext *const f = t->f; 2286| 13.8k| pixel *const pal = t->frame_thread.pass ? ------------------ | Branch (2286:24): [True: 13.8k, False: 18.4E] ------------------ 2287| 13.8k| f->frame_thread.pal[((t->by >> 1) + (t->bx & 1)) * (f->b4_stride >> 1) + 2288| 13.8k| ((t->bx >> 1) + (t->by & 1))][2] : 2289| 18.4E| bytefn(t->scratch.pal)[2]; ------------------ | | 87| 18.4E|#define bytefn(x) bitfn(x) | | ------------------ | | | | 77| 13.8k|#define bitfn(x) x##_16bpc | | ------------------ ------------------ 2290| 13.8k| const int bpc = BITDEPTH == 8 ? 8 : f->cur.p.bpc; ------------------ | Branch (2290:21): [Folded, False: 13.8k] ------------------ 2291| 13.8k| if (dav1d_msac_decode_bool_equi(&ts->msac)) { ------------------ | | 53| 13.8k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (2291:9): [True: 7.18k, False: 6.68k] ------------------ 2292| 7.18k| const int bits = bpc - 4 + dav1d_msac_decode_bools(&ts->msac, 2); 2293| 7.18k| int prev = pal[0] = dav1d_msac_decode_bools(&ts->msac, bpc); 2294| 7.18k| const int max = (1 << bpc) - 1; 2295| 28.2k| for (int i = 1; i < b->pal_sz[1]; i++) { ------------------ | Branch (2295:25): [True: 21.0k, False: 7.18k] ------------------ 2296| 21.0k| int delta = dav1d_msac_decode_bools(&ts->msac, bits); 2297| 21.0k| if (delta && dav1d_msac_decode_bool_equi(&ts->msac)) delta = -delta; ------------------ | | 53| 20.5k|#define dav1d_msac_decode_bool_equi dav1d_msac_decode_bool_equi_sse2 ------------------ | Branch (2297:17): [True: 20.5k, False: 509] | Branch (2297:26): [True: 10.4k, False: 10.0k] ------------------ 2298| 21.0k| prev = pal[i] = (prev + delta) & max; 2299| 21.0k| } 2300| 7.18k| } else { 2301| 30.5k| for (int i = 0; i < b->pal_sz[1]; i++) ------------------ | Branch (2301:25): [True: 23.8k, False: 6.68k] ------------------ 2302| 23.8k| pal[i] = dav1d_msac_decode_bools(&ts->msac, bpc); 2303| 6.68k| } 2304| 13.8k| if (DEBUG_BLOCK_INFO) { ------------------ | | 34| 13.8k|#define DEBUG_BLOCK_INFO 0 && \ | | ------------------ | | | Branch (34:26): [Folded, False: 13.8k] | | ------------------ | | 35| 13.8k| f->frame_hdr->frame_offset == 2 && t->by >= 0 && t->by < 4 && \ | | ------------------ | | | Branch (35:9): [True: 0, False: 0] | | | Branch (35:44): [True: 0, False: 0] | | | Branch (35:58): [True: 0, False: 0] | | ------------------ | | 36| 13.8k| t->bx >= 8 && t->bx < 12 | | ------------------ | | | Branch (36:9): [True: 0, False: 0] | | | Branch (36:23): [True: 0, False: 0] | | ------------------ ------------------ 2305| 0| printf("Post-pal[pl=2]: r=%d ", ts->msac.rng); 2306| 0| for (int n = 0; n < b->pal_sz[1]; n++) ------------------ | Branch (2306:25): [True: 0, False: 0] ------------------ 2307| 0| printf("%c%02x", n ? ' ' : '[', pal[n]); ------------------ | Branch (2307:30): [True: 0, False: 0] ------------------ 2308| 0| printf("]\n"); 2309| 0| } 2310| 13.8k|} dav1d_ref_create: 37| 412k|Dav1dRef *dav1d_ref_create(const enum AllocationType type, size_t size) { 38| 412k| size = (size + sizeof(void*) - 1) & ~(sizeof(void*) - 1); 39| | 40| 412k| uint8_t *const data = dav1d_alloc_aligned(type, size + sizeof(Dav1dRef), 64); ------------------ | | 134| 412k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 41| 412k| if (!data) return NULL; ------------------ | Branch (41:9): [True: 0, False: 412k] ------------------ 42| | 43| 412k| Dav1dRef *const res = (Dav1dRef*)(data + size); 44| 412k| res->const_data = res->user_data = res->data = data; 45| 412k| atomic_init(&res->ref_cnt, 1); 46| 412k| res->free_ref = 0; 47| 412k| res->free_callback = default_free_callback; 48| | 49| 412k| return res; 50| 412k|} dav1d_ref_create_using_pool: 56| 789k|Dav1dRef *dav1d_ref_create_using_pool(Dav1dMemPool *const pool, size_t size) { 57| 789k| void *const buf = dav1d_mem_pool_pop(pool, size); 58| 789k| if (!buf) return NULL; ------------------ | Branch (58:9): [True: 0, False: 789k] ------------------ 59| | 60| | /* Store Dav1dRef inside the Dav1dMemPoolBuffer alignment padding */ 61| 789k| assert(sizeof(Dav1dMemPoolBuffer) + sizeof(Dav1dRef) <= 64); ------------------ | Branch (61:5): [True: 789k, Folded] ------------------ 62| 789k| Dav1dRef *const res = &((Dav1dRef*)buf)[-1]; 63| 789k| res->data = buf; 64| 789k| res->const_data = pool; 65| 789k| atomic_init(&res->ref_cnt, 1); 66| 789k| res->free_ref = 0; 67| 789k| res->free_callback = pool_free_callback; 68| 789k| res->user_data = buf; 69| | 70| 789k| return res; 71| 789k|} dav1d_ref_dec: 73| 50.7M|void dav1d_ref_dec(Dav1dRef **const pref) { 74| 50.7M| assert(pref != NULL); ------------------ | Branch (74:5): [True: 50.7M, False: 18.4E] ------------------ 75| | 76| 50.7M| Dav1dRef *const ref = *pref; 77| 50.7M| if (!ref) return; ------------------ | Branch (77:9): [True: 33.5M, False: 17.1M] ------------------ 78| | 79| 17.1M| *pref = NULL; 80| 17.1M| if (atomic_fetch_sub(&ref->ref_cnt, 1) == 1) { ------------------ | Branch (80:9): [True: 1.60M, False: 15.5M] ------------------ 81| 1.60M| const int free_ref = ref->free_ref; 82| 1.60M| ref->free_callback(ref->const_data, ref->user_data); 83| 1.60M| if (free_ref) dav1d_free(ref); ------------------ | | 135| 0|#define dav1d_free(ptr) free(ptr) ------------------ | Branch (83:13): [True: 0, False: 1.60M] ------------------ 84| 1.60M| } 85| 17.1M|} ref.c:default_free_callback: 32| 412k|static void default_free_callback(const uint8_t *const data, void *const user_data) { 33| 412k| assert(data == user_data); ------------------ | Branch (33:5): [True: 412k, False: 0] ------------------ 34| 412k| dav1d_free_aligned(user_data); ------------------ | | 136| 412k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 35| 412k|} ref.c:pool_free_callback: 52| 789k|static void pool_free_callback(const uint8_t *const data, void *const user_data) { 53| 789k| dav1d_mem_pool_push((Dav1dMemPool*)data, user_data); 54| 789k|} obu.c:dav1d_ref_is_writable: 73| 417k|static inline int dav1d_ref_is_writable(Dav1dRef *const ref) { 74| 417k| return atomic_load(&ref->ref_cnt) == 1 && ref->data; ------------------ | Branch (74:12): [True: 417k, False: 0] | Branch (74:47): [True: 417k, False: 0] ------------------ 75| 417k|} obu.c:dav1d_ref_init: 59| 972|{ 60| 972| ref->data = NULL; 61| 972| ref->const_data = ptr; 62| | atomic_init(&ref->ref_cnt, 1); 63| 972| ref->free_ref = free_ref; 64| 972| ref->free_callback = free_callback; 65| 972| ref->user_data = user_data; 66| 972| return ref; 67| 972|} obu.c:dav1d_ref_inc: 69| 9.17k|static inline void dav1d_ref_inc(Dav1dRef *const ref) { 70| | atomic_fetch_add_explicit(&ref->ref_cnt, 1, memory_order_relaxed); 71| 9.17k|} picture.c:dav1d_ref_inc: 69| 13.0M|static inline void dav1d_ref_inc(Dav1dRef *const ref) { 70| | atomic_fetch_add_explicit(&ref->ref_cnt, 1, memory_order_relaxed); 71| 13.0M|} picture.c:dav1d_ref_init: 59| 405k|{ 60| 405k| ref->data = NULL; 61| 405k| ref->const_data = ptr; 62| | atomic_init(&ref->ref_cnt, 1); 63| 405k| ref->free_ref = free_ref; 64| 405k| ref->free_callback = free_callback; 65| 405k| ref->user_data = user_data; 66| 405k| return ref; 67| 405k|} cdf.c:dav1d_ref_inc: 69| 463k|static inline void dav1d_ref_inc(Dav1dRef *const ref) { 70| | atomic_fetch_add_explicit(&ref->ref_cnt, 1, memory_order_relaxed); 71| 463k|} data.c:dav1d_ref_inc: 69| 771k|static inline void dav1d_ref_inc(Dav1dRef *const ref) { 70| | atomic_fetch_add_explicit(&ref->ref_cnt, 1, memory_order_relaxed); 71| 771k|} decode.c:dav1d_ref_inc: 69| 1.25M|static inline void dav1d_ref_inc(Dav1dRef *const ref) { 70| | atomic_fetch_add_explicit(&ref->ref_cnt, 1, memory_order_relaxed); 71| 1.25M|} dav1d_refmvs_find: 354| 5.27M|{ 355| 5.27M| const refmvs_frame *const rf = rt->rf; 356| 5.27M| const uint8_t *const b_dim = dav1d_block_dimensions[bs]; 357| 5.27M| const int bw4 = b_dim[0], w4 = imin(imin(bw4, 16), rt->tile_col.end - bx4); 358| 5.27M| const int bh4 = b_dim[1], h4 = imin(imin(bh4, 16), rt->tile_row.end - by4); 359| 5.27M| mv gmv[2], tgmv[2]; 360| | 361| 5.27M| *cnt = 0; 362| 5.27M| assert(ref.ref[0] >= 0 && ref.ref[0] <= 8 && ------------------ | Branch (362:5): [True: 5.27M, False: 18.4E] | Branch (362:5): [True: 5.27M, False: 338] | Branch (362:5): [True: 5.27M, False: 18.4E] | Branch (362:5): [True: 5.27M, False: 18.4E] ------------------ 363| 5.27M| ref.ref[1] >= -1 && ref.ref[1] <= 8); 364| 5.27M| if (ref.ref[0] > 0) { ------------------ | Branch (364:9): [True: 3.93M, False: 1.33M] ------------------ 365| 3.93M| tgmv[0] = get_gmv_2d(&rf->frm_hdr->gmv[ref.ref[0] - 1], 366| 3.93M| bx4, by4, bw4, bh4, rf->frm_hdr); 367| 3.93M| gmv[0] = rf->frm_hdr->gmv[ref.ref[0] - 1].type > DAV1D_WM_TYPE_TRANSLATION ? ------------------ | Branch (367:18): [True: 1.29M, False: 2.64M] ------------------ 368| 2.64M| tgmv[0] : (mv) { .n = INVALID_MV }; ------------------ | | 40| 2.64M|#define INVALID_MV 0x80008000 ------------------ 369| 3.93M| } else { 370| 1.33M| tgmv[0] = (mv) { .n = 0 }; 371| 1.33M| gmv[0] = (mv) { .n = INVALID_MV }; ------------------ | | 40| 1.33M|#define INVALID_MV 0x80008000 ------------------ 372| 1.33M| } 373| 5.27M| if (ref.ref[1] > 0) { ------------------ | Branch (373:9): [True: 397k, False: 4.87M] ------------------ 374| 397k| tgmv[1] = get_gmv_2d(&rf->frm_hdr->gmv[ref.ref[1] - 1], 375| 397k| bx4, by4, bw4, bh4, rf->frm_hdr); 376| 397k| gmv[1] = rf->frm_hdr->gmv[ref.ref[1] - 1].type > DAV1D_WM_TYPE_TRANSLATION ? ------------------ | Branch (376:18): [True: 88.9k, False: 308k] ------------------ 377| 308k| tgmv[1] : (mv) { .n = INVALID_MV }; ------------------ | | 40| 308k|#define INVALID_MV 0x80008000 ------------------ 378| 397k| } 379| | 380| | // top 381| 5.27M| int have_newmv = 0, have_col_mvs = 0, have_row_mvs = 0; 382| 5.27M| unsigned max_rows = 0, n_rows = ~0; 383| 5.27M| const refmvs_block *b_top; 384| 5.27M| if (by4 > rt->tile_row.start) { ------------------ | Branch (384:9): [True: 4.67M, False: 602k] ------------------ 385| 4.67M| max_rows = imin((by4 - rt->tile_row.start + 1) >> 1, 2 + (bh4 > 1)); 386| 4.67M| b_top = &rt->r[(by4 & 31) - 1 + 5][bx4]; 387| 4.67M| n_rows = scan_row(mvstack, cnt, ref, gmv, b_top, 388| 4.67M| bw4, w4, max_rows, bw4 >= 16 ? 4 : 1, ------------------ | Branch (388:46): [True: 1.89M, False: 2.77M] ------------------ 389| 4.67M| &have_newmv, &have_row_mvs); 390| 4.67M| } 391| | 392| | // left 393| 5.27M| unsigned max_cols = 0, n_cols = ~0U; 394| 5.27M| refmvs_block *const *b_left; 395| 5.27M| if (bx4 > rt->tile_col.start) { ------------------ | Branch (395:9): [True: 3.68M, False: 1.59M] ------------------ 396| 3.68M| max_cols = imin((bx4 - rt->tile_col.start + 1) >> 1, 2 + (bw4 > 1)); 397| 3.68M| b_left = &rt->r[(by4 & 31) + 5]; 398| 3.68M| n_cols = scan_col(mvstack, cnt, ref, gmv, b_left, 399| 3.68M| bh4, h4, bx4 - 1, max_cols, bh4 >= 16 ? 4 : 1, ------------------ | Branch (399:55): [True: 815k, False: 2.86M] ------------------ 400| 3.68M| &have_newmv, &have_col_mvs); 401| 3.68M| } 402| | 403| | // top/right 404| 5.27M| if (n_rows != ~0U && edge_flags & EDGE_I444_TOP_HAS_RIGHT && ------------------ | Branch (404:9): [True: 4.65M, False: 617k] | Branch (404:26): [True: 3.36M, False: 1.28M] ------------------ 405| 3.36M| imax(bw4, bh4) <= 16 && bw4 + bx4 < rt->tile_col.end) ------------------ | Branch (405:9): [True: 2.67M, False: 694k] | Branch (405:33): [True: 1.75M, False: 918k] ------------------ 406| 1.75M| { 407| 1.75M| add_spatial_candidate(mvstack, cnt, 4, &b_top[bw4], ref, gmv, 408| 1.75M| &have_newmv, &have_row_mvs); 409| 1.75M| } 410| | 411| 5.27M| const int nearest_match = have_col_mvs + have_row_mvs; 412| 5.27M| const int nearest_cnt = *cnt; 413| 11.4M| for (int n = 0; n < nearest_cnt; n++) ------------------ | Branch (413:21): [True: 6.17M, False: 5.27M] ------------------ 414| 6.17M| mvstack[n].weight += 640; 415| | 416| | // temporal 417| 5.27M| int globalmv_ctx = rf->frm_hdr->use_ref_frame_mvs; 418| 5.27M| if (rf->use_ref_frame_mvs) { ------------------ | Branch (418:9): [True: 805k, False: 4.46M] ------------------ 419| 805k| const ptrdiff_t stride = rf->rp_stride; 420| 805k| const int by8 = by4 >> 1, bx8 = bx4 >> 1; 421| 805k| const refmvs_temporal_block *const rbi = &rt->rp_proj[(by8 & 15) * stride + bx8]; 422| 805k| const refmvs_temporal_block *rb = rbi; 423| 805k| const int step_h = bw4 >= 16 ? 2 : 1, step_v = bh4 >= 16 ? 2 : 1; ------------------ | Branch (423:28): [True: 361k, False: 444k] | Branch (423:56): [True: 363k, False: 442k] ------------------ 424| 805k| const int w8 = imin((w4 + 1) >> 1, 8), h8 = imin((h4 + 1) >> 1, 8); 425| 3.08M| for (int y = 0; y < h8; y += step_v) { ------------------ | Branch (425:25): [True: 2.27M, False: 805k] ------------------ 426| 9.07M| for (int x = 0; x < w8; x+= step_h) { ------------------ | Branch (426:29): [True: 6.79M, False: 2.27M] ------------------ 427| 6.79M| add_temporal_candidate(rf, mvstack, cnt, &rb[x], ref, 428| 6.79M| !(x | y) ? &globalmv_ctx : NULL, tgmv); ------------------ | Branch (428:40): [True: 807k, False: 5.98M] ------------------ 429| 6.79M| } 430| 2.27M| rb += stride * step_v; 431| 2.27M| } 432| 805k| if (imin(bw4, bh4) >= 2 && imax(bw4, bh4) < 16) { ------------------ | Branch (432:13): [True: 724k, False: 81.8k] | Branch (432:36): [True: 355k, False: 368k] ------------------ 433| 355k| const int bh8 = bh4 >> 1, bw8 = bw4 >> 1; 434| 355k| rb = &rbi[bh8 * stride]; 435| 355k| const int has_bottom = by8 + bh8 < imin(rt->tile_row.end >> 1, 436| 355k| (by8 & ~7) + 8); 437| 355k| if (has_bottom && bx8 - 1 >= imax(rt->tile_col.start >> 1, bx8 & ~7)) { ------------------ | Branch (437:17): [True: 246k, False: 109k] | Branch (437:31): [True: 185k, False: 60.8k] ------------------ 438| 185k| add_temporal_candidate(rf, mvstack, cnt, &rb[-1], ref, 439| 185k| NULL, NULL); 440| 185k| } 441| 355k| if (bx8 + bw8 < imin(rt->tile_col.end >> 1, (bx8 & ~7) + 8)) { ------------------ | Branch (441:17): [True: 264k, False: 91.1k] ------------------ 442| 264k| if (has_bottom) { ------------------ | Branch (442:21): [True: 183k, False: 81.0k] ------------------ 443| 183k| add_temporal_candidate(rf, mvstack, cnt, &rb[bw8], ref, 444| 183k| NULL, NULL); 445| 183k| } 446| 264k| if (by8 + bh8 - 1 < imin(rt->tile_row.end >> 1, (by8 & ~7) + 8)) { ------------------ | Branch (446:21): [True: 264k, False: 260] ------------------ 447| 264k| add_temporal_candidate(rf, mvstack, cnt, &rb[bw8 - stride], 448| 264k| ref, NULL, NULL); 449| 264k| } 450| 264k| } 451| 355k| } 452| 805k| } 453| 5.27M| assert(*cnt <= 8); ------------------ | Branch (453:5): [True: 5.26M, False: 13.9k] ------------------ 454| | 455| | // top/left (which, confusingly, is part of "secondary" references) 456| 5.26M| int have_dummy_newmv_match; 457| 5.26M| if ((n_rows | n_cols) != ~0U) { ------------------ | Branch (457:9): [True: 3.10M, False: 2.16M] ------------------ 458| 3.10M| add_spatial_candidate(mvstack, cnt, 4, &b_top[-1], ref, gmv, 459| 3.10M| &have_dummy_newmv_match, &have_row_mvs); 460| 3.10M| } 461| | 462| | // "secondary" (non-direct neighbour) top & left edges 463| | // what is different about secondary is that everything is now in 8x8 resolution 464| 15.7M| for (int n = 2; n <= 3; n++) { ------------------ | Branch (464:21): [True: 10.4M, False: 5.26M] ------------------ 465| 10.4M| if ((unsigned) n > n_rows && (unsigned) n <= max_rows) { ------------------ | Branch (465:13): [True: 4.16M, False: 6.33M] | Branch (465:38): [True: 3.17M, False: 991k] ------------------ 466| 3.17M| n_rows += scan_row(mvstack, cnt, ref, gmv, 467| 3.17M| &rt->r[(((by4 & 31) - 2 * n + 1) | 1) + 5][bx4 | 1], 468| 3.17M| bw4, w4, 1 + max_rows - n, bw4 >= 16 ? 4 : 2, ------------------ | Branch (468:59): [True: 45.5k, False: 3.12M] ------------------ 469| 3.17M| &have_dummy_newmv_match, &have_row_mvs); 470| 3.17M| } 471| | 472| 10.4M| if ((unsigned) n > n_cols && (unsigned) n <= max_cols) { ------------------ | Branch (472:13): [True: 4.60M, False: 5.89M] | Branch (472:38): [True: 3.85M, False: 749k] ------------------ 473| 3.85M| n_cols += scan_col(mvstack, cnt, ref, gmv, &rt->r[((by4 & 31) | 1) + 5], 474| 3.85M| bh4, h4, (bx4 - n * 2 + 1) | 1, 475| 3.85M| 1 + max_cols - n, bh4 >= 16 ? 4 : 2, ------------------ | Branch (475:50): [True: 413k, False: 3.44M] ------------------ 476| 3.85M| &have_dummy_newmv_match, &have_col_mvs); 477| 3.85M| } 478| 10.4M| } 479| 5.26M| assert(*cnt <= 8); ------------------ | Branch (479:5): [True: 5.26M, False: 18.4E] ------------------ 480| | 481| 5.26M| const int ref_match_count = have_col_mvs + have_row_mvs; 482| | 483| | // context build-up 484| 5.26M| int refmv_ctx, newmv_ctx; 485| 5.26M| switch (nearest_match) { ------------------ | Branch (485:13): [True: 5.27M, False: 18.4E] ------------------ 486| 672k| case 0: ------------------ | Branch (486:5): [True: 672k, False: 4.59M] ------------------ 487| 672k| refmv_ctx = imin(2, ref_match_count); 488| 672k| newmv_ctx = ref_match_count > 0; 489| 672k| break; 490| 2.80M| case 1: ------------------ | Branch (490:5): [True: 2.80M, False: 2.45M] ------------------ 491| 2.80M| refmv_ctx = imin(ref_match_count * 3, 4); 492| 2.80M| newmv_ctx = 3 - have_newmv; 493| 2.80M| break; 494| 1.79M| case 2: ------------------ | Branch (494:5): [True: 1.79M, False: 3.46M] ------------------ 495| 1.79M| refmv_ctx = 5; 496| 1.79M| newmv_ctx = 5 - have_newmv; 497| 1.79M| break; 498| 5.26M| } 499| | 500| | // sorting (nearest, then "secondary") 501| 5.26M| int len = nearest_cnt; 502| 10.4M| while (len) { ------------------ | Branch (502:12): [True: 5.22M, False: 5.26M] ------------------ 503| 5.22M| int last = 0; 504| 7.04M| for (int n = 1; n < len; n++) { ------------------ | Branch (504:25): [True: 1.82M, False: 5.22M] ------------------ 505| 1.82M| if (mvstack[n - 1].weight < mvstack[n].weight) { ------------------ | Branch (505:17): [True: 783k, False: 1.03M] ------------------ 506| 783k|#define EXCHANGE(a, b) do { refmvs_candidate tmp = a; a = b; b = tmp; } while (0) 507| 783k| EXCHANGE(mvstack[n - 1], mvstack[n]); ------------------ | | 506| 783k|#define EXCHANGE(a, b) do { refmvs_candidate tmp = a; a = b; b = tmp; } while (0) | | ------------------ | | | Branch (506:80): [Folded, False: 783k] | | ------------------ ------------------ 508| 783k| last = n; 509| 783k| } 510| 1.82M| } 511| 5.22M| len = last; 512| 5.22M| } 513| 5.26M| len = *cnt; 514| 7.47M| while (len > nearest_cnt) { ------------------ | Branch (514:12): [True: 2.20M, False: 5.26M] ------------------ 515| 2.20M| int last = nearest_cnt; 516| 4.11M| for (int n = nearest_cnt + 1; n < len; n++) { ------------------ | Branch (516:39): [True: 1.90M, False: 2.20M] ------------------ 517| 1.90M| if (mvstack[n - 1].weight < mvstack[n].weight) { ------------------ | Branch (517:17): [True: 670k, False: 1.23M] ------------------ 518| 670k| EXCHANGE(mvstack[n - 1], mvstack[n]); ------------------ | | 506| 670k|#define EXCHANGE(a, b) do { refmvs_candidate tmp = a; a = b; b = tmp; } while (0) | | ------------------ | | | Branch (506:80): [Folded, False: 670k] | | ------------------ ------------------ 519| 670k|#undef EXCHANGE 520| 670k| last = n; 521| 670k| } 522| 1.90M| } 523| 2.20M| len = last; 524| 2.20M| } 525| | 526| 5.26M| if (ref.ref[1] > 0) { ------------------ | Branch (526:9): [True: 396k, False: 4.86M] ------------------ 527| 396k| if (*cnt < 2) { ------------------ | Branch (527:13): [True: 249k, False: 146k] ------------------ 528| 249k| const int sign0 = rf->sign_bias[ref.ref[0] - 1]; 529| 249k| const int sign1 = rf->sign_bias[ref.ref[1] - 1]; 530| 249k| const int sz4 = imin(w4, h4); 531| 249k| refmvs_candidate *const same = &mvstack[*cnt]; 532| 249k| int same_count[4] = { 0 }; 533| | 534| | // non-self references in top 535| 402k| if (n_rows != ~0U) for (int x = 0; x < sz4;) { ------------------ | Branch (535:17): [True: 194k, False: 55.4k] | Branch (535:48): [True: 207k, False: 194k] ------------------ 536| 207k| const refmvs_block *const cand_b = &b_top[x]; 537| 207k| add_compound_extended_candidate(same, same_count, cand_b, 538| 207k| sign0, sign1, ref, rf->sign_bias); 539| 207k| x += dav1d_block_dimensions[cand_b->bs][0]; 540| 207k| } 541| | 542| | // non-self references in left 543| 521k| if (n_cols != ~0U) for (int y = 0; y < sz4;) { ------------------ | Branch (543:17): [True: 237k, False: 12.1k] | Branch (543:48): [True: 284k, False: 237k] ------------------ 544| 284k| const refmvs_block *const cand_b = &b_left[y][bx4 - 1]; 545| 284k| add_compound_extended_candidate(same, same_count, cand_b, 546| 284k| sign0, sign1, ref, rf->sign_bias); 547| 284k| y += dav1d_block_dimensions[cand_b->bs][1]; 548| 284k| } 549| | 550| 249k| refmvs_candidate *const diff = &same[2]; 551| 249k| const int *const diff_count = &same_count[2]; 552| | 553| | // merge together 554| 750k| for (int n = 0; n < 2; n++) { ------------------ | Branch (554:29): [True: 500k, False: 249k] ------------------ 555| 500k| int m = same_count[n]; 556| | 557| 500k| if (m >= 2) continue; ------------------ | Branch (557:21): [True: 147k, False: 353k] ------------------ 558| | 559| 353k| const int l = diff_count[n]; 560| 353k| if (l) { ------------------ | Branch (560:21): [True: 322k, False: 30.5k] ------------------ 561| 322k| same[m].mv.mv[n] = diff[0].mv.mv[n]; 562| 322k| if (++m == 2) continue; ------------------ | Branch (562:25): [True: 211k, False: 111k] ------------------ 563| 111k| if (l == 2) { ------------------ | Branch (563:25): [True: 90.5k, False: 20.5k] ------------------ 564| 90.5k| same[1].mv.mv[n] = diff[1].mv.mv[n]; 565| 90.5k| continue; 566| 90.5k| } 567| 111k| } 568| 68.3k| do { 569| 68.3k| same[m].mv.mv[n] = tgmv[n]; 570| 68.3k| } while (++m < 2); ------------------ | Branch (570:26): [True: 17.2k, False: 51.1k] ------------------ 571| 51.1k| } 572| | 573| | // if the first extended was the same as the non-extended one, 574| | // then replace it with the second extended one 575| 249k| int n = *cnt; 576| 249k| if (n == 1 && mvstack[0].mv.n == same[0].mv.n) ------------------ | Branch (576:17): [True: 141k, False: 108k] | Branch (576:27): [True: 95.4k, False: 45.7k] ------------------ 577| 95.4k| mvstack[1].mv = mvstack[2].mv; 578| 359k| do { 579| 359k| mvstack[n].weight = 2; 580| 359k| } while (++n < 2); ------------------ | Branch (580:22): [True: 109k, False: 249k] ------------------ 581| 249k| *cnt = 2; 582| 249k| } 583| | 584| | // clamping 585| 396k| const int left = -(bx4 + bw4 + 4) * 4 * 8; 586| 396k| const int right = (rf->iw4 - bx4 + 4) * 4 * 8; 587| 396k| const int top = -(by4 + bh4 + 4) * 4 * 8; 588| 396k| const int bottom = (rf->ih4 - by4 + 4) * 4 * 8; 589| | 590| 396k| const int n_refmvs = *cnt; 591| 396k| int n = 0; 592| 950k| do { 593| 950k| mvstack[n].mv.mv[0].x = iclip(mvstack[n].mv.mv[0].x, left, right); 594| 950k| mvstack[n].mv.mv[0].y = iclip(mvstack[n].mv.mv[0].y, top, bottom); 595| 950k| mvstack[n].mv.mv[1].x = iclip(mvstack[n].mv.mv[1].x, left, right); 596| 950k| mvstack[n].mv.mv[1].y = iclip(mvstack[n].mv.mv[1].y, top, bottom); 597| 950k| } while (++n < n_refmvs); ------------------ | Branch (597:18): [True: 554k, False: 396k] ------------------ 598| | 599| 396k| switch (refmv_ctx >> 1) { ------------------ | Branch (599:17): [True: 398k, False: 18.4E] ------------------ 600| 144k| case 0: ------------------ | Branch (600:9): [True: 144k, False: 252k] ------------------ 601| 144k| *ctx = imin(newmv_ctx, 1); 602| 144k| break; 603| 148k| case 1: ------------------ | Branch (603:9): [True: 148k, False: 248k] ------------------ 604| 148k| *ctx = 1 + imin(newmv_ctx, 3); 605| 148k| break; 606| 105k| case 2: ------------------ | Branch (606:9): [True: 105k, False: 290k] ------------------ 607| 105k| *ctx = iclip(3 + newmv_ctx, 4, 7); 608| 105k| break; 609| 396k| } 610| | 611| 397k| return; 612| 4.86M| } else if (*cnt < 2 && ref.ref[0] > 0) { ------------------ | Branch (612:16): [True: 3.08M, False: 1.78M] | Branch (612:28): [True: 2.50M, False: 572k] ------------------ 613| 2.50M| const int sign = rf->sign_bias[ref.ref[0] - 1]; 614| 2.50M| const int sz4 = imin(w4, h4); 615| | 616| | // non-self references in top 617| 4.73M| if (n_rows != ~0U) for (int x = 0; x < sz4 && *cnt < 2;) { ------------------ | Branch (617:13): [True: 2.34M, False: 160k] | Branch (617:44): [True: 2.39M, False: 2.34M] | Branch (617:55): [True: 2.38M, False: 6.00k] ------------------ 618| 2.38M| const refmvs_block *const cand_b = &b_top[x]; 619| 2.38M| add_single_extended_candidate(mvstack, cnt, cand_b, sign, rf->sign_bias); 620| 2.38M| x += dav1d_block_dimensions[cand_b->bs][0]; 621| 2.38M| } 622| | 623| | // non-self references in left 624| 2.50M| if (n_cols != ~0U) for (int y = 0; y < sz4 && *cnt < 2;) { ------------------ | Branch (624:13): [True: 967k, False: 1.54M] | Branch (624:44): [True: 1.01M, False: 835k] | Branch (624:55): [True: 881k, False: 132k] ------------------ 625| 881k| const refmvs_block *const cand_b = &b_left[y][bx4 - 1]; 626| 881k| add_single_extended_candidate(mvstack, cnt, cand_b, sign, rf->sign_bias); 627| 881k| y += dav1d_block_dimensions[cand_b->bs][1]; 628| 881k| } 629| 2.50M| } 630| 5.26M| assert(*cnt <= 8); ------------------ | Branch (630:5): [True: 4.87M, False: 18.4E] ------------------ 631| | 632| | // clamping 633| 4.87M| int n_refmvs = *cnt; 634| 4.87M| if (n_refmvs) { ------------------ | Branch (634:9): [True: 4.58M, False: 292k] ------------------ 635| 4.58M| const int left = -(bx4 + bw4 + 4) * 4 * 8; 636| 4.58M| const int right = (rf->iw4 - bx4 + 4) * 4 * 8; 637| 4.58M| const int top = -(by4 + bh4 + 4) * 4 * 8; 638| 4.58M| const int bottom = (rf->ih4 - by4 + 4) * 4 * 8; 639| | 640| 4.58M| int n = 0; 641| 9.12M| do { 642| 9.12M| mvstack[n].mv.mv[0].x = iclip(mvstack[n].mv.mv[0].x, left, right); 643| 9.12M| mvstack[n].mv.mv[0].y = iclip(mvstack[n].mv.mv[0].y, top, bottom); 644| 9.12M| } while (++n < n_refmvs); ------------------ | Branch (644:18): [True: 4.54M, False: 4.58M] ------------------ 645| 4.58M| } 646| | 647| 7.96M| for (int n = *cnt; n < 2; n++) ------------------ | Branch (647:24): [True: 3.09M, False: 4.87M] ------------------ 648| 3.09M| mvstack[n].mv.mv[0] = tgmv[0]; 649| | 650| 4.87M| *ctx = (refmv_ctx << 4) | (globalmv_ctx << 3) | newmv_ctx; 651| 4.87M|} dav1d_refmvs_tile_sbrow_init: 657| 4.01M|{ 658| 4.01M| if (rf->n_tile_threads == 1) tile_row_idx = 0; ------------------ | Branch (658:9): [True: 0, False: 4.01M] ------------------ 659| 4.01M| rt->rp_proj = &rf->rp_proj[16 * rf->rp_stride * tile_row_idx]; 660| 4.01M| const ptrdiff_t r_stride = rf->rp_stride * 2; 661| 4.01M| const ptrdiff_t pass_off = (rf->n_frame_threads > 1 && pass == 2) ? ------------------ | Branch (661:33): [True: 4.01M, False: 44] | Branch (661:60): [True: 1.94M, False: 2.07M] ------------------ 662| 2.07M| 35 * 2 * rf->n_blocks : 0; 663| 4.01M| refmvs_block *r = &rf->r[35 * r_stride * tile_row_idx + pass_off]; 664| 4.01M| const int sbsz = rf->sbsz; 665| 4.01M| const int off = (sbsz * sby) & 16; 666| 100M| for (int i = 0; i < sbsz; i++, r += r_stride) ------------------ | Branch (666:21): [True: 96.8M, False: 4.01M] ------------------ 667| 96.8M| rt->r[off + 5 + i] = r; 668| 4.01M| rt->r[off + 0] = r; 669| 4.01M| r += r_stride; 670| 4.01M| rt->r[off + 1] = NULL; 671| 4.01M| rt->r[off + 2] = r; 672| 4.01M| r += r_stride; 673| 4.01M| rt->r[off + 3] = NULL; 674| 4.01M| rt->r[off + 4] = r; 675| 4.01M| if (sby & 1) { ------------------ | Branch (675:9): [True: 1.80M, False: 2.20M] ------------------ 676| 1.80M|#define EXCHANGE(a, b) do { void *const tmp = a; a = b; b = tmp; } while (0) 677| 1.80M| EXCHANGE(rt->r[off + 0], rt->r[off + sbsz + 0]); ------------------ | | 676| 1.80M|#define EXCHANGE(a, b) do { void *const tmp = a; a = b; b = tmp; } while (0) | | ------------------ | | | Branch (676:75): [Folded, False: 1.80M] | | ------------------ ------------------ 678| 1.80M| EXCHANGE(rt->r[off + 2], rt->r[off + sbsz + 2]); ------------------ | | 676| 1.80M|#define EXCHANGE(a, b) do { void *const tmp = a; a = b; b = tmp; } while (0) | | ------------------ | | | Branch (676:75): [Folded, False: 1.80M] | | ------------------ ------------------ 679| 1.80M| EXCHANGE(rt->r[off + 4], rt->r[off + sbsz + 4]); ------------------ | | 676| 1.80M|#define EXCHANGE(a, b) do { void *const tmp = a; a = b; b = tmp; } while (0) | | ------------------ | | | Branch (676:75): [Folded, False: 1.80M] | | ------------------ ------------------ 680| 1.80M|#undef EXCHANGE 681| 1.80M| } 682| | 683| 4.01M| rt->rf = rf; 684| 4.01M| rt->tile_row.start = tile_row_start4; 685| 4.01M| rt->tile_row.end = imin(tile_row_end4, rf->ih4); 686| 4.01M| rt->tile_col.start = tile_col_start4; 687| 4.01M| rt->tile_col.end = imin(tile_col_end4, rf->iw4); 688| 4.01M|} dav1d_refmvs_init_frame: 812| 325k|{ 813| 325k| const int rp_stride = ((frm_hdr->width[0] + 127) & ~127) >> 3; 814| 18.4E| const int n_tile_rows = n_tile_threads > 1 ? frm_hdr->tiling.rows : 1; ------------------ | Branch (814:29): [True: 325k, False: 18.4E] ------------------ 815| 325k| const int n_blocks = rp_stride * n_tile_rows; 816| | 817| 325k| rf->sbsz = 16 << seq_hdr->sb128; 818| 325k| rf->frm_hdr = frm_hdr; 819| 325k| rf->iw8 = (frm_hdr->width[0] + 7) >> 3; 820| 325k| rf->ih8 = (frm_hdr->height + 7) >> 3; 821| 325k| rf->iw4 = rf->iw8 << 1; 822| 325k| rf->ih4 = rf->ih8 << 1; 823| 325k| rf->rp = rp; 824| 325k| rf->rp_stride = rp_stride; 825| 325k| rf->n_tile_threads = n_tile_threads; 826| 325k| rf->n_frame_threads = n_frame_threads; 827| | 828| 325k| if (n_blocks != rf->n_blocks) { ------------------ | Branch (828:9): [True: 30.5k, False: 295k] ------------------ 829| 30.5k| const size_t r_sz = sizeof(*rf->r) * 35 * 2 * n_blocks * (1 + (n_frame_threads > 1)); 830| 30.5k| const size_t rp_proj_sz = sizeof(*rf->rp_proj) * 16 * n_blocks; 831| | /* Note that sizeof(*rf->r) == 12, but it's accessed using 16-byte unaligned 832| | * loads in save_tmvs() asm which can overread 4 bytes into rp_proj. */ 833| 30.5k| dav1d_free_aligned(rf->r); ------------------ | | 136| 30.5k|#define dav1d_free_aligned(ptr) dav1d_free_aligned_internal(ptr) ------------------ 834| 30.5k| rf->r = dav1d_alloc_aligned(ALLOC_REFMVS, r_sz + rp_proj_sz, 64); ------------------ | | 134| 30.5k|#define dav1d_alloc_aligned(type, sz, align) dav1d_alloc_aligned_internal(sz, align) ------------------ 835| 30.5k| if (!rf->r) { ------------------ | Branch (835:13): [True: 0, False: 30.5k] ------------------ 836| 0| rf->n_blocks = 0; 837| 0| return DAV1D_ERR(ENOMEM); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 838| 0| } 839| | 840| 30.5k| rf->rp_proj = (refmvs_temporal_block*)((uintptr_t)rf->r + r_sz); 841| 30.5k| rf->n_blocks = n_blocks; 842| 30.5k| } 843| | 844| 325k| const int poc = frm_hdr->frame_offset; 845| 2.60M| for (int i = 0; i < 7; i++) { ------------------ | Branch (845:21): [True: 2.27M, False: 325k] ------------------ 846| 2.27M| const int poc_diff = get_poc_diff(seq_hdr->order_hint_n_bits, 847| 2.27M| ref_poc[i], poc); 848| 2.27M| rf->sign_bias[i] = poc_diff > 0; 849| 2.27M| rf->mfmv_sign[i] = poc_diff < 0; 850| 2.27M| rf->pocdiff[i] = iclip(get_poc_diff(seq_hdr->order_hint_n_bits, 851| 2.27M| poc, ref_poc[i]), -31, 31); 852| 2.27M| } 853| | 854| | // temporal MV setup 855| 325k| rf->n_mfmvs = 0; 856| 325k| rf->rp_ref = rp_ref; 857| 325k| if (frm_hdr->use_ref_frame_mvs && seq_hdr->order_hint_n_bits) { ------------------ | Branch (857:9): [True: 75.4k, False: 250k] | Branch (857:39): [True: 75.4k, False: 0] ------------------ 858| 75.4k| int total = 2; 859| 75.4k| if (rp_ref[0] && ref_ref_poc[0][6] != ref_poc[3] /* alt-of-last != gold */) { ------------------ | Branch (859:13): [True: 63.0k, False: 12.4k] | Branch (859:26): [True: 43.2k, False: 19.7k] ------------------ 860| 43.2k| rf->mfmv_ref[rf->n_mfmvs++] = 0; // last 861| 43.2k| total = 3; 862| 43.2k| } 863| 75.4k| if (rp_ref[4] && get_poc_diff(seq_hdr->order_hint_n_bits, ref_poc[4], ------------------ | Branch (863:13): [True: 63.2k, False: 12.2k] | Branch (863:26): [True: 28.6k, False: 34.5k] ------------------ 864| 63.2k| frm_hdr->frame_offset) > 0) 865| 28.6k| { 866| 28.6k| rf->mfmv_ref[rf->n_mfmvs++] = 4; // bwd 867| 28.6k| } 868| 75.4k| if (rp_ref[5] && get_poc_diff(seq_hdr->order_hint_n_bits, ref_poc[5], ------------------ | Branch (868:13): [True: 60.1k, False: 15.3k] | Branch (868:26): [True: 26.8k, False: 33.3k] ------------------ 869| 60.1k| frm_hdr->frame_offset) > 0) 870| 26.8k| { 871| 26.8k| rf->mfmv_ref[rf->n_mfmvs++] = 5; // altref2 872| 26.8k| } 873| 75.4k| if (rf->n_mfmvs < total && rp_ref[6] && ------------------ | Branch (873:13): [True: 53.0k, False: 22.4k] | Branch (873:36): [True: 37.1k, False: 15.8k] ------------------ 874| 37.1k| get_poc_diff(seq_hdr->order_hint_n_bits, ref_poc[6], ------------------ | Branch (874:13): [True: 17.8k, False: 19.2k] ------------------ 875| 37.1k| frm_hdr->frame_offset) > 0) 876| 17.8k| { 877| 17.8k| rf->mfmv_ref[rf->n_mfmvs++] = 6; // altref 878| 17.8k| } 879| 75.4k| if (rf->n_mfmvs < total && rp_ref[1]) ------------------ | Branch (879:13): [True: 42.8k, False: 32.6k] | Branch (879:36): [True: 31.3k, False: 11.5k] ------------------ 880| 31.3k| rf->mfmv_ref[rf->n_mfmvs++] = 1; // last2 881| | 882| 223k| for (int n = 0; n < rf->n_mfmvs; n++) { ------------------ | Branch (882:25): [True: 147k, False: 75.4k] ------------------ 883| 147k| const int rpoc = ref_poc[rf->mfmv_ref[n]]; 884| 147k| const int diff1 = get_poc_diff(seq_hdr->order_hint_n_bits, 885| 147k| rpoc, frm_hdr->frame_offset); 886| 147k| if (abs(diff1) > 31) { ------------------ | Branch (886:17): [True: 695, False: 147k] ------------------ 887| 695| rf->mfmv_ref2cur[n] = INVALID_REF2CUR; ------------------ | | 41| 695|#define INVALID_REF2CUR (-32) ------------------ 888| 147k| } else { 889| 147k| rf->mfmv_ref2cur[n] = rf->mfmv_ref[n] < 4 ? -diff1 : diff1; ------------------ | Branch (889:39): [True: 74.1k, False: 73.0k] ------------------ 890| 1.17M| for (int m = 0; m < 7; m++) { ------------------ | Branch (890:33): [True: 1.02M, False: 147k] ------------------ 891| 1.02M| const int rrpoc = ref_ref_poc[rf->mfmv_ref[n]][m]; 892| 1.02M| const int diff2 = get_poc_diff(seq_hdr->order_hint_n_bits, 893| 1.02M| rpoc, rrpoc); 894| | // unsigned comparison also catches the < 0 case 895| 1.02M| rf->mfmv_ref2ref[n][m] = (unsigned) diff2 > 31U ? 0 : diff2; ------------------ | Branch (895:46): [True: 260k, False: 769k] ------------------ 896| 1.02M| } 897| 147k| } 898| 147k| } 899| 75.4k| } 900| 325k| rf->use_ref_frame_mvs = rf->n_mfmvs > 0; 901| | 902| 325k| return 0; 903| 325k|} dav1d_refmvs_dsp_init: 926| 9.41k|{ 927| 9.41k| c->load_tmvs = load_tmvs_c; 928| 9.41k| c->save_tmvs = save_tmvs_c; 929| 9.41k| c->splat_mv = splat_mv_c; 930| | 931| 9.41k|#if HAVE_ASM 932| |#if ARCH_AARCH64 || ARCH_ARM 933| | refmvs_dsp_init_arm(c); 934| |#elif ARCH_LOONGARCH64 935| | refmvs_dsp_init_loongarch(c); 936| |#elif ARCH_X86 937| | refmvs_dsp_init_x86(c); 938| 9.41k|#endif 939| 9.41k|#endif 940| 9.41k|} refmvs.c:scan_row: 102| 7.83M|{ 103| 7.83M| const refmvs_block *cand_b = b; 104| 7.83M| const enum BlockSize first_cand_bs = cand_b->bs; 105| 7.83M| const uint8_t *const first_cand_b_dim = dav1d_block_dimensions[first_cand_bs]; 106| 7.83M| int cand_bw4 = first_cand_b_dim[0]; 107| 7.83M| int len = imax(step, imin(bw4, cand_bw4)); 108| | 109| 7.83M| if (bw4 <= cand_bw4) { ------------------ | Branch (109:9): [True: 6.96M, False: 869k] ------------------ 110| | // FIXME weight can be higher for odd blocks (bx4 & 1), but then the 111| | // position of the first block has to be odd already, i.e. not just 112| | // for row_offset=-3/-5 113| | // FIXME why can this not be cand_bw4? 114| 6.96M| const int weight = bw4 == 1 ? 2 : ------------------ | Branch (114:28): [True: 1.53M, False: 5.43M] ------------------ 115| 6.96M| imax(2, imin(2 * max_rows, first_cand_b_dim[1])); 116| 6.96M| add_spatial_candidate(mvstack, cnt, len * weight, cand_b, ref, gmv, 117| 6.96M| have_newmv_match, have_refmv_match); 118| 6.96M| return weight >> 1; 119| 6.96M| } 120| | 121| 1.77M| for (int x = 0;;) { 122| | // FIXME if we overhang above, we could fill a bitmask so we don't have 123| | // to repeat the add_spatial_candidate() for the next row, but just increase 124| | // the weight here 125| 1.77M| add_spatial_candidate(mvstack, cnt, len * 2, cand_b, ref, gmv, 126| 1.77M| have_newmv_match, have_refmv_match); 127| 1.77M| x += len; 128| 1.77M| if (x >= w4) return 1; ------------------ | Branch (128:13): [True: 862k, False: 912k] ------------------ 129| 912k| cand_b = &b[x]; 130| 912k| cand_bw4 = dav1d_block_dimensions[cand_b->bs][0]; 131| 912k| assert(cand_bw4 < bw4); ------------------ | Branch (131:9): [True: 914k, False: 18.4E] ------------------ 132| 914k| len = imax(step, cand_bw4); 133| 914k| } 134| 869k|} refmvs.c:scan_col: 141| 7.50M|{ 142| 7.50M| const refmvs_block *cand_b = &b[0][bx4]; 143| 7.50M| const enum BlockSize first_cand_bs = cand_b->bs; 144| 7.50M| const uint8_t *const first_cand_b_dim = dav1d_block_dimensions[first_cand_bs]; 145| 7.50M| int cand_bh4 = first_cand_b_dim[1]; 146| 7.50M| int len = imax(step, imin(bh4, cand_bh4)); 147| | 148| 7.50M| if (bh4 <= cand_bh4) { ------------------ | Branch (148:9): [True: 6.25M, False: 1.25M] ------------------ 149| | // FIXME weight can be higher for odd blocks (by4 & 1), but then the 150| | // position of the first block has to be odd already, i.e. not just 151| | // for col_offset=-3/-5 152| | // FIXME why can this not be cand_bh4? 153| 6.25M| const int weight = bh4 == 1 ? 2 : ------------------ | Branch (153:28): [True: 1.89M, False: 4.36M] ------------------ 154| 6.25M| imax(2, imin(2 * max_cols, first_cand_b_dim[0])); 155| 6.25M| add_spatial_candidate(mvstack, cnt, len * weight, cand_b, ref, gmv, 156| 6.25M| have_newmv_match, have_refmv_match); 157| 6.25M| return weight >> 1; 158| 6.25M| } 159| | 160| 2.40M| for (int y = 0;;) { 161| | // FIXME if we overhang above, we could fill a bitmask so we don't have 162| | // to repeat the add_spatial_candidate() for the next row, but just increase 163| | // the weight here 164| 2.40M| add_spatial_candidate(mvstack, cnt, len * 2, cand_b, ref, gmv, 165| 2.40M| have_newmv_match, have_refmv_match); 166| 2.40M| y += len; 167| 2.40M| if (y >= h4) return 1; ------------------ | Branch (167:13): [True: 1.26M, False: 1.14M] ------------------ 168| 1.14M| cand_b = &b[y][bx4]; 169| 1.14M| cand_bh4 = dav1d_block_dimensions[cand_b->bs][1]; 170| 1.14M| assert(cand_bh4 < bh4); ------------------ | Branch (170:9): [True: 1.14M, False: 18.4E] ------------------ 171| 1.14M| len = imax(step, cand_bh4); 172| 1.14M| } 173| 1.25M|} refmvs.c:add_spatial_candidate: 46| 21.9M|{ 47| 21.9M| if (b->mv.mv[0].n == INVALID_MV) return; // intra block, no intrabc ------------------ | | 40| 21.9M|#define INVALID_MV 0x80008000 ------------------ | Branch (47:9): [True: 4.28M, False: 17.6M] ------------------ 48| | 49| 17.6M| if (ref.ref[1] == -1) { ------------------ | Branch (49:9): [True: 15.7M, False: 1.89M] ------------------ 50| 19.8M| for (int n = 0; n < 2; n++) { ------------------ | Branch (50:25): [True: 17.9M, False: 1.89M] ------------------ 51| 17.9M| if (b->ref.ref[n] == ref.ref[0]) { ------------------ | Branch (51:17): [True: 13.8M, False: 4.09M] ------------------ 52| 13.8M| const mv cand_mv = ((b->mf & 1) && gmv[0].n != INVALID_MV) ? ------------------ | | 40| 3.88M|#define INVALID_MV 0x80008000 ------------------ | Branch (52:37): [True: 3.88M, False: 9.98M] | Branch (52:52): [True: 1.34M, False: 2.53M] ------------------ 53| 12.5M| gmv[0] : b->mv.mv[n]; 54| | 55| 13.8M| *have_refmv_match = 1; 56| 13.8M| *have_newmv_match |= b->mf >> 1; 57| | 58| 13.8M| const int last = *cnt; 59| 23.7M| for (int m = 0; m < last; m++) ------------------ | Branch (59:33): [True: 15.5M, False: 8.16M] ------------------ 60| 15.5M| if (mvstack[m].mv.mv[0].n == cand_mv.n) { ------------------ | Branch (60:25): [True: 5.70M, False: 9.83M] ------------------ 61| 5.70M| mvstack[m].weight += weight; 62| 5.70M| return; 63| 5.70M| } 64| | 65| 8.21M| if (last < 8) { ------------------ | Branch (65:21): [True: 8.21M, False: 18.4E] ------------------ 66| 8.21M| mvstack[last].mv.mv[0] = cand_mv; 67| 8.21M| mvstack[last].weight = weight; 68| 8.21M| *cnt = last + 1; 69| 8.21M| } 70| 8.16M| return; 71| 13.8M| } 72| 17.9M| } 73| 15.7M| } else if (b->ref.pair == ref.pair) { ------------------ | Branch (73:16): [True: 728k, False: 1.16M] ------------------ 74| 728k| const refmvs_mvpair cand_mv = { .mv = { 75| 728k| [0] = ((b->mf & 1) && gmv[0].n != INVALID_MV) ? gmv[0] : b->mv.mv[0], ------------------ | | 40| 55.0k|#define INVALID_MV 0x80008000 ------------------ | Branch (75:20): [True: 55.0k, False: 672k] | Branch (75:35): [True: 29.9k, False: 25.1k] ------------------ 76| 728k| [1] = ((b->mf & 1) && gmv[1].n != INVALID_MV) ? gmv[1] : b->mv.mv[1], ------------------ | | 40| 55.0k|#define INVALID_MV 0x80008000 ------------------ | Branch (76:20): [True: 55.0k, False: 672k] | Branch (76:35): [True: 20.6k, False: 34.3k] ------------------ 77| 728k| }}; 78| | 79| 728k| *have_refmv_match = 1; 80| 728k| *have_newmv_match |= b->mf >> 1; 81| | 82| 728k| const int last = *cnt; 83| 1.10M| for (int n = 0; n < last; n++) ------------------ | Branch (83:25): [True: 682k, False: 426k] ------------------ 84| 682k| if (mvstack[n].mv.n == cand_mv.n) { ------------------ | Branch (84:17): [True: 301k, False: 381k] ------------------ 85| 301k| mvstack[n].weight += weight; 86| 301k| return; 87| 301k| } 88| | 89| 426k| if (last < 8) { ------------------ | Branch (89:13): [True: 425k, False: 712] ------------------ 90| 425k| mvstack[last].mv = cand_mv; 91| 425k| mvstack[last].weight = weight; 92| 425k| *cnt = last + 1; 93| 425k| } 94| 426k| } 95| 17.6M|} refmvs.c:add_temporal_candidate: 198| 7.38M|{ 199| 7.38M| if (rb->mv.n == INVALID_MV) return; ------------------ | | 40| 7.38M|#define INVALID_MV 0x80008000 ------------------ | Branch (199:9): [True: 5.13M, False: 2.25M] ------------------ 200| | 201| 2.25M| union mv mv = mv_projection(rb->mv, rf->pocdiff[ref.ref[0] - 1], rb->ref); 202| 2.25M| fix_mv_precision(rf->frm_hdr, &mv); 203| | 204| 2.25M| const int last = *cnt; 205| 2.25M| if (ref.ref[1] == -1) { ------------------ | Branch (205:9): [True: 1.59M, False: 664k] ------------------ 206| 1.59M| if (globalmv_ctx) ------------------ | Branch (206:13): [True: 312k, False: 1.28M] ------------------ 207| 312k| *globalmv_ctx = (abs(mv.x - gmv[0].x) | abs(mv.y - gmv[0].y)) >= 16; 208| | 209| 4.30M| for (int n = 0; n < last; n++) ------------------ | Branch (209:25): [True: 3.83M, False: 476k] ------------------ 210| 3.83M| if (mvstack[n].mv.mv[0].n == mv.n) { ------------------ | Branch (210:17): [True: 1.11M, False: 2.71M] ------------------ 211| 1.11M| mvstack[n].weight += 2; 212| 1.11M| return; 213| 1.11M| } 214| 494k| if (last < 8) { ------------------ | Branch (214:13): [True: 494k, False: 18.4E] ------------------ 215| 494k| mvstack[last].mv.mv[0] = mv; 216| 494k| mvstack[last].weight = 2; 217| 494k| *cnt = last + 1; 218| 494k| } 219| 664k| } else { 220| 664k| refmvs_mvpair mvp = { .mv = { 221| 664k| [0] = mv, 222| 664k| [1] = mv_projection(rb->mv, rf->pocdiff[ref.ref[1] - 1], rb->ref), 223| 664k| }}; 224| 664k| fix_mv_precision(rf->frm_hdr, &mvp.mv[1]); 225| | 226| 1.51M| for (int n = 0; n < last; n++) ------------------ | Branch (226:25): [True: 1.36M, False: 149k] ------------------ 227| 1.36M| if (mvstack[n].mv.n == mvp.n) { ------------------ | Branch (227:17): [True: 514k, False: 847k] ------------------ 228| 514k| mvstack[n].weight += 2; 229| 514k| return; 230| 514k| } 231| 164k| if (last < 8) { ------------------ | Branch (231:13): [True: 164k, False: 18.4E] ------------------ 232| 164k| mvstack[last].mv = mvp; 233| 164k| mvstack[last].weight = 2; 234| 164k| *cnt = last + 1; 235| 164k| } 236| 149k| } 237| 2.25M|} refmvs.c:mv_projection: 175| 2.93M|static inline union mv mv_projection(const union mv mv, const int num, const int den) { 176| 2.93M| static const uint16_t div_mult[32] = { 177| 2.93M| 0, 16384, 8192, 5461, 4096, 3276, 2730, 2340, 178| 2.93M| 2048, 1820, 1638, 1489, 1365, 1260, 1170, 1092, 179| 2.93M| 1024, 963, 910, 862, 819, 780, 744, 712, 180| 2.93M| 682, 655, 630, 606, 585, 564, 546, 528 181| 2.93M| }; 182| 2.93M| assert(den > 0 && den < 32); ------------------ | Branch (182:5): [True: 2.93M, False: 18.4E] | Branch (182:5): [True: 2.93M, False: 18.4E] ------------------ 183| 2.93M| assert(num > -32 && num < 32); ------------------ | Branch (183:5): [True: 2.93M, False: 18.4E] | Branch (183:5): [True: 2.93M, False: 18.4E] ------------------ 184| 2.93M| const int frac = num * div_mult[den]; 185| 2.93M| const int y = mv.y * frac, x = mv.x * frac; 186| | // Round and clip according to AV1 spec section 7.9.3 187| 2.93M| return (union mv) { // 0x3fff == (1 << 14) - 1 188| 2.93M| .y = iclip((y + 8192 + (y >> 31)) >> 14, -0x3fff, 0x3fff), 189| 2.93M| .x = iclip((x + 8192 + (x >> 31)) >> 14, -0x3fff, 0x3fff) 190| 2.93M| }; 191| 2.93M|} refmvs.c:add_compound_extended_candidate: 245| 490k|{ 246| 490k| refmvs_candidate *const diff = &same[2]; 247| 490k| int *const diff_count = &same_count[2]; 248| | 249| 1.24M| for (int n = 0; n < 2; n++) { ------------------ | Branch (249:21): [True: 949k, False: 296k] ------------------ 250| 949k| const int cand_ref = cand_b->ref.ref[n]; 251| | 252| 949k| if (cand_ref <= 0) break; ------------------ | Branch (252:13): [True: 193k, False: 755k] ------------------ 253| | 254| 755k| mv cand_mv = cand_b->mv.mv[n]; 255| 755k| if (cand_ref == ref.ref[0]) { ------------------ | Branch (255:13): [True: 283k, False: 471k] ------------------ 256| 283k| if (same_count[0] < 2) ------------------ | Branch (256:17): [True: 272k, False: 11.2k] ------------------ 257| 272k| same[same_count[0]++].mv.mv[0] = cand_mv; 258| 283k| if (diff_count[1] < 2) { ------------------ | Branch (258:17): [True: 234k, False: 49.5k] ------------------ 259| 234k| if (sign1 ^ sign_bias[cand_ref - 1]) { ------------------ | Branch (259:21): [True: 21.5k, False: 212k] ------------------ 260| 21.5k| cand_mv.y = -cand_mv.y; 261| 21.5k| cand_mv.x = -cand_mv.x; 262| 21.5k| } 263| 234k| diff[diff_count[1]++].mv.mv[1] = cand_mv; 264| 234k| } 265| 471k| } else if (cand_ref == ref.ref[1]) { ------------------ | Branch (265:20): [True: 254k, False: 216k] ------------------ 266| 254k| if (same_count[1] < 2) ------------------ | Branch (266:17): [True: 246k, False: 8.28k] ------------------ 267| 246k| same[same_count[1]++].mv.mv[1] = cand_mv; 268| 254k| if (diff_count[0] < 2) { ------------------ | Branch (268:17): [True: 209k, False: 45.4k] ------------------ 269| 209k| if (sign0 ^ sign_bias[cand_ref - 1]) { ------------------ | Branch (269:21): [True: 22.2k, False: 187k] ------------------ 270| 22.2k| cand_mv.y = -cand_mv.y; 271| 22.2k| cand_mv.x = -cand_mv.x; 272| 22.2k| } 273| 209k| diff[diff_count[0]++].mv.mv[0] = cand_mv; 274| 209k| } 275| 254k| } else { 276| 216k| mv i_cand_mv = (union mv) { 277| 216k| .x = -cand_mv.x, 278| 216k| .y = -cand_mv.y 279| 216k| }; 280| | 281| 216k| if (diff_count[0] < 2) { ------------------ | Branch (281:17): [True: 170k, False: 46.5k] ------------------ 282| 170k| diff[diff_count[0]++].mv.mv[0] = 283| 170k| sign0 ^ sign_bias[cand_ref - 1] ? ------------------ | Branch (283:21): [True: 2.43k, False: 167k] ------------------ 284| 167k| i_cand_mv : cand_mv; 285| 170k| } 286| | 287| 216k| if (diff_count[1] < 2) { ------------------ | Branch (287:17): [True: 154k, False: 61.9k] ------------------ 288| 154k| diff[diff_count[1]++].mv.mv[1] = 289| 154k| sign1 ^ sign_bias[cand_ref - 1] ? ------------------ | Branch (289:21): [True: 5.06k, False: 149k] ------------------ 290| 149k| i_cand_mv : cand_mv; 291| 154k| } 292| 216k| } 293| 755k| } 294| 490k|} refmvs.c:add_single_extended_candidate: 299| 3.26M|{ 300| 6.42M| for (int n = 0; n < 2; n++) { ------------------ | Branch (300:21): [True: 6.34M, False: 71.5k] ------------------ 301| 6.34M| const int cand_ref = cand_b->ref.ref[n]; 302| | 303| 6.34M| if (cand_ref <= 0) break; ------------------ | Branch (303:13): [True: 3.19M, False: 3.15M] ------------------ 304| | // we need to continue even if cand_ref == ref.ref[0], since 305| | // the candidate could have been added as a globalmv variant, 306| | // which changes the value 307| | // FIXME if scan_{row,col}() returned a mask for the nearest 308| | // edge, we could skip the appropriate ones here 309| | 310| 3.15M| mv cand_mv = cand_b->mv.mv[n]; 311| 3.15M| if (sign ^ sign_bias[cand_ref - 1]) { ------------------ | Branch (311:13): [True: 16.9k, False: 3.13M] ------------------ 312| 16.9k| cand_mv.y = -cand_mv.y; 313| 16.9k| cand_mv.x = -cand_mv.x; 314| 16.9k| } 315| | 316| 3.15M| int m; 317| 3.15M| const int last = *cnt; 318| 3.45M| for (m = 0; m < last; m++) ------------------ | Branch (318:21): [True: 3.09M, False: 366k] ------------------ 319| 3.09M| if (cand_mv.n == mvstack[m].mv.mv[0].n) ------------------ | Branch (319:17): [True: 2.78M, False: 301k] ------------------ 320| 2.78M| break; 321| 3.15M| if (m == last) { ------------------ | Branch (321:13): [True: 370k, False: 2.78M] ------------------ 322| 370k| mvstack[m].mv.mv[0] = cand_mv; 323| 370k| mvstack[m].weight = 2; // "minimal" 324| 370k| *cnt = last + 1; 325| 370k| } 326| 3.15M| } 327| 3.26M|} decode.c:dav1d_refmvs_save_tmvs: 145| 832k|{ 146| 832k| const refmvs_frame *const rf = rt->rf; 147| | 148| 832k| assert(row_start8 >= 0); ------------------ | Branch (148:5): [True: 832k, False: 48] ------------------ 149| 832k| assert((unsigned) (row_end8 - row_start8) <= 16U); ------------------ | Branch (149:5): [True: 832k, False: 18.4E] ------------------ 150| 832k| row_end8 = imin(row_end8, rf->ih8); 151| 832k| col_end8 = imin(col_end8, rf->iw8); 152| | 153| 832k| const ptrdiff_t stride = rf->rp_stride; 154| 832k| const uint8_t *const ref_sign = rf->mfmv_sign; 155| 832k| refmvs_temporal_block *rp = &rf->rp[row_start8 * stride]; 156| | 157| 832k| dsp->save_tmvs(rp, stride, rt->r + 6, ref_sign, 158| 832k| col_end8, row_end8, col_start8, row_start8); 159| 832k|} dav1d_init_last_nonzero_col_from_eob_tables: 350| 2.99k|COLD void dav1d_init_last_nonzero_col_from_eob_tables(void) { 351| | static pthread_once_t initted = PTHREAD_ONCE_INIT; 352| 2.99k| pthread_once(&initted, init_internal); 353| 2.99k|} scan.c:init_internal: 333| 1|static COLD void init_internal(void) { 334| 1| init_tbl(last_nonzero_col_from_eob_4x4, scan_4x4, 4, 4); 335| 1| init_tbl(last_nonzero_col_from_eob_8x8, scan_8x8, 8, 8); 336| 1| init_tbl(last_nonzero_col_from_eob_16x16, scan_16x16, 16, 16); 337| 1| init_tbl(last_nonzero_col_from_eob_32x32, scan_32x32, 32, 32); 338| 1| init_tbl(last_nonzero_col_from_eob_4x8, scan_4x8, 4, 8); 339| 1| init_tbl(last_nonzero_col_from_eob_8x4, scan_8x4, 8, 4); 340| 1| init_tbl(last_nonzero_col_from_eob_8x16, scan_8x16, 8, 16); 341| 1| init_tbl(last_nonzero_col_from_eob_16x8, scan_16x8, 16, 8); 342| 1| init_tbl(last_nonzero_col_from_eob_16x32, scan_16x32, 16, 32); 343| 1| init_tbl(last_nonzero_col_from_eob_32x16, scan_32x16, 32, 16); 344| 1| init_tbl(last_nonzero_col_from_eob_4x16, scan_4x16, 4, 16); 345| 1| init_tbl(last_nonzero_col_from_eob_16x4, scan_16x4, 16, 4); 346| 1| init_tbl(last_nonzero_col_from_eob_8x32, scan_8x32, 8, 32); 347| 1| init_tbl(last_nonzero_col_from_eob_32x8, scan_32x8, 32, 8); 348| 1|} scan.c:init_tbl: 321| 14|{ 322| 14| int max_col = 0; 323| 218| for (int y = 0, n = 0; y < h; y++) { ------------------ | Branch (323:28): [True: 204, False: 14] ------------------ 324| 3.54k| for (int x = 0; x < w; x++, n++) { ------------------ | Branch (324:25): [True: 3.34k, False: 204] ------------------ 325| 3.34k| const int rc = scan[n]; 326| 3.34k| const int rcx = rc & (h - 1); 327| 3.34k| max_col = imax(max_col, rcx); 328| 3.34k| last_nonzero_col_from_eob[n] = max_col; 329| 3.34k| } 330| 204| } 331| 14|} thread_task.c:dav1d_set_thread_name: 152| 37.6k|static inline void dav1d_set_thread_name(const char *const name) { 153| | prctl(PR_SET_NAME, name); 154| 37.6k|} dav1d_task_create_tile_sbrow: 270| 589k|{ 271| 589k| Dav1dTask *tasks = f->task_thread.tile_tasks[0]; 272| 589k| const int uses_2pass = f->c->n_fc > 1; 273| 589k| const int n_tasks_per_pass = f->frame_hdr->tiling.cols * f->frame_hdr->tiling.rows; 274| 589k| const int n_tasks = n_tasks_per_pass * (1 + uses_2pass); 275| 589k| if (pass < 2) { ------------------ | Branch (275:9): [True: 294k, False: 294k] ------------------ 276| 294k| if (n_tasks > f->task_thread.num_tile_tasks) { ------------------ | Branch (276:13): [True: 16.0k, False: 278k] ------------------ 277| 16.0k| const size_t size = sizeof(Dav1dTask) * n_tasks; 278| 16.0k| tasks = dav1d_realloc(ALLOC_COMMON_CTX, f->task_thread.tile_tasks[0], size); ------------------ | | 133| 16.0k|#define dav1d_realloc(type, ptr, sz) realloc(ptr, sz) ------------------ 279| 16.0k| if (!tasks) return -1; ------------------ | Branch (279:17): [True: 0, False: 16.0k] ------------------ 280| 16.0k| memset(tasks, 0, size); 281| 16.0k| f->task_thread.tile_tasks[0] = tasks; 282| 16.0k| f->task_thread.num_tile_tasks = n_tasks; 283| 16.0k| } 284| 294k| f->task_thread.tile_tasks[1] = tasks + n_tasks_per_pass; 285| 294k| } 286| 589k| assert(n_tasks <= f->task_thread.num_tile_tasks); ------------------ | Branch (286:5): [True: 589k, False: 66] ------------------ 287| | 288| 589k| Dav1dTask *pf_t; 289| 589k| if (create_filter_sbrow(f, pass, &pf_t)) ------------------ | Branch (289:9): [True: 0, False: 589k] ------------------ 290| 0| return -1; 291| | 292| 589k| Dav1dTask *const p1_tasks = f->task_thread.tile_tasks[1]; 293| 589k| Dav1dTask *prev_t = NULL; 294| 589k| if (pass == 2) { ------------------ | Branch (294:9): [True: 295k, False: 294k] ------------------ 295| 295k| prev_t = &p1_tasks[n_tasks_per_pass - 1]; 296| | // PF task is scheduled after the last sby=0 TILE task 297| 295k| if (f->frame_hdr->tiling.rows == 1) ------------------ | Branch (297:13): [True: 287k, False: 7.35k] ------------------ 298| 287k| prev_t = prev_t->next; 299| 295k| } 300| 589k| tasks += (pass & 1) * n_tasks_per_pass; 301| 1.21M| for (int tile_idx = 0; tile_idx < n_tasks_per_pass; tile_idx++) { ------------------ | Branch (301:28): [True: 629k, False: 589k] ------------------ 302| 629k| Dav1dTileState *const ts = &f->ts[tile_idx]; 303| 629k| Dav1dTask *t = &tasks[tile_idx]; 304| 629k| t->sby = ts->tiling.row_start >> f->sb_shift; 305| 629k| if (pf_t && t->sby) { ------------------ | Branch (305:13): [True: 616k, False: 13.0k] | Branch (305:21): [True: 14.7k, False: 602k] ------------------ 306| 14.7k| prev_t->next = pf_t; 307| 14.7k| prev_t = pf_t; 308| 14.7k| pf_t = NULL; 309| 14.7k| } 310| 629k| t->recon_progress = 0; 311| 629k| t->deblock_progress = 0; 312| 629k| t->deps_skip = 0; 313| 629k| t->type = pass != 1 ? DAV1D_TASK_TYPE_TILE_RECONSTRUCTION : ------------------ | Branch (313:19): [True: 314k, False: 314k] ------------------ 314| 629k| DAV1D_TASK_TYPE_TILE_ENTROPY; 315| 629k| t->frame_idx = (int)(f - f->c->fc); 316| 629k| if (prev_t) prev_t->next = t; ------------------ | Branch (316:13): [True: 335k, False: 294k] ------------------ 317| 629k| prev_t = t; 318| 629k| } 319| 589k| if (pf_t) { ------------------ | Branch (319:9): [True: 575k, False: 14.1k] ------------------ 320| 575k| prev_t->next = pf_t; 321| 575k| prev_t = pf_t; 322| 575k| } 323| 589k| prev_t->next = NULL; 324| | 325| 589k| atomic_store(&f->task_thread.done[pass & 1], 0); 326| | 327| | // XXX in theory this could be done locklessly, at this point they are no 328| | // tasks in the frameQ, so no other runner should be using this lock, but 329| | // we must add both passes at once 330| 589k| if (!(pass & 1)) { ------------------ | Branch (330:9): [True: 295k, False: 294k] ------------------ 331| 295k| pthread_mutex_lock(&f->task_thread.pending_tasks.lock); 332| 295k| assert(f->task_thread.pending_tasks.head == NULL); ------------------ | Branch (332:9): [True: 295k, False: 18.4E] ------------------ 333| 295k| f->task_thread.pending_tasks.head = f->task_thread.tile_tasks[pass == 2]; 334| 295k| f->task_thread.pending_tasks.tail = prev_t; 335| 295k| atomic_store(&f->task_thread.pending_tasks.merge, 1); 336| 295k| atomic_store(&f->task_thread.init_done, 1); 337| 295k| pthread_mutex_unlock(&f->task_thread.pending_tasks.lock); 338| 295k| } 339| 589k| return 0; 340| 589k|} dav1d_task_frame_init: 342| 356k|void dav1d_task_frame_init(Dav1dFrameContext *const f) { 343| 356k| const Dav1dContext *const c = f->c; 344| | 345| 356k| atomic_store(&f->task_thread.init_done, 0); 346| | // schedule init task, which will schedule the remaining tasks 347| 356k| Dav1dTask *const t = &f->task_thread.init_task; 348| 356k| t->type = DAV1D_TASK_TYPE_INIT; 349| 356k| t->frame_idx = (int)(f - c->fc); 350| 356k| t->sby = 0; 351| 356k| t->recon_progress = t->deblock_progress = 0; 352| 356k| insert_task(f, t, 1); 353| 356k|} dav1d_task_delayed_fg: 357| 10.1k|{ 358| 10.1k| struct TaskThreadData *const ttd = &c->task_thread; 359| 10.1k| ttd->delayed_fg.in = in; 360| 10.1k| ttd->delayed_fg.out = out; 361| 10.1k| ttd->delayed_fg.type = DAV1D_TASK_TYPE_FG_PREP; 362| 10.1k| atomic_init(&ttd->delayed_fg.progress[0], 0); 363| 10.1k| atomic_init(&ttd->delayed_fg.progress[1], 0); 364| 10.1k| pthread_mutex_lock(&ttd->lock); 365| 10.1k| ttd->delayed_fg.exec = 1; 366| 10.1k| ttd->delayed_fg.finished = 0; 367| 10.1k| pthread_cond_signal(&ttd->cond); 368| 10.1k| do { 369| 10.1k| pthread_cond_wait(&ttd->delayed_fg.cond, &ttd->lock); 370| 10.1k| } while (!ttd->delayed_fg.finished); ------------------ | Branch (370:14): [True: 0, False: 10.1k] ------------------ 371| 10.1k| pthread_mutex_unlock(&ttd->lock); 372| 10.1k|} dav1d_worker_task: 556| 37.6k|void *dav1d_worker_task(void *data) { 557| 37.6k| Dav1dTaskContext *const tc = data; 558| 37.6k| const Dav1dContext *const c = tc->c; 559| 37.6k| struct TaskThreadData *const ttd = tc->task_thread.ttd; 560| | 561| 37.6k| dav1d_set_thread_name("dav1d-worker"); 562| | 563| 37.6k| pthread_mutex_lock(&ttd->lock); 564| 21.7M| for (;;) { 565| 21.7M| if (tc->task_thread.die) break; ------------------ | Branch (565:13): [True: 37.6k, False: 21.6M] ------------------ 566| 21.6M| if (atomic_load(c->flush)) goto park; ------------------ | Branch (566:13): [True: 9.51k, False: 21.6M] ------------------ 567| | 568| 21.6M| merge_pending(c); 569| 21.6M| if (ttd->delayed_fg.exec) { // run delayed film grain first ------------------ | Branch (569:13): [True: 16.2k, False: 21.6M] ------------------ 570| 16.2k| delayed_fg_task(c, ttd); 571| 16.2k| continue; 572| 16.2k| } 573| 21.6M| Dav1dFrameContext *f; 574| 21.6M| Dav1dTask *t, *prev_t = NULL; 575| 21.6M| if (c->n_fc > 1) { // run init tasks second ------------------ | Branch (575:13): [True: 21.6M, False: 0] ------------------ 576| 107M| for (unsigned i = 0; i < c->n_fc; i++) { ------------------ | Branch (576:34): [True: 86.2M, False: 21.2M] ------------------ 577| 86.2M| const unsigned first = atomic_load(&ttd->first); 578| 86.2M| f = &c->fc[(first + i) % c->n_fc]; 579| 86.2M| if (atomic_load(&f->task_thread.init_done)) continue; ------------------ | Branch (579:21): [True: 77.0M, False: 9.29M] ------------------ 580| 9.29M| t = f->task_thread.task_head; 581| 9.29M| if (!t) continue; ------------------ | Branch (581:21): [True: 7.80M, False: 1.48M] ------------------ 582| 1.48M| if (t->type == DAV1D_TASK_TYPE_INIT) goto found; ------------------ | Branch (582:21): [True: 355k, False: 1.13M] ------------------ 583| 1.13M| if (t->type == DAV1D_TASK_TYPE_INIT_CDF) { ------------------ | Branch (583:21): [True: 1.13M, False: 0] ------------------ 584| | // XXX This can be a simple else, if adding tasks of both 585| | // passes at once (in dav1d_task_create_tile_sbrow). 586| | // Adding the tasks to the pending Q can result in a 587| | // thread merging them before setting init_done. 588| | // We will need to set init_done before adding to the 589| | // pending Q, so maybe return the tasks, set init_done, 590| | // and add to pending Q only then. 591| 1.13M| const int p1 = f->in_cdf.progress ? ------------------ | Branch (591:36): [True: 1.13M, False: 0] ------------------ 592| 1.13M| atomic_load(f->in_cdf.progress) : 1; 593| 1.13M| if (p1) { ------------------ | Branch (593:25): [True: 11.0k, False: 1.12M] ------------------ 594| 11.0k| atomic_fetch_or(&f->task_thread.error, p1 == TILE_ERROR); 595| 11.0k| goto found; 596| 11.0k| } 597| 1.13M| } 598| 1.13M| } 599| 21.6M| } 600| 24.0M| while (ttd->cur < c->n_fc) { // run decoding tasks last ------------------ | Branch (600:16): [True: 23.1M, False: 918k] ------------------ 601| 23.1M| const unsigned first = atomic_load(&ttd->first); 602| 23.1M| f = &c->fc[(first + ttd->cur) % c->n_fc]; 603| 23.1M| merge_pending_frame(f); 604| 23.1M| prev_t = f->task_thread.task_cur_prev; 605| 23.1M| t = prev_t ? prev_t->next : f->task_thread.task_head; ------------------ | Branch (605:17): [True: 150k, False: 23.0M] ------------------ 606| 28.9M| while (t) { ------------------ | Branch (606:20): [True: 26.1M, False: 2.80M] ------------------ 607| 26.1M| if (t->type == DAV1D_TASK_TYPE_INIT_CDF) goto next; ------------------ | Branch (607:21): [True: 27.5k, False: 26.0M] ------------------ 608| 26.0M| else if (t->type == DAV1D_TASK_TYPE_TILE_ENTROPY || ------------------ | Branch (608:26): [True: 2.27M, False: 23.8M] ------------------ 609| 23.8M| t->type == DAV1D_TASK_TYPE_TILE_RECONSTRUCTION) ------------------ | Branch (609:26): [True: 2.55M, False: 21.2M] ------------------ 610| 4.82M| { 611| | // if not bottom sbrow of tile, this task will be re-added 612| | // after it's finished 613| 4.82M| if (!check_tile(t, f, c->n_fc > 1)) ------------------ | Branch (613:25): [True: 4.46M, False: 359k] ------------------ 614| 4.46M| goto found; 615| 21.2M| } else if (t->recon_progress) { ------------------ | Branch (615:28): [True: 19.5M, False: 1.73M] ------------------ 616| 19.5M| const int p = t->type == DAV1D_TASK_TYPE_ENTROPY_PROGRESS; 617| 19.5M| int error = atomic_load(&f->task_thread.error); 618| 19.5M| assert(!atomic_load(&f->task_thread.done[p]) || error); ------------------ | Branch (618:21): [True: 13.5M, False: 5.97M] | Branch (618:21): [True: 5.97M, False: 0] ------------------ 619| 19.5M| const int tile_row_base = f->frame_hdr->tiling.cols * 620| 19.5M| f->frame_thread.next_tile_row[p]; 621| 19.5M| if (p) { ------------------ | Branch (621:25): [True: 8.28M, False: 11.2M] ------------------ 622| 8.28M| atomic_int *const prog = &f->frame_thread.entropy_progress; 623| 8.28M| const int p1 = atomic_load(prog); 624| 8.28M| if (p1 < t->sby) goto next; ------------------ | Branch (624:29): [True: 14.6k, False: 8.26M] ------------------ 625| 8.28M| atomic_fetch_or(&f->task_thread.error, p1 == TILE_ERROR); 626| 8.26M| } 627| 36.4M| for (int tc = 0; tc < f->frame_hdr->tiling.cols; tc++) { ------------------ | Branch (627:38): [True: 20.7M, False: 15.7M] ------------------ 628| 20.7M| Dav1dTileState *const ts = &f->ts[tile_row_base + tc]; 629| 20.7M| const int p2 = atomic_load(&ts->progress[p]); 630| 20.7M| if (p2 < t->recon_progress) goto next; ------------------ | Branch (630:29): [True: 3.79M, False: 16.9M] ------------------ 631| 20.7M| atomic_fetch_or(&f->task_thread.error, p2 == TILE_ERROR); 632| 16.9M| } 633| 15.7M| if (t->sby + 1 < f->sbh) { ------------------ | Branch (633:25): [True: 15.1M, False: 588k] ------------------ 634| | // add sby+1 to list to replace this one 635| 15.1M| Dav1dTask *next_t = &t[1]; 636| 15.1M| *next_t = *t; 637| 15.1M| next_t->sby++; 638| 15.1M| const int ntr = f->frame_thread.next_tile_row[p] + 1; 639| 15.1M| const int start = f->frame_hdr->tiling.row_start_sb[ntr]; 640| 15.1M| if (next_t->sby == start) ------------------ | Branch (640:29): [True: 22.6k, False: 15.1M] ------------------ 641| 22.6k| f->frame_thread.next_tile_row[p] = ntr; 642| 15.1M| next_t->recon_progress = next_t->sby + 1; 643| 15.1M| insert_task(f, next_t, 0); 644| 15.1M| } 645| 15.7M| goto found; 646| 19.5M| } else if (t->type == DAV1D_TASK_TYPE_CDEF) { ------------------ | Branch (646:28): [True: 1.40M, False: 323k] ------------------ 647| 1.40M| atomic_uint *prog = f->frame_thread.copy_lpf_progress; 648| 1.40M| const int p1 = atomic_load(&prog[(t->sby - 1) >> 5]); 649| 1.40M| if (p1 & (1U << ((t->sby - 1) & 31))) ------------------ | Branch (649:25): [True: 150k, False: 1.25M] ------------------ 650| 150k| goto found; 651| 1.40M| } else { 652| 323k| assert(t->deblock_progress); ------------------ | Branch (652:21): [True: 323k, False: 0] ------------------ 653| 323k| const int p1 = atomic_load(&f->frame_thread.deblock_progress); 654| 323k| if (p1 >= t->deblock_progress) { ------------------ | Branch (654:25): [True: 18.4k, False: 304k] ------------------ 655| 18.4k| atomic_fetch_or(&f->task_thread.error, p1 == TILE_ERROR); 656| 18.4k| goto found; 657| 18.4k| } 658| 323k| } 659| 5.75M| next: 660| 5.75M| prev_t = t; 661| 5.75M| t = t->next; 662| 5.75M| f->task_thread.task_cur_prev = prev_t; 663| 5.75M| } 664| 2.80M| ttd->cur++; 665| 2.80M| } 666| 918k| if (reset_task_cur(c, ttd, UINT_MAX)) continue; ------------------ | Branch (666:13): [True: 9.94k, False: 908k] ------------------ 667| 908k| if (merge_pending(c)) continue; ------------------ | Branch (667:13): [True: 5.02k, False: 903k] ------------------ 668| 913k| park: 669| 913k| tc->task_thread.flushed = 1; 670| 913k| pthread_cond_signal(&tc->task_thread.td.cond); 671| | // we want to be woken up next time progress is signaled 672| 913k| atomic_store(&ttd->cond_signaled, 0); 673| 913k| pthread_cond_wait(&ttd->cond, &ttd->lock); 674| 913k| tc->task_thread.flushed = 0; 675| 913k| reset_task_cur(c, ttd, UINT_MAX); 676| 913k| continue; 677| | 678| 20.7M| found: 679| | // remove t from list 680| 20.7M| if (prev_t) prev_t->next = t->next; ------------------ | Branch (680:13): [True: 3.08M, False: 17.6M] ------------------ 681| 17.6M| else f->task_thread.task_head = t->next; 682| 20.7M| if (!t->next) f->task_thread.task_tail = prev_t; ------------------ | Branch (682:13): [True: 938k, False: 19.7M] ------------------ 683| 20.7M| if (t->type > DAV1D_TASK_TYPE_INIT_CDF && !f->task_thread.task_head) ------------------ | Branch (683:13): [True: 20.3M, False: 366k] | Branch (683:51): [True: 294k, False: 20.0M] ------------------ 684| 294k| ttd->cur++; 685| 20.7M| t->next = NULL; 686| | // we don't need to check cond_signaled here, since we found a task 687| | // after the last signal so we want to re-signal the next waiting thread 688| | // and again won't need to signal after that 689| 20.7M| atomic_store(&ttd->cond_signaled, 1); 690| 20.7M| pthread_cond_signal(&ttd->cond); 691| 20.7M| pthread_mutex_unlock(&ttd->lock); 692| 33.1M| found_unlocked:; 693| 33.1M| const int flush = atomic_load(c->flush); 694| 33.1M| int error = atomic_fetch_or(&f->task_thread.error, flush) | flush; 695| | 696| | // run it 697| 33.1M| tc->f = f; 698| 33.1M| int sby = t->sby; 699| 33.1M| switch (t->type) { 700| 355k| case DAV1D_TASK_TYPE_INIT: { ------------------ | Branch (700:9): [True: 355k, False: 32.7M] ------------------ 701| 355k| assert(c->n_fc > 1); ------------------ | Branch (701:13): [True: 355k, False: 8] ------------------ 702| 355k| int res = dav1d_decode_frame_init(f); 703| 355k| int p1 = f->in_cdf.progress ? atomic_load(f->in_cdf.progress) : 1; ------------------ | Branch (703:22): [True: 69.4k, False: 285k] ------------------ 704| 355k| if (res || p1 == TILE_ERROR) { ------------------ | | 36| 354k|#define TILE_ERROR (INT_MAX - 1) ------------------ | Branch (704:17): [True: 653, False: 354k] | Branch (704:24): [True: 46.4k, False: 308k] ------------------ 705| 46.5k| pthread_mutex_lock(&ttd->lock); 706| 46.5k| abort_frame(f, res ? res : DAV1D_ERR(EINVAL)); ------------------ | | 58| 46.5k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (706:32): [True: 0, False: 46.5k] ------------------ 707| 46.5k| reset_task_cur(c, ttd, t->frame_idx); 708| 308k| } else { 709| 308k| t->type = DAV1D_TASK_TYPE_INIT_CDF; 710| 308k| if (p1) goto found_unlocked; ------------------ | Branch (710:21): [True: 297k, False: 11.7k] ------------------ 711| 11.7k| add_pending(f, t); 712| 11.7k| pthread_mutex_lock(&ttd->lock); 713| 11.7k| } 714| 58.2k| continue; 715| 355k| } 716| 308k| case DAV1D_TASK_TYPE_INIT_CDF: { ------------------ | Branch (716:9): [True: 308k, False: 32.8M] ------------------ 717| 308k| assert(c->n_fc > 1); ------------------ | Branch (717:13): [True: 308k, False: 18.4E] ------------------ 718| 308k| int res = DAV1D_ERR(EINVAL); ------------------ | | 58| 308k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 719| 308k| if (!atomic_load(&f->task_thread.error)) ------------------ | Branch (719:17): [True: 302k, False: 6.07k] ------------------ 720| 302k| res = dav1d_decode_frame_init_cdf(f); 721| 308k| if (f->frame_hdr->refresh_context && !f->task_thread.update_set) ------------------ | Branch (721:17): [True: 26.7k, False: 282k] | Branch (721:50): [True: 2.14k, False: 24.5k] ------------------ 722| 308k| atomic_store(f->out_cdf.progress, res < 0 ? TILE_ERROR : 1); ------------------ | Branch (722:17): [True: 2.14k, False: 0] ------------------ 723| 898k| for (int p = 1; p <= 2 && !res; p++) ------------------ | Branch (723:29): [True: 602k, False: 295k] | Branch (723:39): [True: 589k, False: 13.5k] ------------------ 724| 589k| res = dav1d_task_create_tile_sbrow(f, p, 0); 725| 308k| pthread_mutex_lock(&ttd->lock); 726| 308k| if (res) { ------------------ | Branch (726:17): [True: 13.5k, False: 295k] ------------------ 727| 13.5k| abort_frame(f, DAV1D_ERR(ENOMEM)); ------------------ | | 58| 13.5k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ 728| 13.5k| reset_task_cur(c, ttd, t->frame_idx); 729| 13.5k| atomic_store(&f->task_thread.init_done, 1); 730| 13.5k| } 731| 308k| continue; 732| 308k| } 733| 8.29M| case DAV1D_TASK_TYPE_TILE_ENTROPY: ------------------ | Branch (733:9): [True: 8.29M, False: 24.8M] ------------------ 734| 16.5M| case DAV1D_TASK_TYPE_TILE_RECONSTRUCTION: { ------------------ | Branch (734:9): [True: 8.28M, False: 24.8M] ------------------ 735| 16.5M| const int p = t->type == DAV1D_TASK_TYPE_TILE_ENTROPY; 736| 16.5M| const int tile_idx = (int)(t - f->task_thread.tile_tasks[p]); 737| 16.5M| Dav1dTileState *const ts = &f->ts[tile_idx]; 738| | 739| 16.5M| tc->ts = ts; 740| 16.5M| tc->by = sby << f->sb_shift; 741| 16.5M| const int uses_2pass = c->n_fc > 1; 742| 16.5M| tc->frame_thread.pass = !uses_2pass ? 0 : ------------------ | Branch (742:37): [True: 0, False: 16.5M] ------------------ 743| 16.5M| 1 + (t->type == DAV1D_TASK_TYPE_TILE_RECONSTRUCTION); 744| 16.5M| if (!error) error = dav1d_decode_tile_sbrow(tc); ------------------ | Branch (744:17): [True: 4.12M, False: 12.4M] ------------------ 745| 16.5M| const int progress = error ? TILE_ERROR : 1 + sby; ------------------ | | 36| 12.5M|#define TILE_ERROR (INT_MAX - 1) ------------------ | Branch (745:34): [True: 12.5M, False: 4.03M] ------------------ 746| | 747| | // signal progress 748| 16.5M| atomic_fetch_or(&f->task_thread.error, error); 749| 16.5M| if (((sby + 1) << f->sb_shift) < ts->tiling.row_end) { ------------------ | Branch (749:17): [True: 15.9M, False: 642k] ------------------ 750| 15.9M| t->sby++; 751| 15.9M| t->deps_skip = 0; 752| 15.9M| if (!check_tile(t, f, uses_2pass)) { ------------------ | Branch (752:21): [True: 12.0M, False: 3.84M] ------------------ 753| 12.0M| atomic_store(&ts->progress[p], progress); 754| 12.0M| reset_task_cur_async(ttd, t->frame_idx, c->n_fc); 755| 12.0M| if (!atomic_fetch_or(&ttd->cond_signaled, 1)) ------------------ | Branch (755:25): [True: 104, False: 12.0M] ------------------ 756| 104| pthread_cond_signal(&ttd->cond); 757| 12.0M| goto found_unlocked; 758| 12.0M| } 759| 15.9M| atomic_store(&ts->progress[p], progress); 760| 3.84M| add_pending(f, t); 761| 3.84M| pthread_mutex_lock(&ttd->lock); 762| 3.84M| } else { 763| 642k| pthread_mutex_lock(&ttd->lock); 764| 642k| atomic_store(&ts->progress[p], progress); 765| 642k| reset_task_cur(c, ttd, t->frame_idx); 766| 642k| error = atomic_load(&f->task_thread.error); 767| 642k| if (f->frame_hdr->refresh_context && ------------------ | Branch (767:21): [True: 53.0k, False: 589k] ------------------ 768| 53.0k| tc->frame_thread.pass <= 1 && f->task_thread.update_set && ------------------ | Branch (768:21): [True: 26.6k, False: 26.4k] | Branch (768:51): [True: 26.6k, False: 0] ------------------ 769| 26.6k| f->frame_hdr->tiling.update == tile_idx) ------------------ | Branch (769:21): [True: 20.6k, False: 5.91k] ------------------ 770| 20.6k| { 771| 20.6k| if (!error) ------------------ | Branch (771:25): [True: 9.85k, False: 10.8k] ------------------ 772| 9.85k| dav1d_cdf_thread_update(f->frame_hdr, f->out_cdf.data.cdf, 773| 9.85k| &f->ts[f->frame_hdr->tiling.update].cdf); 774| 20.6k| if (c->n_fc > 1) ------------------ | Branch (774:25): [True: 20.6k, False: 0] ------------------ 775| 20.6k| atomic_store(f->out_cdf.progress, error ? TILE_ERROR : 1); ------------------ | Branch (775:25): [True: 10.8k, False: 9.85k] ------------------ 776| 20.6k| } 777| 642k| if (atomic_fetch_sub(&f->task_thread.task_counter, 1) - 1 == 0 && ------------------ | Branch (777:21): [True: 8.58k, False: 633k] ------------------ 778| 642k| atomic_load(&f->task_thread.done[0]) && ------------------ | Branch (778:21): [True: 8.58k, False: 0] ------------------ 779| 8.58k| (!uses_2pass || atomic_load(&f->task_thread.done[1]))) ------------------ | Branch (779:22): [True: 0, False: 8.58k] | Branch (779:37): [True: 8.58k, False: 0] ------------------ 780| 8.58k| { 781| 8.58k| error = atomic_load(&f->task_thread.error); 782| 8.58k| dav1d_decode_frame_exit(f, error == 1 ? DAV1D_ERR(EINVAL) : ------------------ | | 58| 8.58k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (782:48): [True: 8.58k, False: 0] ------------------ 783| 8.58k| error ? DAV1D_ERR(ENOMEM) : 0); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (783:45): [True: 0, False: 0] ------------------ 784| 8.58k| f->n_tile_data = 0; 785| 8.58k| pthread_cond_signal(&f->task_thread.cond); 786| 8.58k| } 787| 642k| assert(atomic_load(&f->task_thread.task_counter) >= 0); ------------------ | Branch (787:17): [True: 628k, False: 14.1k] ------------------ 788| 628k| if (!atomic_fetch_or(&ttd->cond_signaled, 1)) ------------------ | Branch (788:21): [True: 108k, False: 519k] ------------------ 789| 108k| pthread_cond_signal(&ttd->cond); 790| 628k| } 791| 4.47M| continue; 792| 16.5M| } 793| 4.47M| case DAV1D_TASK_TYPE_DEBLOCK_COLS: ------------------ | Branch (793:9): [True: 2.03M, False: 31.0M] ------------------ 794| 2.03M| if (!atomic_load(&f->task_thread.error)) ------------------ | Branch (794:17): [True: 1.22M, False: 810k] ------------------ 795| 1.22M| f->bd_fn.filter_sbrow_deblock_cols(f, sby); 796| 2.03M| if (ensure_progress(ttd, f, t, DAV1D_TASK_TYPE_DEBLOCK_ROWS, ------------------ | Branch (796:17): [True: 18.4k, False: 2.01M] ------------------ 797| 2.03M| &f->frame_thread.deblock_progress, 798| 2.03M| &t->deblock_progress)) continue; 799| | // fall-through 800| 7.40M| case DAV1D_TASK_TYPE_DEBLOCK_ROWS: ------------------ | Branch (800:9): [True: 5.39M, False: 27.7M] ------------------ 801| 7.40M| if (!atomic_load(&f->task_thread.error)) ------------------ | Branch (801:17): [True: 1.49M, False: 5.90M] ------------------ 802| 1.49M| f->bd_fn.filter_sbrow_deblock_rows(f, sby); 803| | // signal deblock progress 804| 7.40M| if (f->frame_hdr->loopfilter.level_y[0] || ------------------ | Branch (804:17): [True: 1.72M, False: 5.68M] ------------------ 805| 5.68M| f->frame_hdr->loopfilter.level_y[1]) ------------------ | Branch (805:17): [True: 306k, False: 5.37M] ------------------ 806| 2.03M| { 807| 2.03M| error = atomic_load(&f->task_thread.error); 808| 2.03M| atomic_store(&f->frame_thread.deblock_progress, ------------------ | Branch (808:17): [True: 810k, False: 1.22M] ------------------ 809| 2.03M| error ? TILE_ERROR : sby + 1); 810| 2.03M| reset_task_cur_async(ttd, t->frame_idx, c->n_fc); 811| 2.03M| if (!atomic_fetch_or(&ttd->cond_signaled, 1)) ------------------ | Branch (811:21): [True: 10.8k, False: 2.01M] ------------------ 812| 10.8k| pthread_cond_signal(&ttd->cond); 813| 5.37M| } else if (f->seq_hdr->cdef || f->lf.restore_planes) { ------------------ | Branch (813:24): [True: 5.31M, False: 66.2k] | Branch (813:44): [True: 66.2k, False: 0] ------------------ 814| 5.37M| atomic_fetch_or(&f->frame_thread.copy_lpf_progress[sby >> 5], 815| 5.37M| 1U << (sby & 31)); 816| | // CDEF needs the top buffer to be saved by lr_copy_lpf of the 817| | // previous sbrow 818| 5.37M| if (sby) { ------------------ | Branch (818:21): [True: 5.19M, False: 178k] ------------------ 819| 5.19M| int prog = atomic_load(&f->frame_thread.copy_lpf_progress[(sby - 1) >> 5]); 820| 5.19M| if (~prog & (1U << ((sby - 1) & 31))) { ------------------ | Branch (820:25): [True: 150k, False: 5.04M] ------------------ 821| 150k| t->type = DAV1D_TASK_TYPE_CDEF; 822| 150k| t->recon_progress = t->deblock_progress = 0; 823| 150k| add_pending(f, t); 824| 150k| pthread_mutex_lock(&ttd->lock); 825| 150k| continue; 826| 150k| } 827| 5.19M| } 828| 5.37M| } 829| | // fall-through 830| 7.40M| case DAV1D_TASK_TYPE_CDEF: ------------------ | Branch (830:9): [True: 150k, False: 32.9M] ------------------ 831| 7.40M| if (f->seq_hdr->cdef) { ------------------ | Branch (831:17): [True: 6.11M, False: 1.29M] ------------------ 832| 6.11M| if (!atomic_load(&f->task_thread.error)) ------------------ | Branch (832:21): [True: 932k, False: 5.18M] ------------------ 833| 932k| f->bd_fn.filter_sbrow_cdef(tc, sby); 834| 6.11M| reset_task_cur_async(ttd, t->frame_idx, c->n_fc); 835| 6.11M| if (!atomic_fetch_or(&ttd->cond_signaled, 1)) ------------------ | Branch (835:21): [True: 31.4k, False: 6.08M] ------------------ 836| 31.4k| pthread_cond_signal(&ttd->cond); 837| 6.11M| } 838| | // fall-through 839| 7.41M| case DAV1D_TASK_TYPE_SUPER_RESOLUTION: ------------------ | Branch (839:9): [True: 3.25k, False: 33.1M] ------------------ 840| 7.41M| if (f->frame_hdr->width[0] != f->frame_hdr->width[1]) ------------------ | Branch (840:17): [True: 587k, False: 6.82M] ------------------ 841| 587k| if (!atomic_load(&f->task_thread.error)) ------------------ | Branch (841:21): [True: 43.9k, False: 543k] ------------------ 842| 43.9k| f->bd_fn.filter_sbrow_resize(f, sby); 843| | // fall-through 844| 7.41M| case DAV1D_TASK_TYPE_LOOP_RESTORATION: ------------------ | Branch (844:9): [True: 0, False: 33.1M] ------------------ 845| 7.41M| if (!atomic_load(&f->task_thread.error) && f->lf.restore_planes) ------------------ | Branch (845:17): [True: 1.49M, False: 5.91M] | Branch (845:56): [True: 217k, False: 1.28M] ------------------ 846| 217k| f->bd_fn.filter_sbrow_lr(f, sby); 847| | // fall-through 848| 7.86M| case DAV1D_TASK_TYPE_RECONSTRUCTION_PROGRESS: ------------------ | Branch (848:9): [True: 450k, False: 32.6M] ------------------ 849| | // dummy to cover for no post-filters 850| 15.7M| case DAV1D_TASK_TYPE_ENTROPY_PROGRESS: ------------------ | Branch (850:9): [True: 7.86M, False: 25.2M] ------------------ 851| | // dummy to convert tile progress to frame 852| 15.7M| break; 853| 0| default: abort(); ------------------ | Branch (853:9): [True: 0, False: 33.1M] ------------------ 854| 33.1M| } 855| | // if task completed [typically LR], signal picture progress as per below 856| 15.7M| const int uses_2pass = c->n_fc > 1; 857| 15.7M| const int sbh = f->sbh; 858| 15.7M| const int sbsz = f->sb_step * 4; 859| 15.7M| if (t->type == DAV1D_TASK_TYPE_ENTROPY_PROGRESS) { ------------------ | Branch (859:13): [True: 7.86M, False: 7.85M] ------------------ 860| 7.86M| error = atomic_load(&f->task_thread.error); 861| 7.86M| const unsigned y = sby + 1 == sbh ? UINT_MAX : (unsigned)(sby + 1) * sbsz; ------------------ | Branch (861:32): [True: 294k, False: 7.56M] ------------------ 862| 7.86M| assert(c->n_fc > 1); ------------------ | Branch (862:13): [True: 7.86M, False: 18.4E] ------------------ 863| 7.86M| if (f->sr_cur.p.data[0] /* upon flush, this can be free'ed already */) ------------------ | Branch (863:17): [True: 7.86M, False: 18.4E] ------------------ 864| 7.86M| atomic_store(&f->sr_cur.progress[0], error ? FRAME_ERROR : y); ------------------ | Branch (864:17): [True: 6.16M, False: 1.70M] ------------------ 865| 7.86M| atomic_store(&f->frame_thread.entropy_progress, ------------------ | Branch (865:13): [True: 6.16M, False: 1.70M] ------------------ 866| 7.86M| error ? TILE_ERROR : sby + 1); 867| 7.86M| if (sby + 1 == sbh) ------------------ | Branch (867:17): [True: 294k, False: 7.56M] ------------------ 868| 7.86M| atomic_store(&f->task_thread.done[1], 1); 869| 7.86M| pthread_mutex_lock(&ttd->lock); 870| 7.86M| const int num_tasks = atomic_fetch_sub(&f->task_thread.task_counter, 1) - 1; 871| 7.86M| if (sby + 1 < sbh && num_tasks) { ------------------ | Branch (871:17): [True: 7.56M, False: 293k] | Branch (871:34): [True: 7.56M, False: 7.63k] ------------------ 872| 7.56M| reset_task_cur(c, ttd, t->frame_idx); 873| 7.56M| continue; 874| 7.56M| } 875| 301k| if (!num_tasks && atomic_load(&f->task_thread.done[0]) && ------------------ | Branch (875:17): [True: 40.3k, False: 260k] | Branch (875:31): [True: 40.3k, False: 0] ------------------ 876| 301k| atomic_load(&f->task_thread.done[1])) ------------------ | Branch (876:17): [True: 40.3k, False: 0] ------------------ 877| 40.3k| { 878| 40.3k| error = atomic_load(&f->task_thread.error); 879| 40.3k| dav1d_decode_frame_exit(f, error == 1 ? DAV1D_ERR(EINVAL) : ------------------ | | 58| 29.1k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (879:44): [True: 29.1k, False: 11.2k] ------------------ 880| 40.3k| error ? DAV1D_ERR(ENOMEM) : 0); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (880:41): [True: 0, False: 11.2k] ------------------ 881| 40.3k| f->n_tile_data = 0; 882| 40.3k| pthread_cond_signal(&f->task_thread.cond); 883| 40.3k| } 884| 301k| reset_task_cur(c, ttd, t->frame_idx); 885| 301k| continue; 886| 7.86M| } 887| | // t->type != DAV1D_TASK_TYPE_ENTROPY_PROGRESS 888| 15.7M| atomic_fetch_or(&f->frame_thread.frame_progress[sby >> 5], 889| 7.85M| 1U << (sby & 31)); 890| 7.85M| pthread_mutex_lock(&f->task_thread.lock); 891| 7.85M| sby = get_frame_progress(c, f); 892| 7.85M| error = atomic_load(&f->task_thread.error); 893| 7.85M| const unsigned y = sby + 1 == sbh ? UINT_MAX : (unsigned)(sby + 1) * sbsz; ------------------ | Branch (893:28): [True: 6.27M, False: 1.58M] ------------------ 894| 7.86M| if (c->n_fc > 1 && f->sr_cur.p.data[0] /* upon flush, this can be free'ed already */) ------------------ | Branch (894:13): [True: 7.86M, False: 18.4E] | Branch (894:28): [True: 7.86M, False: 12] ------------------ 895| 7.86M| atomic_store(&f->sr_cur.progress[1], error ? FRAME_ERROR : y); ------------------ | Branch (895:13): [True: 6.18M, False: 1.67M] ------------------ 896| 7.85M| pthread_mutex_unlock(&f->task_thread.lock); 897| 7.85M| if (sby + 1 == sbh) ------------------ | Branch (897:13): [True: 6.27M, False: 1.58M] ------------------ 898| 7.85M| atomic_store(&f->task_thread.done[0], 1); 899| 7.85M| pthread_mutex_lock(&ttd->lock); 900| 7.85M| const int num_tasks = atomic_fetch_sub(&f->task_thread.task_counter, 1) - 1; 901| 7.85M| if (sby + 1 < sbh && num_tasks) { ------------------ | Branch (901:13): [True: 1.58M, False: 6.27M] | Branch (901:30): [True: 1.58M, False: 4.02k] ------------------ 902| 1.58M| reset_task_cur(c, ttd, t->frame_idx); 903| 1.58M| continue; 904| 1.58M| } 905| 6.27M| if (!num_tasks && atomic_load(&f->task_thread.done[0]) && ------------------ | Branch (905:13): [True: 245k, False: 6.03M] | Branch (905:27): [True: 245k, False: 0] ------------------ 906| 245k| (!uses_2pass || atomic_load(&f->task_thread.done[1]))) ------------------ | Branch (906:14): [True: 0, False: 245k] | Branch (906:29): [True: 245k, False: 0] ------------------ 907| 245k| { 908| 245k| error = atomic_load(&f->task_thread.error); 909| 245k| dav1d_decode_frame_exit(f, error == 1 ? DAV1D_ERR(EINVAL) : ------------------ | | 58| 85.2k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (909:40): [True: 85.2k, False: 160k] ------------------ 910| 245k| error ? DAV1D_ERR(ENOMEM) : 0); ------------------ | | 58| 0|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (910:37): [True: 0, False: 160k] ------------------ 911| 245k| f->n_tile_data = 0; 912| 245k| pthread_cond_signal(&f->task_thread.cond); 913| 245k| } 914| 6.27M| reset_task_cur(c, ttd, t->frame_idx); 915| 6.27M| } 916| 41.3k| pthread_mutex_unlock(&ttd->lock); 917| | 918| | return NULL; 919| 37.6k|} thread_task.c:create_filter_sbrow: 215| 589k|{ 216| 589k| const int has_deblock = f->frame_hdr->loopfilter.level_y[0] || ------------------ | Branch (216:29): [True: 88.9k, False: 500k] ------------------ 217| 500k| f->frame_hdr->loopfilter.level_y[1]; ------------------ | Branch (217:29): [True: 3.85k, False: 496k] ------------------ 218| 589k| const int has_cdef = f->seq_hdr->cdef; 219| 589k| const int has_resize = f->frame_hdr->width[0] != f->frame_hdr->width[1]; 220| 589k| const int has_lr = f->lf.restore_planes; 221| | 222| 589k| Dav1dTask *tasks = f->task_thread.tasks; 223| 589k| const int uses_2pass = f->c->n_fc > 1; 224| 589k| int num_tasks = f->sbh * (1 + uses_2pass); 225| 589k| if (num_tasks > f->task_thread.num_tasks) { ------------------ | Branch (225:9): [True: 16.0k, False: 573k] ------------------ 226| 16.0k| const size_t size = sizeof(Dav1dTask) * num_tasks; 227| 16.0k| tasks = dav1d_realloc(ALLOC_COMMON_CTX, f->task_thread.tasks, size); ------------------ | | 133| 16.0k|#define dav1d_realloc(type, ptr, sz) realloc(ptr, sz) ------------------ 228| 16.0k| if (!tasks) return -1; ------------------ | Branch (228:13): [True: 0, False: 16.0k] ------------------ 229| 16.0k| memset(tasks, 0, size); 230| 16.0k| f->task_thread.tasks = tasks; 231| 16.0k| f->task_thread.num_tasks = num_tasks; 232| 16.0k| } 233| 589k| tasks += f->sbh * (pass & 1); 234| | 235| 589k| if (pass & 1) { ------------------ | Branch (235:9): [True: 294k, False: 294k] ------------------ 236| 294k| f->frame_thread.entropy_progress = 0; 237| 294k| } else { 238| 294k| const int prog_sz = ((f->sbh + 31) & ~31) >> 5; 239| 294k| if (prog_sz > f->frame_thread.prog_sz) { ------------------ | Branch (239:13): [True: 16.3k, False: 278k] ------------------ 240| 16.3k| atomic_uint *const prog = dav1d_realloc(ALLOC_COMMON_CTX, f->frame_thread.frame_progress, ------------------ | | 133| 16.3k|#define dav1d_realloc(type, ptr, sz) realloc(ptr, sz) ------------------ 241| 16.3k| 2 * prog_sz * sizeof(*prog)); 242| 16.3k| if (!prog) return -1; ------------------ | Branch (242:17): [True: 0, False: 16.3k] ------------------ 243| 16.3k| f->frame_thread.frame_progress = prog; 244| 16.3k| f->frame_thread.copy_lpf_progress = prog + prog_sz; 245| 16.3k| } 246| 294k| f->frame_thread.prog_sz = prog_sz; 247| 294k| memset(f->frame_thread.frame_progress, 0, prog_sz * sizeof(atomic_uint)); 248| 294k| memset(f->frame_thread.copy_lpf_progress, 0, prog_sz * sizeof(atomic_uint)); 249| 294k| atomic_store(&f->frame_thread.deblock_progress, 0); 250| 294k| } 251| 589k| f->frame_thread.next_tile_row[pass & 1] = 0; 252| | 253| 589k| Dav1dTask *t = &tasks[0]; 254| 589k| t->sby = 0; 255| 589k| t->recon_progress = 1; 256| 589k| t->deblock_progress = 0; 257| 589k| t->type = pass == 1 ? DAV1D_TASK_TYPE_ENTROPY_PROGRESS : ------------------ | Branch (257:15): [True: 295k, False: 294k] ------------------ 258| 589k| has_deblock ? DAV1D_TASK_TYPE_DEBLOCK_COLS : ------------------ | Branch (258:15): [True: 46.4k, False: 247k] ------------------ 259| 294k| has_cdef || has_lr /* i.e. LR backup */ ? DAV1D_TASK_TYPE_DEBLOCK_ROWS : ------------------ | Branch (259:15): [True: 177k, False: 69.9k] | Branch (259:27): [True: 719, False: 69.2k] ------------------ 260| 247k| has_resize ? DAV1D_TASK_TYPE_SUPER_RESOLUTION : ------------------ | Branch (260:15): [True: 1.12k, False: 67.2k] ------------------ 261| 68.3k| DAV1D_TASK_TYPE_RECONSTRUCTION_PROGRESS; 262| 589k| t->frame_idx = (int)(f - f->c->fc); 263| | 264| 589k| *res_t = t; 265| 589k| return 0; 266| 589k|} thread_task.c:insert_task: 172| 20.7M|{ 173| 20.7M| insert_tasks(f, t, t, cond_signal); 174| 20.7M|} thread_task.c:insert_tasks: 118| 20.7M|{ 119| | // insert task back into task queue 120| 20.7M| Dav1dTask *t_ptr, *prev_t = NULL; 121| 20.7M| for (t_ptr = f->task_thread.task_head; 122| 57.9M| t_ptr; prev_t = t_ptr, t_ptr = t_ptr->next) ------------------ | Branch (122:10): [True: 40.2M, False: 17.6M] ------------------ 123| 40.2M| { 124| | // entropy coding precedes other steps 125| 40.2M| if (t_ptr->type == DAV1D_TASK_TYPE_TILE_ENTROPY) { ------------------ | Branch (125:13): [True: 1.54M, False: 38.7M] ------------------ 126| 1.54M| if (first->type > DAV1D_TASK_TYPE_TILE_ENTROPY) continue; ------------------ | Branch (126:17): [True: 1.19M, False: 348k] ------------------ 127| | // both are entropy 128| 348k| if (first->sby > t_ptr->sby) continue; ------------------ | Branch (128:17): [True: 173k, False: 175k] ------------------ 129| 175k| if (first->sby < t_ptr->sby) { ------------------ | Branch (129:17): [True: 39.8k, False: 135k] ------------------ 130| 39.8k| insert_tasks_between(f, first, last, prev_t, t_ptr, cond_signal); 131| 39.8k| return; 132| 39.8k| } 133| | // same sby 134| 38.7M| } else { 135| 38.7M| if (first->type == DAV1D_TASK_TYPE_TILE_ENTROPY) { ------------------ | Branch (135:17): [True: 1.89M, False: 36.8M] ------------------ 136| 1.89M| insert_tasks_between(f, first, last, prev_t, t_ptr, cond_signal); 137| 1.89M| return; 138| 1.89M| } 139| 36.8M| if (first->sby > t_ptr->sby) continue; ------------------ | Branch (139:17): [True: 26.8M, False: 9.97M] ------------------ 140| 9.97M| if (first->sby < t_ptr->sby) { ------------------ | Branch (140:17): [True: 1.06M, False: 8.91M] ------------------ 141| 1.06M| insert_tasks_between(f, first, last, prev_t, t_ptr, cond_signal); 142| 1.06M| return; 143| 1.06M| } 144| | // same sby 145| 8.91M| if (first->type > t_ptr->type) continue; ------------------ | Branch (145:17): [True: 8.82M, False: 93.5k] ------------------ 146| 93.5k| if (first->type < t_ptr->type) { ------------------ | Branch (146:17): [True: 39.5k, False: 53.9k] ------------------ 147| 39.5k| insert_tasks_between(f, first, last, prev_t, t_ptr, cond_signal); 148| 39.5k| return; 149| 39.5k| } 150| | // same task type 151| 93.5k| } 152| | 153| | // sort by tile-id 154| 40.2M| assert(first->type == DAV1D_TASK_TYPE_TILE_RECONSTRUCTION || ------------------ | Branch (154:9): [True: 53.9k, False: 135k] | Branch (154:9): [True: 135k, False: 0] ------------------ 155| 189k| first->type == DAV1D_TASK_TYPE_TILE_ENTROPY); 156| 189k| assert(first->type == t_ptr->type); ------------------ | Branch (156:9): [True: 189k, False: 0] ------------------ 157| 189k| assert(t_ptr->sby == first->sby); ------------------ | Branch (157:9): [True: 189k, False: 0] ------------------ 158| 189k| const int p = first->type == DAV1D_TASK_TYPE_TILE_ENTROPY; 159| 189k| const int t_tile_idx = (int) (first - f->task_thread.tile_tasks[p]); 160| 189k| const int p_tile_idx = (int) (t_ptr - f->task_thread.tile_tasks[p]); 161| 189k| assert(t_tile_idx != p_tile_idx); ------------------ | Branch (161:9): [True: 189k, False: 0] ------------------ 162| 189k| if (t_tile_idx > p_tile_idx) continue; ------------------ | Branch (162:13): [True: 170k, False: 19.2k] ------------------ 163| 19.2k| insert_tasks_between(f, first, last, prev_t, t_ptr, cond_signal); 164| 19.2k| return; 165| 189k| } 166| | // append at the end 167| 17.6M| insert_tasks_between(f, first, last, prev_t, NULL, cond_signal); 168| 17.6M|} thread_task.c:insert_tasks_between: 102| 20.7M|{ 103| 20.7M| struct TaskThreadData *const ttd = f->task_thread.ttd; 104| 20.7M| if (atomic_load(f->c->flush)) return; ------------------ | Branch (104:9): [True: 47, False: 20.7M] ------------------ 105| 20.7M| assert(!a || a->next == b); ------------------ | Branch (105:5): [True: 2.58M, False: 18.1M] | Branch (105:5): [True: 18.1M, False: 0] ------------------ 106| 20.7M| if (!a) f->task_thread.task_head = first; ------------------ | Branch (106:9): [True: 2.58M, False: 18.1M] ------------------ 107| 18.1M| else a->next = first; 108| 20.7M| if (!b) f->task_thread.task_tail = last; ------------------ | Branch (108:9): [True: 17.6M, False: 3.05M] ------------------ 109| 20.7M| last->next = b; 110| 20.7M| reset_task_cur(f->c, ttd, first->frame_idx); 111| 20.7M| if (cond_signal && !atomic_fetch_or(&ttd->cond_signaled, 1)) ------------------ | Branch (111:9): [True: 356k, False: 20.3M] | Branch (111:24): [True: 118k, False: 237k] ------------------ 112| 118k| pthread_cond_signal(&ttd->cond); 113| 20.7M|} thread_task.c:merge_pending: 206| 22.5M|static inline int merge_pending(const Dav1dContext *const c) { 207| 22.5M| int res = 0; 208| 112M| for (unsigned i = 0; i < c->n_fc; i++) ------------------ | Branch (208:26): [True: 90.2M, False: 22.5M] ------------------ 209| 90.2M| res |= merge_pending_frame(&c->fc[i]); 210| 22.5M| return res; 211| 22.5M|} thread_task.c:delayed_fg_task: 473| 16.2k|{ 474| 16.2k| const Dav1dPicture *const in = ttd->delayed_fg.in; 475| 16.2k| Dav1dPicture *const out = ttd->delayed_fg.out; 476| 16.2k|#if CONFIG_16BPC 477| 16.2k| int off; 478| 16.2k| if (out->p.bpc != 8) ------------------ | Branch (478:9): [True: 3.81k, False: 12.4k] ------------------ 479| 3.81k| off = (out->p.bpc >> 1) - 4; 480| 16.2k|#endif 481| 16.2k| switch (ttd->delayed_fg.type) { 482| 10.1k| case DAV1D_TASK_TYPE_FG_PREP: ------------------ | Branch (482:5): [True: 10.1k, False: 6.16k] ------------------ 483| 10.1k| ttd->delayed_fg.exec = 0; 484| 10.1k| if (atomic_load(&ttd->cond_signaled)) ------------------ | Branch (484:13): [True: 3.08k, False: 7.03k] ------------------ 485| 3.08k| pthread_cond_signal(&ttd->cond); 486| 10.1k| pthread_mutex_unlock(&ttd->lock); 487| 10.1k| switch (out->p.bpc) { 488| 0|#if CONFIG_8BPC 489| 7.70k| case 8: ------------------ | Branch (489:9): [True: 7.70k, False: 2.40k] ------------------ 490| 7.70k| dav1d_prep_grain_8bpc(&c->dsp[0].fg, out, in, 491| 7.70k| ttd->delayed_fg.scaling_8bpc, 492| 7.70k| ttd->delayed_fg.grain_lut_8bpc); 493| 7.70k| break; 494| 0|#endif 495| 0|#if CONFIG_16BPC 496| 2.06k| case 10: ------------------ | Branch (496:9): [True: 2.06k, False: 8.05k] ------------------ 497| 2.40k| case 12: ------------------ | Branch (497:9): [True: 348, False: 9.76k] ------------------ 498| 2.40k| dav1d_prep_grain_16bpc(&c->dsp[off].fg, out, in, 499| 2.40k| ttd->delayed_fg.scaling_16bpc, 500| 2.40k| ttd->delayed_fg.grain_lut_16bpc); 501| 2.40k| break; 502| 0|#endif 503| 0| default: abort(); ------------------ | Branch (503:9): [True: 0, False: 10.1k] ------------------ 504| 10.1k| } 505| 10.1k| ttd->delayed_fg.type = DAV1D_TASK_TYPE_FG_APPLY; 506| 10.1k| pthread_mutex_lock(&ttd->lock); 507| 10.1k| ttd->delayed_fg.exec = 1; 508| | // fall-through 509| 16.2k| case DAV1D_TASK_TYPE_FG_APPLY:; ------------------ | Branch (509:5): [True: 6.16k, False: 10.1k] ------------------ 510| 16.2k| int row = atomic_fetch_add(&ttd->delayed_fg.progress[0], 1); 511| 16.2k| pthread_mutex_unlock(&ttd->lock); 512| 16.2k| int progmax = (out->p.h + FG_BLOCK_SIZE - 1) / FG_BLOCK_SIZE; ------------------ | | 37| 16.2k|#define FG_BLOCK_SIZE 32 ------------------ int progmax = (out->p.h + FG_BLOCK_SIZE - 1) / FG_BLOCK_SIZE; ------------------ | | 37| 16.2k|#define FG_BLOCK_SIZE 32 ------------------ 513| 55.2k| while (row < progmax) { ------------------ | Branch (513:16): [True: 38.9k, False: 16.2k] ------------------ 514| 38.9k| if (row + 1 < progmax) ------------------ | Branch (514:17): [True: 28.8k, False: 10.1k] ------------------ 515| 28.8k| pthread_cond_signal(&ttd->cond); 516| 10.1k| else { 517| 10.1k| pthread_mutex_lock(&ttd->lock); 518| 10.1k| ttd->delayed_fg.exec = 0; 519| 10.1k| pthread_mutex_unlock(&ttd->lock); 520| 10.1k| } 521| 38.9k| switch (out->p.bpc) { 522| 0|#if CONFIG_8BPC 523| 29.8k| case 8: ------------------ | Branch (523:13): [True: 29.8k, False: 9.16k] ------------------ 524| 29.8k| dav1d_apply_grain_row_8bpc(&c->dsp[0].fg, out, in, 525| 29.8k| ttd->delayed_fg.scaling_8bpc, 526| 29.8k| ttd->delayed_fg.grain_lut_8bpc, row); 527| 29.8k| break; 528| 0|#endif 529| 0|#if CONFIG_16BPC 530| 7.79k| case 10: ------------------ | Branch (530:13): [True: 7.79k, False: 31.2k] ------------------ 531| 9.16k| case 12: ------------------ | Branch (531:13): [True: 1.37k, False: 37.6k] ------------------ 532| 9.16k| dav1d_apply_grain_row_16bpc(&c->dsp[off].fg, out, in, 533| 9.16k| ttd->delayed_fg.scaling_16bpc, 534| 9.16k| ttd->delayed_fg.grain_lut_16bpc, row); 535| 9.16k| break; 536| 0|#endif 537| 0| default: abort(); ------------------ | Branch (537:13): [True: 0, False: 38.9k] ------------------ 538| 38.9k| } 539| 38.9k| row = atomic_fetch_add(&ttd->delayed_fg.progress[0], 1); 540| 38.9k| atomic_fetch_add(&ttd->delayed_fg.progress[1], 1); 541| 38.9k| } 542| 16.2k| pthread_mutex_lock(&ttd->lock); 543| 16.2k| ttd->delayed_fg.exec = 0; 544| 16.2k| int done = atomic_fetch_add(&ttd->delayed_fg.progress[1], 1) + 1; 545| 16.2k| progmax = atomic_load(&ttd->delayed_fg.progress[0]); 546| | // signal for completion only once the last runner reaches this 547| 16.2k| if (done >= progmax) { ------------------ | Branch (547:13): [True: 10.1k, False: 6.08k] ------------------ 548| 10.1k| ttd->delayed_fg.finished = 1; 549| 10.1k| pthread_cond_signal(&ttd->delayed_fg.cond); 550| 10.1k| } 551| 16.2k| break; 552| 0| default: abort(); ------------------ | Branch (552:5): [True: 0, False: 16.2k] ------------------ 553| 16.2k| } 554| 16.2k|} thread_task.c:merge_pending_frame: 188| 113M|static inline int merge_pending_frame(Dav1dFrameContext *const f) { 189| 113M| int const merge = atomic_load(&f->task_thread.pending_tasks.merge); 190| 113M| if (merge) { ------------------ | Branch (190:9): [True: 4.20M, False: 109M] ------------------ 191| 4.20M| pthread_mutex_lock(&f->task_thread.pending_tasks.lock); 192| 4.20M| Dav1dTask *t = f->task_thread.pending_tasks.head; 193| 4.20M| f->task_thread.pending_tasks.head = NULL; 194| 4.20M| f->task_thread.pending_tasks.tail = NULL; 195| 4.20M| atomic_store(&f->task_thread.pending_tasks.merge, 0); 196| 4.20M| pthread_mutex_unlock(&f->task_thread.pending_tasks.lock); 197| 9.44M| while (t) { ------------------ | Branch (197:16): [True: 5.23M, False: 4.20M] ------------------ 198| 5.23M| Dav1dTask *const tmp = t->next; 199| 5.23M| insert_task(f, t, 0); 200| 5.23M| t = tmp; 201| 5.23M| } 202| 4.20M| } 203| 113M| return merge; 204| 113M|} thread_task.c:check_tile: 395| 20.6M|{ 396| 20.6M| const int tp = t->type == DAV1D_TASK_TYPE_TILE_ENTROPY; 397| 20.6M| const int tile_idx = (int)(t - f->task_thread.tile_tasks[tp]); 398| 20.6M| Dav1dTileState *const ts = &f->ts[tile_idx]; 399| 20.6M| const int p1 = atomic_load(&ts->progress[tp]); 400| 20.6M| if (p1 < t->sby) return 1; ------------------ | Branch (400:9): [True: 3.83M, False: 16.8M] ------------------ 401| 16.8M| int error = p1 == TILE_ERROR; ------------------ | | 36| 16.8M|#define TILE_ERROR (INT_MAX - 1) ------------------ 402| 16.8M| error |= atomic_fetch_or(&f->task_thread.error, error); 403| 16.8M| if (!error && frame_mt && !tp) { ------------------ | Branch (403:9): [True: 4.51M, False: 12.3M] | Branch (403:19): [True: 4.51M, False: 0] | Branch (403:31): [True: 2.32M, False: 2.18M] ------------------ 404| 2.32M| const int p2 = atomic_load(&ts->progress[1]); 405| 2.32M| if (p2 <= t->sby) return 1; ------------------ | Branch (405:13): [True: 244k, False: 2.08M] ------------------ 406| 2.08M| error = p2 == TILE_ERROR; ------------------ | | 36| 2.08M|#define TILE_ERROR (INT_MAX - 1) ------------------ 407| 2.08M| error |= atomic_fetch_or(&f->task_thread.error, error); 408| 2.08M| } 409| 16.5M| if (!error && frame_mt && !IS_KEY_OR_INTRA(f->frame_hdr)) { ------------------ | | 43| 4.26M| (!IS_INTER_OR_SWITCH(frame_header)) | | ------------------ | | | | 36| 4.26M| ((frame_header)->frame_type & 1) | | ------------------ ------------------ | Branch (409:9): [True: 4.26M, False: 12.3M] | Branch (409:19): [True: 4.26M, False: 0] | Branch (409:31): [True: 3.19M, False: 1.06M] ------------------ 410| | // check reference state 411| 3.19M| const Dav1dThreadPicture *p = &f->sr_cur; 412| 3.19M| const int ss_ver = p->p.p.layout == DAV1D_PIXEL_LAYOUT_I420; 413| 3.19M| const unsigned p_b = (t->sby + 1) << (f->sb_shift + 2); 414| 3.19M| const int tile_sby = t->sby - (ts->tiling.row_start >> f->sb_shift); 415| 3.19M| const int (*const lowest_px)[2] = ts->lowest_pixel[tile_sby]; 416| 24.7M| for (int n = t->deps_skip; n < 7; n++, t->deps_skip++) { ------------------ | Branch (416:36): [True: 21.7M, False: 3.08M] ------------------ 417| 21.7M| unsigned lowest; 418| 21.7M| if (tp) { ------------------ | Branch (418:17): [True: 11.0M, False: 10.6M] ------------------ 419| | // if temporal mv refs are disabled, we only need this 420| | // for the primary ref; if segmentation is disabled, we 421| | // don't even need that 422| 11.0M| lowest = p_b; 423| 11.0M| } else { 424| | // +8 is postfilter-induced delay 425| 10.6M| const int y = lowest_px[n][0] == INT_MIN ? INT_MIN : ------------------ | Branch (425:31): [True: 9.02M, False: 1.63M] ------------------ 426| 10.6M| lowest_px[n][0] + 8; 427| 10.6M| const int uv = lowest_px[n][1] == INT_MIN ? INT_MIN : ------------------ | Branch (427:32): [True: 10.5M, False: 157k] ------------------ 428| 10.6M| lowest_px[n][1] * (1 << ss_ver) + 8; 429| 10.6M| const int max = imax(y, uv); 430| 10.6M| if (max == INT_MIN) continue; ------------------ | Branch (430:21): [True: 9.02M, False: 1.63M] ------------------ 431| 1.63M| lowest = iclip(max, 1, f->refp[n].p.p.h); 432| 1.63M| } 433| 12.6M| const unsigned p3 = atomic_load(&f->refp[n].progress[!tp]); 434| 12.6M| if (p3 < lowest) return 1; ------------------ | Branch (434:17): [True: 115k, False: 12.5M] ------------------ 435| 12.6M| atomic_fetch_or(&f->task_thread.error, p3 == FRAME_ERROR); 436| 12.5M| } 437| 3.19M| } 438| 16.4M| return 0; 439| 16.5M|} thread_task.c:reset_task_cur: 50| 38.9M|{ 51| 38.9M| const unsigned first = atomic_load(&ttd->first); 52| 38.9M| unsigned reset_frame_idx = atomic_exchange(&ttd->reset_task_cur, UINT_MAX); 53| 38.9M| if (reset_frame_idx < first) { ------------------ | Branch (53:9): [True: 0, False: 38.9M] ------------------ 54| 0| if (frame_idx == UINT_MAX) return 0; ------------------ | Branch (54:13): [True: 0, False: 0] ------------------ 55| 0| reset_frame_idx = UINT_MAX; 56| 0| } 57| 38.9M| if (!ttd->cur && c->fc[first].task_thread.task_cur_prev == NULL) ------------------ | Branch (57:9): [True: 26.5M, False: 12.4M] | Branch (57:22): [True: 24.3M, False: 2.22M] ------------------ 58| 24.3M| return 0; 59| 14.6M| if (reset_frame_idx != UINT_MAX) { ------------------ | Branch (59:9): [True: 1.98M, False: 12.6M] ------------------ 60| 1.98M| if (frame_idx == UINT_MAX) { ------------------ | Branch (60:13): [True: 37.7k, False: 1.94M] ------------------ 61| 37.7k| if (reset_frame_idx > first + ttd->cur) ------------------ | Branch (61:17): [True: 361, False: 37.3k] ------------------ 62| 361| return 0; 63| 37.3k| ttd->cur = reset_frame_idx - first; 64| 37.3k| goto cur_found; 65| 37.7k| } 66| 12.6M| } else if (frame_idx == UINT_MAX) ------------------ | Branch (66:16): [True: 1.47M, False: 11.2M] ------------------ 67| 1.47M| return 0; 68| 13.1M| if (frame_idx < first) frame_idx += c->n_fc; ------------------ | Branch (68:9): [True: 3.50M, False: 9.64M] ------------------ 69| 13.1M| const unsigned min_frame_idx = umin(reset_frame_idx, frame_idx); 70| 13.1M| const unsigned cur_frame_idx = first + ttd->cur; 71| 13.1M| if (ttd->cur < c->n_fc && cur_frame_idx < min_frame_idx) ------------------ | Branch (71:9): [True: 12.5M, False: 644k] | Branch (71:31): [True: 798k, False: 11.7M] ------------------ 72| 798k| return 0; 73| 13.0M| for (ttd->cur = min_frame_idx - first; ttd->cur < c->n_fc; ttd->cur++) ------------------ | Branch (73:44): [True: 12.9M, False: 136k] ------------------ 74| 12.9M| if (c->fc[(first + ttd->cur) % c->n_fc].task_thread.task_head) ------------------ | Branch (74:13): [True: 12.2M, False: 691k] ------------------ 75| 12.2M| break; 76| 12.3M|cur_found: 77| 48.4M| for (unsigned i = ttd->cur; i < c->n_fc; i++) ------------------ | Branch (77:33): [True: 36.0M, False: 12.3M] ------------------ 78| 36.0M| c->fc[(first + i) % c->n_fc].task_thread.task_cur_prev = NULL; 79| 12.3M| return 1; 80| 12.3M|} thread_task.c:abort_frame: 459| 60.1k|static inline void abort_frame(Dav1dFrameContext *const f, const int error) { 460| 60.1k| atomic_store(&f->task_thread.error, error == DAV1D_ERR(EINVAL) ? 1 : -1); ------------------ | Branch (460:5): [True: 46.5k, False: 13.5k] ------------------ 461| 60.1k| atomic_store(&f->task_thread.task_counter, 0); 462| 60.1k| atomic_store(&f->task_thread.done[0], 1); 463| 60.1k| atomic_store(&f->task_thread.done[1], 1); 464| 60.1k| atomic_store(&f->sr_cur.progress[0], FRAME_ERROR); 465| | atomic_store(&f->sr_cur.progress[1], FRAME_ERROR); 466| 60.1k| dav1d_decode_frame_exit(f, error); 467| 60.1k| f->n_tile_data = 0; 468| 60.1k| pthread_cond_signal(&f->task_thread.cond); 469| 60.1k|} thread_task.c:add_pending: 176| 4.01M|static inline void add_pending(Dav1dFrameContext *const f, Dav1dTask *const t) { 177| 4.01M| pthread_mutex_lock(&f->task_thread.pending_tasks.lock); 178| 4.01M| t->next = NULL; 179| 4.01M| if (!f->task_thread.pending_tasks.head) ------------------ | Branch (179:9): [True: 3.90M, False: 109k] ------------------ 180| 3.90M| f->task_thread.pending_tasks.head = t; 181| 109k| else 182| 109k| f->task_thread.pending_tasks.tail->next = t; 183| 4.01M| f->task_thread.pending_tasks.tail = t; 184| | atomic_store(&f->task_thread.pending_tasks.merge, 1); 185| 4.01M| pthread_mutex_unlock(&f->task_thread.pending_tasks.lock); 186| 4.01M|} thread_task.c:reset_task_cur_async: 84| 20.2M|{ 85| 20.2M| const unsigned first = atomic_load(&ttd->first); 86| 20.2M| if (frame_idx < first) frame_idx += n_frames; ------------------ | Branch (86:9): [True: 3.56M, False: 16.6M] ------------------ 87| 20.2M| unsigned last_idx = frame_idx; 88| 20.4M| do { 89| 20.4M| frame_idx = last_idx; 90| 20.4M| last_idx = atomic_exchange(&ttd->reset_task_cur, frame_idx); 91| 20.4M| } while (last_idx < frame_idx); ------------------ | Branch (91:14): [True: 151k, False: 20.2M] ------------------ 92| 20.2M| if (frame_idx == first && atomic_load(&ttd->first) != first) { ------------------ | Branch (92:9): [True: 7.61M, False: 12.6M] | Branch (92:31): [True: 0, False: 7.61M] ------------------ 93| 0| unsigned expected = frame_idx; 94| | atomic_compare_exchange_strong(&ttd->reset_task_cur, &expected, UINT_MAX); 95| 0| } 96| 20.2M|} thread_task.c:ensure_progress: 378| 2.03M|{ 379| | // deblock_rows (non-LR portion) depends on deblock of previous sbrow, 380| | // so ensure that completed. if not, re-add to task-queue; else, fall-through 381| 2.03M| int p1 = atomic_load(state); 382| 2.03M| if (p1 < t->sby) { ------------------ | Branch (382:9): [True: 18.4k, False: 2.01M] ------------------ 383| 18.4k| t->type = type; 384| 18.4k| t->recon_progress = t->deblock_progress = 0; 385| 18.4k| *target = t->sby; 386| 18.4k| add_pending(f, t); 387| 18.4k| pthread_mutex_lock(&ttd->lock); 388| 18.4k| return 1; 389| 18.4k| } 390| 2.01M| return 0; 391| 2.03M|} thread_task.c:get_frame_progress: 443| 7.86M|{ 444| 7.86M| unsigned frame_prog = c->n_fc > 1 ? atomic_load(&f->sr_cur.progress[1]) : 0; ------------------ | Branch (444:27): [True: 7.86M, False: 16] ------------------ 445| 7.86M| if (frame_prog >= FRAME_ERROR) ------------------ | | 35| 7.86M|#define FRAME_ERROR (UINT_MAX - 1) ------------------ | Branch (445:9): [True: 6.06M, False: 1.79M] ------------------ 446| 6.06M| return f->sbh - 1; 447| 1.79M| int idx = frame_prog >> (f->sb_shift + 7); 448| 1.79M| int prog; 449| 1.84M| do { 450| 1.84M| atomic_uint *state = &f->frame_thread.frame_progress[idx]; 451| 1.84M| const unsigned val = ~atomic_load(state); 452| 1.84M| prog = val ? ctz(val) : 32; ------------------ | Branch (452:16): [True: 1.79M, False: 46.3k] ------------------ 453| 1.84M| if (prog != 32) break; ------------------ | Branch (453:13): [True: 1.79M, False: 46.3k] ------------------ 454| 46.3k| prog = 0; 455| 46.3k| } while (++idx < f->frame_thread.prog_sz); ------------------ | Branch (455:14): [True: 43.6k, False: 2.77k] ------------------ 456| 1.79M| return ((idx << 5) | prog) - 1; 457| 7.86M|} dav1d_get_shear_params: 80| 312k|int dav1d_get_shear_params(Dav1dWarpedMotionParams *const wm) { 81| 312k| const int32_t *const mat = wm->matrix; 82| | 83| 312k| if (mat[2] <= 0) return 1; ------------------ | Branch (83:9): [True: 0, False: 312k] ------------------ 84| | 85| 312k| wm->u.p.alpha = iclip_wmp(mat[2] - 0x10000); 86| 312k| wm->u.p.beta = iclip_wmp(mat[3]); 87| | 88| 312k| int shift; 89| 312k| const int y = apply_sign(resolve_divisor_32(abs(mat[2]), &shift), mat[2]); 90| 312k| const int64_t v1 = ((int64_t) mat[4] * 0x10000) * y; 91| 312k| const int rnd = (1 << shift) >> 1; 92| 312k| wm->u.p.gamma = iclip_wmp(apply_sign64((int) ((llabs(v1) + rnd) >> shift), v1)); 93| 312k| const int64_t v2 = ((int64_t) mat[3] * mat[4]) * y; 94| 312k| wm->u.p.delta = iclip_wmp(mat[5] - 95| 312k| apply_sign64((int) ((llabs(v2) + rnd) >> shift), v2) - 96| 312k| 0x10000); 97| | 98| 312k| return (4 * abs(wm->u.p.alpha) + 7 * abs(wm->u.p.beta) >= 0x10000) || ------------------ | Branch (98:12): [True: 8.31k, False: 304k] ------------------ 99| 304k| (4 * abs(wm->u.p.gamma) + 4 * abs(wm->u.p.delta) >= 0x10000); ------------------ | Branch (99:12): [True: 1.71k, False: 302k] ------------------ 100| 312k|} dav1d_set_affine_mv2d: 136| 124k|{ 137| 124k| int32_t *const mat = wm->matrix; 138| 124k| const int rsuy = 2 * bh4 - 1; 139| 124k| const int rsux = 2 * bw4 - 1; 140| 124k| const int isuy = by4 * 4 + rsuy; 141| 124k| const int isux = bx4 * 4 + rsux; 142| | 143| 124k| mat[0] = iclip(mv.x * 0x2000 - (isux * (mat[2] - 0x10000) + isuy * mat[3]), 144| 124k| -0x800000, 0x7fffff); 145| 124k| mat[1] = iclip(mv.y * 0x2000 - (isux * mat[4] + isuy * (mat[5] - 0x10000)), 146| 124k| -0x800000, 0x7fffff); 147| 124k|} dav1d_find_affine_int: 153| 169k|{ 154| 169k| int32_t *const mat = wm->matrix; 155| 169k| int a[2][2] = { { 0, 0 }, { 0, 0 } }; 156| 169k| int bx[2] = { 0, 0 }; 157| 169k| int by[2] = { 0, 0 }; 158| 169k| const int rsuy = 2 * bh4 - 1; 159| 169k| const int rsux = 2 * bw4 - 1; 160| 169k| const int suy = rsuy * 8; 161| 169k| const int sux = rsux * 8; 162| 169k| const int duy = suy + mv.y; 163| 169k| const int dux = sux + mv.x; 164| 169k| const int isuy = by4 * 4 + rsuy; 165| 169k| const int isux = bx4 * 4 + rsux; 166| | 167| 493k| for (int i = 0; i < np; i++) { ------------------ | Branch (167:21): [True: 323k, False: 169k] ------------------ 168| 323k| const int dx = pts[i][1][0] - dux; 169| 323k| const int dy = pts[i][1][1] - duy; 170| 323k| const int sx = pts[i][0][0] - sux; 171| 323k| const int sy = pts[i][0][1] - suy; 172| 323k| if (abs(sx - dx) < 256 && abs(sy - dy) < 256) { ------------------ | Branch (172:13): [True: 319k, False: 4.20k] | Branch (172:35): [True: 315k, False: 4.04k] ------------------ 173| 315k| a[0][0] += ((sx * sx) >> 2) + sx * 2 + 8; 174| 315k| a[0][1] += ((sx * sy) >> 2) + sx + sy + 4; 175| 315k| a[1][1] += ((sy * sy) >> 2) + sy * 2 + 8; 176| 315k| bx[0] += ((sx * dx) >> 2) + sx + dx + 8; 177| 315k| bx[1] += ((sy * dx) >> 2) + sy + dx + 4; 178| 315k| by[0] += ((sx * dy) >> 2) + sx + dy + 4; 179| 315k| by[1] += ((sy * dy) >> 2) + sy + dy + 8; 180| 315k| } 181| 323k| } 182| | 183| | // compute determinant of a 184| 169k| const int64_t det = (int64_t) a[0][0] * a[1][1] - (int64_t) a[0][1] * a[0][1]; 185| 169k| if (det == 0) return 1; ------------------ | Branch (185:9): [True: 8.25k, False: 161k] ------------------ 186| 161k| int shift, idet = apply_sign64(resolve_divisor_64(llabs(det), &shift), det); 187| 161k| shift -= 16; 188| 161k| if (shift < 0) { ------------------ | Branch (188:9): [True: 0, False: 161k] ------------------ 189| 0| idet <<= -shift; 190| 0| shift = 0; 191| 0| } 192| | 193| | // solve the least-squares 194| 161k| mat[2] = get_mult_shift_diag((int64_t) a[1][1] * bx[0] - 195| 161k| (int64_t) a[0][1] * bx[1], idet, shift); 196| 161k| mat[3] = get_mult_shift_ndiag((int64_t) a[0][0] * bx[1] - 197| 161k| (int64_t) a[0][1] * bx[0], idet, shift); 198| 161k| mat[4] = get_mult_shift_ndiag((int64_t) a[1][1] * by[0] - 199| 161k| (int64_t) a[0][1] * by[1], idet, shift); 200| 161k| mat[5] = get_mult_shift_diag((int64_t) a[0][0] * by[1] - 201| 161k| (int64_t) a[0][1] * by[0], idet, shift); 202| | 203| 161k| mat[0] = iclip(mv.x * 0x2000 - (isux * (mat[2] - 0x10000) + isuy * mat[3]), 204| 161k| -0x800000, 0x7fffff); 205| 161k| mat[1] = iclip(mv.y * 0x2000 - (isux * mat[4] + isuy * (mat[5] - 0x10000)), 206| 161k| -0x800000, 0x7fffff); 207| | 208| 161k| return 0; 209| 169k|} warpmv.c:iclip_wmp: 63| 1.24M|static inline int iclip_wmp(const int v) { 64| 1.24M| const int cv = iclip(v, INT16_MIN, INT16_MAX); 65| | 66| 1.24M| return apply_sign((abs(cv) + 32) >> 6, cv) * (1 << 6); 67| 1.24M|} warpmv.c:resolve_divisor_32: 69| 312k|static inline int resolve_divisor_32(const unsigned d, int *const shift) { 70| 312k| *shift = ulog2(d); 71| 312k| const int e = d - (1 << *shift); 72| 312k| const int f = *shift > 8 ? (e + (1 << (*shift - 9))) >> (*shift - 8) : ------------------ | Branch (72:19): [True: 312k, False: 9] ------------------ 73| 312k| e << (8 - *shift); 74| 312k| assert(f <= 256); ------------------ | Branch (74:5): [True: 311k, False: 26] ------------------ 75| 311k| *shift += 14; 76| | // Use f as lookup into the precomputed table of multipliers 77| 311k| return div_lut[f]; 78| 312k|} warpmv.c:resolve_divisor_64: 102| 161k|static int resolve_divisor_64(const uint64_t d, int *const shift) { 103| 161k| *shift = u64log2(d); 104| 161k| const int64_t e = d - (1LL << *shift); 105| 161k| const int64_t f = *shift > 8 ? (e + (1LL << (*shift - 9))) >> (*shift - 8) : ------------------ | Branch (105:23): [True: 161k, False: 18.4E] ------------------ 106| 18.4E| e << (8 - *shift); 107| 161k| assert(f <= 256); ------------------ | Branch (107:5): [True: 161k, False: 18.4E] ------------------ 108| 161k| *shift += 14; 109| | // Use f as lookup into the precomputed table of multipliers 110| 161k| return div_lut[f]; 111| 161k|} warpmv.c:get_mult_shift_diag: 125| 322k|{ 126| 322k| const int64_t v1 = px * idet; 127| 322k| const int v2 = apply_sign64((int) ((llabs(v1) + 128| 322k| ((1LL << shift) >> 1)) >> shift), 129| 322k| v1); 130| 322k| return iclip(v2, 0xe001, 0x11fff); 131| 322k|} warpmv.c:get_mult_shift_ndiag: 115| 322k|{ 116| 322k| const int64_t v1 = px * idet; 117| 322k| const int v2 = apply_sign64((int) ((llabs(v1) + 118| 322k| ((1LL << shift) >> 1)) >> shift), 119| 322k| v1); 120| 322k| return iclip(v2, -0x1fff, 0x1fff); 121| 322k|} dav1d_init_ii_wedge_masks: 207| 1|COLD void dav1d_init_ii_wedge_masks(void) { 208| | // This function is guaranteed to be called only once 209| | 210| 1| enum WedgeMasterLineType { 211| 1| WEDGE_MASTER_LINE_ODD, 212| 1| WEDGE_MASTER_LINE_EVEN, 213| 1| WEDGE_MASTER_LINE_VERT, 214| 1| N_WEDGE_MASTER_LINES, 215| 1| }; 216| 1| static const uint8_t wedge_master_border[N_WEDGE_MASTER_LINES][8] = { 217| 1| [WEDGE_MASTER_LINE_ODD] = { 1, 2, 6, 18, 37, 53, 60, 63 }, 218| 1| [WEDGE_MASTER_LINE_EVEN] = { 1, 4, 11, 27, 46, 58, 62, 63 }, 219| 1| [WEDGE_MASTER_LINE_VERT] = { 0, 2, 7, 21, 43, 57, 62, 64 }, 220| 1| }; 221| 1| uint8_t master[6][64 * 64]; 222| | 223| | // create master templates 224| 65| for (int y = 0, off = 0; y < 64; y++, off += 64) ------------------ | Branch (224:30): [True: 64, False: 1] ------------------ 225| 64| insert_border(&master[WEDGE_VERTICAL][off], 226| 64| wedge_master_border[WEDGE_MASTER_LINE_VERT], 32); 227| 33| for (int y = 0, off = 0, ctr = 48; y < 64; y += 2, off += 128, ctr--) ------------------ | Branch (227:40): [True: 32, False: 1] ------------------ 228| 32| { 229| 32| insert_border(&master[WEDGE_OBLIQUE63][off], 230| 32| wedge_master_border[WEDGE_MASTER_LINE_EVEN], ctr); 231| 32| insert_border(&master[WEDGE_OBLIQUE63][off + 64], 232| 32| wedge_master_border[WEDGE_MASTER_LINE_ODD], ctr - 1); 233| 32| } 234| | 235| 1| transpose(master[WEDGE_OBLIQUE27], master[WEDGE_OBLIQUE63]); 236| 1| transpose(master[WEDGE_HORIZONTAL], master[WEDGE_VERTICAL]); 237| 1| hflip(master[WEDGE_OBLIQUE117], master[WEDGE_OBLIQUE63]); 238| 1| hflip(master[WEDGE_OBLIQUE153], master[WEDGE_OBLIQUE27]); 239| | 240| 1|#define fill(w, h, sz_422, sz_420, hvsw, signs) \ 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ 242| 1| master, wedge_codebook_16_##hvsw, \ 243| 1| dav1d_masks.wedge_444_##w##x##h, \ 244| 1| dav1d_masks.wedge_422_##sz_422, \ 245| 1| dav1d_masks.wedge_420_##sz_420, signs) 246| | 247| 1| fill(32, 32, 16x32, 16x16, heqw, 0x7bfb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 248| 1| fill(32, 16, 16x16, 16x8, hltw, 0x7beb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 249| 1| fill(32, 8, 16x8, 16x4, hltw, 0x6beb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 250| 1| fill(16, 32, 8x32, 8x16, hgtw, 0x7beb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 251| 1| fill(16, 16, 8x16, 8x8, heqw, 0x7bfb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 252| 1| fill(16, 8, 8x8, 8x4, hltw, 0x7beb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 253| 1| fill( 8, 32, 4x32, 4x16, hgtw, 0x7aeb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 254| 1| fill( 8, 16, 4x16, 4x8, hgtw, 0x7beb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 255| 1| fill( 8, 8, 4x8, 4x4, heqw, 0x7bfb); ------------------ | | 241| 1| fill2d_16x2(w, h, BS_##w##x##h - BS_32x32, \ | | 242| 1| master, wedge_codebook_16_##hvsw, \ | | 243| 1| dav1d_masks.wedge_444_##w##x##h, \ | | 244| 1| dav1d_masks.wedge_422_##sz_422, \ | | 245| 1| dav1d_masks.wedge_420_##sz_420, signs) ------------------ 256| 1|#undef fill 257| | 258| 1| memset(dav1d_masks.ii_dc, 32, 32 * 32); 259| 4| for (int c = 0; c < 3; c++) { ------------------ | Branch (259:21): [True: 3, False: 1] ------------------ 260| 3| dav1d_masks.offsets[c][BS_32x32-BS_32x32].ii[II_DC_PRED] = 261| 3| dav1d_masks.offsets[c][BS_32x16-BS_32x32].ii[II_DC_PRED] = 262| 3| dav1d_masks.offsets[c][BS_16x32-BS_32x32].ii[II_DC_PRED] = 263| 3| dav1d_masks.offsets[c][BS_16x16-BS_32x32].ii[II_DC_PRED] = 264| 3| dav1d_masks.offsets[c][BS_16x8 -BS_32x32].ii[II_DC_PRED] = 265| 3| dav1d_masks.offsets[c][BS_8x16 -BS_32x32].ii[II_DC_PRED] = 266| 3| dav1d_masks.offsets[c][BS_8x8 -BS_32x32].ii[II_DC_PRED] = 267| 3| MASK_OFFSET(dav1d_masks.ii_dc); ------------------ | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) ------------------ 268| 3| } 269| | 270| 1|#define BUILD_NONDC_II_MASKS(w, h, step) \ 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) 272| | 273| 1|#define ASSIGN_NONDC_II_OFFSET(bs, w444, h444, w422, h422, w420, h420) \ 274| 1| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ 275| 1| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ 276| 1| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ 277| 1| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ 278| 1| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ 279| 1| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) 280| | 281| 1| BUILD_NONDC_II_MASKS(32, 32, 1); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 282| 1| BUILD_NONDC_II_MASKS(16, 32, 1); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 283| 1| BUILD_NONDC_II_MASKS(16, 16, 2); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 284| 1| BUILD_NONDC_II_MASKS( 8, 32, 1); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 285| 1| BUILD_NONDC_II_MASKS( 8, 16, 2); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 286| 1| BUILD_NONDC_II_MASKS( 8, 8, 4); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 287| 1| BUILD_NONDC_II_MASKS( 4, 16, 2); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 288| 1| BUILD_NONDC_II_MASKS( 4, 8, 4); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 289| 1| BUILD_NONDC_II_MASKS( 4, 4, 8); ------------------ | | 271| 1| build_nondc_ii_masks(dav1d_masks.ii_nondc_##w##x##h, w, h, step) ------------------ 290| 4| for (int p = 0; p < 3; p++) { ------------------ | Branch (290:21): [True: 3, False: 1] ------------------ 291| 3| ASSIGN_NONDC_II_OFFSET(BS_32x32, 32, 32, 16, 32, 16, 16); ------------------ | | 274| 3| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ | | 275| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 276| 3| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ | | 277| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 278| 3| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ | | 279| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ ------------------ 292| 3| ASSIGN_NONDC_II_OFFSET(BS_32x16, 32, 32, 16, 16, 16, 16); ------------------ | | 274| 3| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ | | 275| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 276| 3| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ | | 277| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 278| 3| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ | | 279| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ ------------------ 293| 3| ASSIGN_NONDC_II_OFFSET(BS_16x32, 16, 32, 8, 32, 8, 16); ------------------ | | 274| 3| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ | | 275| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 276| 3| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ | | 277| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 278| 3| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ | | 279| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ ------------------ 294| 3| ASSIGN_NONDC_II_OFFSET(BS_16x16, 16, 16, 8, 16, 8, 8); ------------------ | | 274| 3| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ | | 275| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 276| 3| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ | | 277| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 278| 3| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ | | 279| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ ------------------ 295| 3| ASSIGN_NONDC_II_OFFSET(BS_16x8, 16, 16, 8, 8, 8, 8); ------------------ | | 274| 3| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ | | 275| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 276| 3| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ | | 277| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 278| 3| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ | | 279| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ ------------------ 296| 3| ASSIGN_NONDC_II_OFFSET(BS_8x16, 8, 16, 4, 16, 4, 8); ------------------ | | 274| 3| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ | | 275| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 276| 3| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ | | 277| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 278| 3| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ | | 279| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ ------------------ 297| 3| ASSIGN_NONDC_II_OFFSET(BS_8x8, 8, 8, 4, 8, 4, 4); ------------------ | | 274| 3| dav1d_masks.offsets[0][bs-BS_32x32].ii[p + 1] = \ | | 275| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w444##x##h444[p*w444*h444]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 276| 3| dav1d_masks.offsets[1][bs-BS_32x32].ii[p + 1] = \ | | 277| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w422##x##h422[p*w422*h422]); \ | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ | | 278| 3| dav1d_masks.offsets[2][bs-BS_32x32].ii[p + 1] = \ | | 279| 3| MASK_OFFSET(&dav1d_masks.ii_nondc_##w420##x##h420[p*w420*h420]) | | ------------------ | | | | 129| 3|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) | | ------------------ ------------------ 298| 3| } 299| 1|} wedge.c:insert_border: 90| 128|{ 91| 128| if (ctr > 4) memset(dst, 0, ctr - 4); ------------------ | Branch (91:9): [True: 128, False: 0] ------------------ 92| 128| memcpy(dst + imax(ctr, 4) - 4, src + imax(4 - ctr, 0), imin(64 - ctr, 8)); 93| 128| if (ctr < 64 - 4) ------------------ | Branch (93:9): [True: 128, False: 0] ------------------ 94| 128| memset(dst + ctr + 4, 64, 64 - 4 - ctr); 95| 128|} wedge.c:transpose: 97| 2|static void transpose(uint8_t *const dst, const uint8_t *const src) { 98| 130| for (int y = 0, y_off = 0; y < 64; y++, y_off += 64) ------------------ | Branch (98:32): [True: 128, False: 2] ------------------ 99| 8.32k| for (int x = 0, x_off = 0; x < 64; x++, x_off += 64) ------------------ | Branch (99:36): [True: 8.19k, False: 128] ------------------ 100| 8.19k| dst[x_off + y] = src[y_off + x]; 101| 2|} wedge.c:hflip: 103| 2|static void hflip(uint8_t *const dst, const uint8_t *const src) { 104| 130| for (int y = 0, y_off = 0; y < 64; y++, y_off += 64) ------------------ | Branch (104:32): [True: 128, False: 2] ------------------ 105| 8.32k| for (int x = 0; x < 64; x++) ------------------ | Branch (105:25): [True: 8.19k, False: 128] ------------------ 106| 8.19k| dst[y_off + 64 - 1 - x] = src[y_off + x]; 107| 2|} wedge.c:fill2d_16x2: 153| 9|{ 154| 9| const int n_stride_444 = (w * h); 155| 9| const int n_stride_422 = n_stride_444 >> 1; 156| 9| const int n_stride_420 = n_stride_444 >> 2; 157| 9| const int sign_stride_422 = 16 * n_stride_422; 158| 9| const int sign_stride_420 = 16 * n_stride_420; 159| | 160| | // assign pointer offsets in lookup table 161| 153| for (int n = 0; n < 16; n++) { ------------------ | Branch (161:21): [True: 144, False: 9] ------------------ 162| 144| const int sign = signs & 1; 163| | 164| 144| copy2d(masks_444, master[cb[n].direction], sign, w, h, 165| 144| 32 - (w * cb[n].x_offset >> 3), 32 - (h * cb[n].y_offset >> 3)); 166| | 167| | // not using !sign is intentional here, since 444 does not require 168| | // any rounding since no chroma subsampling is applied. 169| 144| dav1d_masks.offsets[0][bs].wedge[0][n] = 170| 144| dav1d_masks.offsets[0][bs].wedge[1][n] = MASK_OFFSET(masks_444); ------------------ | | 129| 144|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) ------------------ 171| | 172| 144| dav1d_masks.offsets[1][bs].wedge[0][n] = 173| 144| init_chroma(&masks_422[ sign * sign_stride_422], masks_444, 0, w, h, 0); 174| 144| dav1d_masks.offsets[1][bs].wedge[1][n] = 175| 144| init_chroma(&masks_422[!sign * sign_stride_422], masks_444, 1, w, h, 0); 176| 144| dav1d_masks.offsets[2][bs].wedge[0][n] = 177| 144| init_chroma(&masks_420[ sign * sign_stride_420], masks_444, 0, w, h, 1); 178| 144| dav1d_masks.offsets[2][bs].wedge[1][n] = 179| 144| init_chroma(&masks_420[!sign * sign_stride_420], masks_444, 1, w, h, 1); 180| | 181| 144| signs >>= 1; 182| 144| masks_444 += n_stride_444; 183| 144| masks_422 += n_stride_422; 184| 144| masks_420 += n_stride_420; 185| 144| } 186| 9|} wedge.c:copy2d: 111| 144|{ 112| 144| src += y_off * 64 + x_off; 113| 144| if (sign) { ------------------ | Branch (113:9): [True: 109, False: 35] ------------------ 114| 2.14k| for (int y = 0; y < h; y++) { ------------------ | Branch (114:25): [True: 2.03k, False: 109] ------------------ 115| 40.4k| for (int x = 0; x < w; x++) ------------------ | Branch (115:29): [True: 38.4k, False: 2.03k] ------------------ 116| 38.4k| dst[x] = 64 - src[x]; 117| 2.03k| src += 64; 118| 2.03k| dst += w; 119| 2.03k| } 120| 109| } else { 121| 691| for (int y = 0; y < h; y++) { ------------------ | Branch (121:25): [True: 656, False: 35] ------------------ 122| 656| memcpy(dst, src, w); 123| 656| src += 64; 124| 656| dst += w; 125| 656| } 126| 35| } 127| 144|} wedge.c:init_chroma: 134| 576|{ 135| 576| const uint16_t offset = MASK_OFFSET(chroma); ------------------ | | 129| 576|#define MASK_OFFSET(x) ((uint16_t)(((uintptr_t)(x) - (uintptr_t)&dav1d_masks) >> 3)) ------------------ 136| 8.64k| for (int y = 0; y < h; y += 1 + ss_ver) { ------------------ | Branch (136:21): [True: 8.06k, False: 576] ------------------ 137| 83.3k| for (int x = 0; x < w; x += 2) { ------------------ | Branch (137:25): [True: 75.2k, False: 8.06k] ------------------ 138| 75.2k| int sum = luma[x] + luma[x + 1] + 1; 139| 75.2k| if (ss_ver) sum += luma[w + x] + luma[w + x + 1] + 1; ------------------ | Branch (139:17): [True: 25.0k, False: 50.1k] ------------------ 140| 75.2k| chroma[x >> 1] = (sum - sign) >> (1 + ss_ver); 141| 75.2k| } 142| 8.06k| luma += w << ss_ver; 143| 8.06k| chroma += w >> 1; 144| 8.06k| } 145| 576| return offset; 146| 576|} wedge.c:build_nondc_ii_masks: 190| 9|{ 191| 9| static const uint8_t ii_weights_1d[32] = { 192| 9| 60, 52, 45, 39, 34, 30, 26, 22, 19, 17, 15, 13, 11, 10, 8, 7, 193| 9| 6, 6, 5, 4, 4, 3, 3, 2, 2, 2, 2, 1, 1, 1, 1, 1, 194| 9| }; 195| | 196| 9| uint8_t *const mask_h = &mask_v[w * h]; 197| 9| uint8_t *const mask_sm = &mask_h[w * h]; 198| 173| for (int y = 0, off = 0; y < h; y++, off += w) { ------------------ | Branch (198:30): [True: 164, False: 9] ------------------ 199| 164| memset(&mask_v[off], ii_weights_1d[y * step], w); 200| 2.51k| for (int x = 0; x < w; x++) { ------------------ | Branch (200:25): [True: 2.35k, False: 164] ------------------ 201| 2.35k| mask_sm[off + x] = ii_weights_1d[imin(x, y) * step]; 202| 2.35k| mask_h[off + x] = ii_weights_1d[x * step]; 203| 2.35k| } 204| 164| } 205| 9|} cdef_tmpl.c:cdef_dsp_init_x86: 46| 3.46k|static ALWAYS_INLINE void cdef_dsp_init_x86(Dav1dCdefDSPContext *const c) { 47| 3.46k| const unsigned flags = dav1d_get_cpu_flags(); 48| | 49| 3.46k|#if BITDEPTH == 8 50| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_SSE2)) return; ------------------ | Branch (50:9): [True: 0, False: 3.46k] ------------------ 51| | 52| 3.46k| c->fb[0] = BF(dav1d_cdef_filter_8x8, sse2); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 53| 3.46k| c->fb[1] = BF(dav1d_cdef_filter_4x8, sse2); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 54| 3.46k| c->fb[2] = BF(dav1d_cdef_filter_4x4, sse2); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 55| 3.46k|#endif 56| | 57| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (57:9): [True: 0, False: 3.46k] ------------------ 58| | 59| 3.46k| c->dir = BF(dav1d_cdef_dir, ssse3); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 60| 3.46k| c->fb[0] = BF(dav1d_cdef_filter_8x8, ssse3); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 61| 3.46k| c->fb[1] = BF(dav1d_cdef_filter_4x8, ssse3); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 62| 3.46k| c->fb[2] = BF(dav1d_cdef_filter_4x4, ssse3); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 63| | 64| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_SSE41)) return; ------------------ | Branch (64:9): [True: 0, False: 3.46k] ------------------ 65| | 66| 3.46k| c->dir = BF(dav1d_cdef_dir, sse4); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 67| 3.46k|#if BITDEPTH == 8 68| 3.46k| c->fb[0] = BF(dav1d_cdef_filter_8x8, sse4); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 69| 3.46k| c->fb[1] = BF(dav1d_cdef_filter_4x8, sse4); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 70| 3.46k| c->fb[2] = BF(dav1d_cdef_filter_4x4, sse4); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 71| 3.46k|#endif 72| | 73| 3.46k|#if ARCH_X86_64 74| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (74:9): [True: 0, False: 3.46k] ------------------ 75| | 76| 3.46k| c->dir = BF(dav1d_cdef_dir, avx2); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 77| 3.46k| c->fb[0] = BF(dav1d_cdef_filter_8x8, avx2); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 78| 3.46k| c->fb[1] = BF(dav1d_cdef_filter_4x8, avx2); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 79| 3.46k| c->fb[2] = BF(dav1d_cdef_filter_4x4, avx2); ------------------ | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 80| | 81| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (81:9): [True: 3.46k, False: 0] ------------------ 82| | 83| 0| c->fb[0] = BF(dav1d_cdef_filter_8x8, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 84| 0| c->fb[1] = BF(dav1d_cdef_filter_4x8, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 85| 0| c->fb[2] = BF(dav1d_cdef_filter_4x4, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 86| 0|#endif 87| 0|} dav1d_get_cpu_flags_x86: 47| 1|COLD unsigned dav1d_get_cpu_flags_x86(void) { 48| 1| union { 49| 1| CpuidRegisters r; 50| 1| struct { 51| 1| uint32_t max_leaf; 52| 1| char vendor[12]; 53| 1| }; 54| 1| } cpu; 55| 1| dav1d_cpu_cpuid(&cpu.r, 0, 0); 56| 1| unsigned flags = dav1d_get_default_cpu_flags(); 57| | 58| 1| if (cpu.max_leaf >= 1) { ------------------ | Branch (58:9): [True: 1, False: 0] ------------------ 59| 1| CpuidRegisters r; 60| 1| dav1d_cpu_cpuid(&r, 1, 0); 61| 1| const unsigned family = ((r.eax >> 8) & 0x0f) + ((r.eax >> 20) & 0xff); 62| | 63| 1| if (X(r.edx, 0x06008000)) /* CMOV/SSE/SSE2 */ { ------------------ | | 45| 1|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 1, False: 0] | | ------------------ ------------------ 64| 1| flags |= DAV1D_X86_CPU_FLAG_SSE2; 65| 1| if (X(r.ecx, 0x00000201)) /* SSE3/SSSE3 */ { ------------------ | | 45| 1|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 1, False: 0] | | ------------------ ------------------ 66| 1| flags |= DAV1D_X86_CPU_FLAG_SSSE3; 67| 1| if (X(r.ecx, 0x00080000)) /* SSE4.1 */ ------------------ | | 45| 1|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 1, False: 0] | | ------------------ ------------------ 68| 1| flags |= DAV1D_X86_CPU_FLAG_SSE41; 69| 1| } 70| 1| } 71| 1|#if ARCH_X86_64 72| | /* We only support >128-bit SIMD on x86-64. */ 73| 1| if (X(r.ecx, 0x18000000)) /* OSXSAVE/AVX */ { ------------------ | | 45| 1|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 1, False: 0] | | ------------------ ------------------ 74| 1| const uint64_t xcr0 = dav1d_cpu_xgetbv(0); 75| 1| if (X(xcr0, 0x00000006)) /* XMM/YMM */ { ------------------ | | 45| 1|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 1, False: 0] | | ------------------ ------------------ 76| 1| if (cpu.max_leaf >= 7) { ------------------ | Branch (76:21): [True: 1, False: 0] ------------------ 77| 1| dav1d_cpu_cpuid(&r, 7, 0); 78| 1| if (X(r.ebx, 0x00000128)) /* BMI1/BMI2/AVX2 */ { ------------------ | | 45| 1|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 1, False: 0] | | ------------------ ------------------ 79| 1| flags |= DAV1D_X86_CPU_FLAG_AVX2; 80| 1| if (X(xcr0, 0x000000e0)) /* ZMM/OPMASK */ { ------------------ | | 45| 1|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 0, False: 1] | | ------------------ ------------------ 81| 0| if (X(r.ebx, 0xd0230000) && X(r.ecx, 0x00005f42)) ------------------ | | 45| 0|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 0, False: 0] | | ------------------ ------------------ if (X(r.ebx, 0xd0230000) && X(r.ecx, 0x00005f42)) ------------------ | | 45| 0|#define X(reg, mask) (((reg) & (mask)) == (mask)) | | ------------------ | | | Branch (45:22): [True: 0, False: 0] | | ------------------ ------------------ 82| 0| flags |= DAV1D_X86_CPU_FLAG_AVX512ICL; 83| 0| } 84| 1| } 85| 1| } 86| 1| } 87| 1| } 88| 1|#endif 89| 1| if (!memcmp(cpu.vendor, "AuthenticAMD", sizeof(cpu.vendor))) { ------------------ | Branch (89:13): [True: 1, False: 0] ------------------ 90| 1| if ((flags & DAV1D_X86_CPU_FLAG_AVX2) && family <= 0x19) { ------------------ | Branch (90:17): [True: 1, False: 0] | Branch (90:54): [True: 1, False: 0] ------------------ 91| | /* Excavator, Zen, Zen+, Zen 2, Zen 3, Zen 3+, Zen 4 */ 92| 1| flags |= DAV1D_X86_CPU_FLAG_SLOW_GATHER; 93| 1| } 94| 1| } 95| 1| } 96| | 97| 1| return flags; 98| 1|} filmgrain_tmpl.c:film_grain_dsp_init_x86: 45| 8.57k|static ALWAYS_INLINE void film_grain_dsp_init_x86(Dav1dFilmGrainDSPContext *const c) { 46| 8.57k| const unsigned flags = dav1d_get_cpu_flags(); 47| | 48| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (48:9): [True: 0, False: 8.57k] ------------------ 49| | 50| 8.57k| c->generate_grain_y = BF(dav1d_generate_grain_y, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 51| 8.57k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I420 - 1] = BF(dav1d_generate_grain_uv_420, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 52| 8.57k| c->fgy_32x32xn = BF(dav1d_fgy_32x32xn, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 53| 8.57k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] = BF(dav1d_fguv_32x32xn_i420, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 54| 8.57k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I422 - 1] = BF(dav1d_generate_grain_uv_422, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 55| 8.57k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I444 - 1] = BF(dav1d_generate_grain_uv_444, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 56| 8.57k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I422 - 1] = BF(dav1d_fguv_32x32xn_i422, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 57| 8.57k| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I444 - 1] = BF(dav1d_fguv_32x32xn_i444, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 58| | 59| 8.57k|#if ARCH_X86_64 60| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (60:9): [True: 0, False: 8.57k] ------------------ 61| | 62| 8.57k| c->generate_grain_y = BF(dav1d_generate_grain_y, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 63| 8.57k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I420 - 1] = BF(dav1d_generate_grain_uv_420, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 64| 8.57k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I422 - 1] = BF(dav1d_generate_grain_uv_422, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 65| 8.57k| c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I444 - 1] = BF(dav1d_generate_grain_uv_444, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 66| | 67| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_SLOW_GATHER)) { ------------------ | Branch (67:9): [True: 0, False: 8.57k] ------------------ 68| 0| c->fgy_32x32xn = BF(dav1d_fgy_32x32xn, avx2); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 69| 0| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] = BF(dav1d_fguv_32x32xn_i420, avx2); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 70| 0| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I422 - 1] = BF(dav1d_fguv_32x32xn_i422, avx2); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 71| 0| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I444 - 1] = BF(dav1d_fguv_32x32xn_i444, avx2); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 72| 0| } 73| | 74| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (74:9): [True: 8.57k, False: 0] ------------------ 75| | 76| 0| if (BITDEPTH == 8 || !(flags & DAV1D_X86_CPU_FLAG_SLOW_GATHER)) { ------------------ | Branch (76:9): [True: 0, Folded] | Branch (76:26): [True: 0, False: 0] ------------------ 77| 0| c->fgy_32x32xn = BF(dav1d_fgy_32x32xn, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 78| 0| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] = BF(dav1d_fguv_32x32xn_i420, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 79| 0| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I422 - 1] = BF(dav1d_fguv_32x32xn_i422, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 80| 0| c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I444 - 1] = BF(dav1d_fguv_32x32xn_i444, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 81| 0| } 82| 0|#endif 83| 0|} ipred_tmpl.c:intra_pred_dsp_init_x86: 71| 8.57k|static ALWAYS_INLINE void intra_pred_dsp_init_x86(Dav1dIntraPredDSPContext *const c) { 72| 8.57k| const unsigned flags = dav1d_get_cpu_flags(); 73| | 74| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (74:9): [True: 0, False: 8.57k] ------------------ 75| | 76| 8.57k| init_angular_ipred_fn(DC_PRED, ipred_dc, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 77| 8.57k| init_angular_ipred_fn(DC_128_PRED, ipred_dc_128, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 78| 8.57k| init_angular_ipred_fn(TOP_DC_PRED, ipred_dc_top, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 79| 8.57k| init_angular_ipred_fn(LEFT_DC_PRED, ipred_dc_left, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 80| 8.57k| init_angular_ipred_fn(HOR_PRED, ipred_h, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 81| 8.57k| init_angular_ipred_fn(VERT_PRED, ipred_v, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 82| 8.57k| init_angular_ipred_fn(PAETH_PRED, ipred_paeth, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 83| 8.57k| init_angular_ipred_fn(SMOOTH_PRED, ipred_smooth, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 84| 8.57k| init_angular_ipred_fn(SMOOTH_H_PRED, ipred_smooth_h, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 85| 8.57k| init_angular_ipred_fn(SMOOTH_V_PRED, ipred_smooth_v, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 86| 8.57k| init_angular_ipred_fn(Z1_PRED, ipred_z1, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 87| 8.57k| init_angular_ipred_fn(Z2_PRED, ipred_z2, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 88| 8.57k| init_angular_ipred_fn(Z3_PRED, ipred_z3, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 89| 8.57k| init_angular_ipred_fn(FILTER_PRED, ipred_filter, ssse3); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 90| | 91| 8.57k| init_cfl_pred_fn(DC_PRED, ipred_cfl, ssse3); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 92| 8.57k| init_cfl_pred_fn(DC_128_PRED, ipred_cfl_128, ssse3); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 93| 8.57k| init_cfl_pred_fn(TOP_DC_PRED, ipred_cfl_top, ssse3); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 94| 8.57k| init_cfl_pred_fn(LEFT_DC_PRED, ipred_cfl_left, ssse3); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 95| | 96| 8.57k| init_cfl_ac_fn(DAV1D_PIXEL_LAYOUT_I420 - 1, ipred_cfl_ac_420, ssse3); ------------------ | | 43| 8.57k| init_fn(cfl_ac, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 97| 8.57k| init_cfl_ac_fn(DAV1D_PIXEL_LAYOUT_I422 - 1, ipred_cfl_ac_422, ssse3); ------------------ | | 43| 8.57k| init_fn(cfl_ac, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 98| 8.57k| init_cfl_ac_fn(DAV1D_PIXEL_LAYOUT_I444 - 1, ipred_cfl_ac_444, ssse3); ------------------ | | 43| 8.57k| init_fn(cfl_ac, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 99| | 100| 8.57k| c->pal_pred = BF(dav1d_pal_pred, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 101| | 102| 8.57k|#if ARCH_X86_64 103| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (103:9): [True: 0, False: 8.57k] ------------------ 104| | 105| 8.57k| init_angular_ipred_fn(DC_PRED, ipred_dc, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 106| 8.57k| init_angular_ipred_fn(DC_128_PRED, ipred_dc_128, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 107| 8.57k| init_angular_ipred_fn(TOP_DC_PRED, ipred_dc_top, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 108| 8.57k| init_angular_ipred_fn(LEFT_DC_PRED, ipred_dc_left, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 109| 8.57k| init_angular_ipred_fn(HOR_PRED, ipred_h, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 110| 8.57k| init_angular_ipred_fn(VERT_PRED, ipred_v, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 111| 8.57k| init_angular_ipred_fn(PAETH_PRED, ipred_paeth, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 112| 8.57k| init_angular_ipred_fn(SMOOTH_PRED, ipred_smooth, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 113| 8.57k| init_angular_ipred_fn(SMOOTH_H_PRED, ipred_smooth_h, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 114| 8.57k| init_angular_ipred_fn(SMOOTH_V_PRED, ipred_smooth_v, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 115| 8.57k| init_angular_ipred_fn(Z1_PRED, ipred_z1, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 116| 8.57k| init_angular_ipred_fn(Z2_PRED, ipred_z2, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 117| 8.57k| init_angular_ipred_fn(Z3_PRED, ipred_z3, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 118| 8.57k| init_angular_ipred_fn(FILTER_PRED, ipred_filter, avx2); ------------------ | | 39| 8.57k| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 119| | 120| 8.57k| init_cfl_pred_fn(DC_PRED, ipred_cfl, avx2); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 121| 8.57k| init_cfl_pred_fn(DC_128_PRED, ipred_cfl_128, avx2); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 122| 8.57k| init_cfl_pred_fn(TOP_DC_PRED, ipred_cfl_top, avx2); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 123| 8.57k| init_cfl_pred_fn(LEFT_DC_PRED, ipred_cfl_left, avx2); ------------------ | | 41| 8.57k| init_fn(cfl_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 124| | 125| 8.57k| init_cfl_ac_fn(DAV1D_PIXEL_LAYOUT_I420 - 1, ipred_cfl_ac_420, avx2); ------------------ | | 43| 8.57k| init_fn(cfl_ac, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 126| 8.57k| init_cfl_ac_fn(DAV1D_PIXEL_LAYOUT_I422 - 1, ipred_cfl_ac_422, avx2); ------------------ | | 43| 8.57k| init_fn(cfl_ac, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 127| 8.57k| init_cfl_ac_fn(DAV1D_PIXEL_LAYOUT_I444 - 1, ipred_cfl_ac_444, avx2); ------------------ | | 43| 8.57k| init_fn(cfl_ac, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 128| | 129| 8.57k| c->pal_pred = BF(dav1d_pal_pred, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 130| | 131| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (131:9): [True: 8.57k, False: 0] ------------------ 132| | 133| 0|#if BITDEPTH == 8 134| 0| init_angular_ipred_fn(DC_PRED, ipred_dc, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 8.57k| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 135| 0| init_angular_ipred_fn(DC_128_PRED, ipred_dc_128, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 136| 0| init_angular_ipred_fn(TOP_DC_PRED, ipred_dc_top, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 137| 0| init_angular_ipred_fn(LEFT_DC_PRED, ipred_dc_left, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 138| 0| init_angular_ipred_fn(HOR_PRED, ipred_h, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 139| 0| init_angular_ipred_fn(VERT_PRED, ipred_v, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 140| 0| init_angular_ipred_fn(Z2_PRED, ipred_z2, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 141| 0|#endif 142| 0| init_angular_ipred_fn(PAETH_PRED, ipred_paeth, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 143| 0| init_angular_ipred_fn(SMOOTH_PRED, ipred_smooth, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 144| 0| init_angular_ipred_fn(SMOOTH_H_PRED, ipred_smooth_h, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 145| 0| init_angular_ipred_fn(SMOOTH_V_PRED, ipred_smooth_v, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 146| 0| init_angular_ipred_fn(Z1_PRED, ipred_z1, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 147| 0| init_angular_ipred_fn(Z2_PRED, ipred_z2, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 148| 0| init_angular_ipred_fn(Z3_PRED, ipred_z3, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 149| 0| init_angular_ipred_fn(FILTER_PRED, ipred_filter, avx512icl); ------------------ | | 39| 0| init_fn(intra_pred, type, name, suffix) | | ------------------ | | | | 36| 0| c->type0[type1] = BF(dav1d_##name, suffix) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 150| | 151| 0| c->pal_pred = BF(dav1d_pal_pred, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 152| 0|#endif 153| 0|} itx_tmpl.c:itx_dsp_init_x86: 112| 3.46k|{ 113| 3.46k|#define assign_itx_bpc_fn(pfx, w, h, type, type_enum, bpc, ext) \ 114| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ 115| 3.46k| BF_BPC(dav1d_inv_txfm_add_##type##_##w##x##h, bpc, ext) 116| | 117| 3.46k|#define assign_itx1_bpc_fn(pfx, w, h, bpc, ext) \ 118| 3.46k| assign_itx_bpc_fn(pfx, w, h, dct_dct, DCT_DCT, bpc, ext) 119| | 120| 3.46k|#define assign_itx2_bpc_fn(pfx, w, h, bpc, ext) \ 121| 3.46k| assign_itx1_bpc_fn(pfx, w, h, bpc, ext); \ 122| 3.46k| assign_itx_bpc_fn(pfx, w, h, identity_identity, IDTX, bpc, ext) 123| | 124| 3.46k|#define assign_itx12_bpc_fn(pfx, w, h, bpc, ext) \ 125| 3.46k| assign_itx2_bpc_fn(pfx, w, h, bpc, ext); \ 126| 3.46k| assign_itx_bpc_fn(pfx, w, h, dct_adst, ADST_DCT, bpc, ext); \ 127| 3.46k| assign_itx_bpc_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, bpc, ext); \ 128| 3.46k| assign_itx_bpc_fn(pfx, w, h, dct_identity, H_DCT, bpc, ext); \ 129| 3.46k| assign_itx_bpc_fn(pfx, w, h, adst_dct, DCT_ADST, bpc, ext); \ 130| 3.46k| assign_itx_bpc_fn(pfx, w, h, adst_adst, ADST_ADST, bpc, ext); \ 131| 3.46k| assign_itx_bpc_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, bpc, ext); \ 132| 3.46k| assign_itx_bpc_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, bpc, ext); \ 133| 3.46k| assign_itx_bpc_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, bpc, ext); \ 134| 3.46k| assign_itx_bpc_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, bpc, ext); \ 135| 3.46k| assign_itx_bpc_fn(pfx, w, h, identity_dct, V_DCT, bpc, ext) 136| | 137| 3.46k|#define assign_itx16_bpc_fn(pfx, w, h, bpc, ext) \ 138| 3.46k| assign_itx12_bpc_fn(pfx, w, h, bpc, ext); \ 139| 3.46k| assign_itx_bpc_fn(pfx, w, h, adst_identity, H_ADST, bpc, ext); \ 140| 3.46k| assign_itx_bpc_fn(pfx, w, h, flipadst_identity, H_FLIPADST, bpc, ext); \ 141| 3.46k| assign_itx_bpc_fn(pfx, w, h, identity_adst, V_ADST, bpc, ext); \ 142| 3.46k| assign_itx_bpc_fn(pfx, w, h, identity_flipadst, V_FLIPADST, bpc, ext) 143| | 144| 3.46k| const unsigned flags = dav1d_get_cpu_flags(); 145| | 146| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_SSE2)) return; ------------------ | Branch (146:9): [True: 0, False: 3.46k] ------------------ 147| | 148| 3.46k| assign_itx_fn(, 4, 4, wht_wht, WHT_WHT, sse2); ------------------ | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | ------------------ | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 149| | 150| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (150:9): [True: 0, False: 3.46k] ------------------ 151| | 152| 3.46k|#if BITDEPTH == 8 153| 3.46k| assign_itx16_fn(, 4, 4, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 154| 3.46k| assign_itx16_fn(R, 4, 8, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 155| 3.46k| assign_itx16_fn(R, 8, 4, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 156| 3.46k| assign_itx16_fn(, 8, 8, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 157| 3.46k| assign_itx16_fn(R, 4, 16, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 158| 3.46k| assign_itx16_fn(R, 16, 4, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 159| 3.46k| assign_itx16_fn(R, 8, 16, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 160| 3.46k| assign_itx16_fn(R, 16, 8, ssse3); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 161| 3.46k| assign_itx12_fn(, 16, 16, ssse3); ------------------ | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | ------------------ | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 162| 3.46k| assign_itx2_fn (R, 8, 32, ssse3); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 163| 3.46k| assign_itx2_fn (R, 32, 8, ssse3); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 164| 3.46k| assign_itx2_fn (R, 16, 32, ssse3); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 165| 3.46k| assign_itx2_fn (R, 32, 16, ssse3); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 166| 3.46k| assign_itx2_fn (, 32, 32, ssse3); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 167| 3.46k| assign_itx1_fn (R, 16, 64, ssse3); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 168| 3.46k| assign_itx1_fn (R, 32, 64, ssse3); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 169| 3.46k| assign_itx1_fn (R, 64, 16, ssse3); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 170| 3.46k| assign_itx1_fn (R, 64, 32, ssse3); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 171| 3.46k| assign_itx1_fn ( , 64, 64, ssse3); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 172| 3.46k| *all_simd = 1; 173| 3.46k|#endif 174| | 175| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_SSE41)) return; ------------------ | Branch (175:9): [True: 0, False: 3.46k] ------------------ 176| | 177| |#if BITDEPTH == 16 178| | if (bpc == 10) { 179| | assign_itx16_fn(, 4, 4, sse4); 180| | assign_itx16_fn(R, 4, 8, sse4); 181| | assign_itx16_fn(R, 4, 16, sse4); 182| | assign_itx16_fn(R, 8, 4, sse4); 183| | assign_itx16_fn(, 8, 8, sse4); 184| | assign_itx16_fn(R, 8, 16, sse4); 185| | assign_itx16_fn(R, 16, 4, sse4); 186| | assign_itx16_fn(R, 16, 8, sse4); 187| | assign_itx12_fn(, 16, 16, sse4); 188| | assign_itx2_fn (R, 8, 32, sse4); 189| | assign_itx2_fn (R, 32, 8, sse4); 190| | assign_itx2_fn (R, 16, 32, sse4); 191| | assign_itx2_fn (R, 32, 16, sse4); 192| | assign_itx2_fn (, 32, 32, sse4); 193| | assign_itx1_fn (R, 16, 64, sse4); 194| | assign_itx1_fn (R, 32, 64, sse4); 195| | assign_itx1_fn (R, 64, 16, sse4); 196| | assign_itx1_fn (R, 64, 32, sse4); 197| | assign_itx1_fn (, 64, 64, sse4); 198| | *all_simd = 1; 199| | } 200| |#endif 201| | 202| 3.46k|#if ARCH_X86_64 203| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (203:9): [True: 0, False: 3.46k] ------------------ 204| | 205| 3.46k| assign_itx_fn(, 4, 4, wht_wht, WHT_WHT, avx2); ------------------ | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | ------------------ | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 206| | 207| 3.46k|#if BITDEPTH == 8 208| 3.46k| assign_itx16_fn( , 4, 4, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 209| 3.46k| assign_itx16_fn(R, 4, 8, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 210| 3.46k| assign_itx16_fn(R, 4, 16, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 211| 3.46k| assign_itx16_fn(R, 8, 4, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 212| 3.46k| assign_itx16_fn( , 8, 8, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 213| 3.46k| assign_itx16_fn(R, 8, 16, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 214| 3.46k| assign_itx2_fn (R, 8, 32, avx2); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 215| 3.46k| assign_itx16_fn(R, 16, 4, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 216| 3.46k| assign_itx16_fn(R, 16, 8, avx2); ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 3.46k| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 3.46k| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 3.46k| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 217| 3.46k| assign_itx12_fn( , 16, 16, avx2); ------------------ | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | ------------------ | | | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 89| 3.46k| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 90| 3.46k| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 91| 3.46k| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 92| 3.46k| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 93| 3.46k| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 94| 3.46k| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 95| 3.46k| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 96| 3.46k| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 97| 3.46k| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 218| 3.46k| assign_itx2_fn (R, 16, 32, avx2); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 219| 3.46k| assign_itx1_fn (R, 16, 64, avx2); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 220| 3.46k| assign_itx2_fn (R, 32, 8, avx2); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 221| 3.46k| assign_itx2_fn (R, 32, 16, avx2); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 222| 3.46k| assign_itx2_fn ( , 32, 32, avx2); ------------------ | | 84| 3.46k| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 223| 3.46k| assign_itx1_fn (R, 32, 64, avx2); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 224| 3.46k| assign_itx1_fn (R, 64, 16, avx2); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 225| 3.46k| assign_itx1_fn (R, 64, 32, avx2); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 226| 3.46k| assign_itx1_fn ( , 64, 64, avx2); ------------------ | | 81| 3.46k| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 227| |#else 228| | if (bpc == 10) { 229| | assign_itx16_bpc_fn( , 4, 4, 10, avx2); 230| | assign_itx16_bpc_fn(R, 4, 8, 10, avx2); 231| | assign_itx16_bpc_fn(R, 4, 16, 10, avx2); 232| | assign_itx16_bpc_fn(R, 8, 4, 10, avx2); 233| | assign_itx16_bpc_fn( , 8, 8, 10, avx2); 234| | assign_itx16_bpc_fn(R, 8, 16, 10, avx2); 235| | assign_itx2_bpc_fn (R, 8, 32, 10, avx2); 236| | assign_itx16_bpc_fn(R, 16, 4, 10, avx2); 237| | assign_itx16_bpc_fn(R, 16, 8, 10, avx2); 238| | assign_itx12_bpc_fn( , 16, 16, 10, avx2); 239| | assign_itx2_bpc_fn (R, 16, 32, 10, avx2); 240| | assign_itx1_bpc_fn (R, 16, 64, 10, avx2); 241| | assign_itx2_bpc_fn (R, 32, 8, 10, avx2); 242| | assign_itx2_bpc_fn (R, 32, 16, 10, avx2); 243| | assign_itx2_bpc_fn ( , 32, 32, 10, avx2); 244| | assign_itx1_bpc_fn (R, 32, 64, 10, avx2); 245| | assign_itx1_bpc_fn (R, 64, 16, 10, avx2); 246| | assign_itx1_bpc_fn (R, 64, 32, 10, avx2); 247| | assign_itx1_bpc_fn ( , 64, 64, 10, avx2); 248| | } else { 249| | assign_itx16_bpc_fn( , 4, 4, 12, avx2); 250| | assign_itx16_bpc_fn(R, 4, 8, 12, avx2); 251| | assign_itx16_bpc_fn(R, 4, 16, 12, avx2); 252| | assign_itx16_bpc_fn(R, 8, 4, 12, avx2); 253| | assign_itx16_bpc_fn( , 8, 8, 12, avx2); 254| | assign_itx16_bpc_fn(R, 8, 16, 12, avx2); 255| | assign_itx2_bpc_fn (R, 8, 32, 12, avx2); 256| | assign_itx16_bpc_fn(R, 16, 4, 12, avx2); 257| | assign_itx16_bpc_fn(R, 16, 8, 12, avx2); 258| | assign_itx12_bpc_fn( , 16, 16, 12, avx2); 259| | assign_itx2_bpc_fn (R, 32, 8, 12, avx2); 260| | assign_itx_bpc_fn(R, 16, 32, identity_identity, IDTX, 12, avx2); 261| | assign_itx_bpc_fn(R, 32, 16, identity_identity, IDTX, 12, avx2); 262| | assign_itx_bpc_fn( , 32, 32, identity_identity, IDTX, 12, avx2); 263| | } 264| |#endif 265| | 266| 3.46k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (266:9): [True: 3.46k, False: 0] ------------------ 267| | 268| 0|#if BITDEPTH == 8 269| 3.46k| assign_itx16_fn( , 4, 4, avx512icl); // no wht ------------------ | | 101| 3.46k| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 3.46k| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 3.46k| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 3.46k| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 3.46k| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 3.46k| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 3.46k| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 3.46k|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 270| 0| assign_itx16_fn(R, 4, 8, avx512icl); ------------------ | | 101| 0| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 0| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 271| 0| assign_itx16_fn(R, 4, 16, avx512icl); ------------------ | | 101| 0| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 0| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 272| 0| assign_itx16_fn(R, 8, 4, avx512icl); ------------------ | | 101| 0| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 0| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 273| 0| assign_itx16_fn( , 8, 8, avx512icl); ------------------ | | 101| 0| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 0| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 274| 0| assign_itx16_fn(R, 8, 16, avx512icl); ------------------ | | 101| 0| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 0| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 275| 0| assign_itx2_fn (R, 8, 32, avx512icl); ------------------ | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 276| 0| assign_itx16_fn(R, 16, 4, avx512icl); ------------------ | | 101| 0| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 0| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 277| 0| assign_itx16_fn(R, 16, 8, avx512icl); ------------------ | | 101| 0| assign_itx12_fn(pfx, w, h, ext); \ | | ------------------ | | | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | | | ------------------ | | | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | | | ------------------ | | | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | | | ------------------ | | | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | | | ------------------ | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 102| 0| assign_itx_fn(pfx, w, h, adst_identity, H_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 103| 0| assign_itx_fn(pfx, w, h, flipadst_identity, H_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 104| 0| assign_itx_fn(pfx, w, h, identity_adst, V_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 105| 0| assign_itx_fn(pfx, w, h, identity_flipadst, V_FLIPADST, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 278| 0| assign_itx12_fn( , 16, 16, avx512icl); ------------------ | | 88| 0| assign_itx2_fn(pfx, w, h, ext); \ | | ------------------ | | | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | | | ------------------ | | | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | | | ------------------ | | | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | | | ------------------ | | | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | | | ------------------ | | | | | | ------------------ | | | | ------------------ | | | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 89| 0| assign_itx_fn(pfx, w, h, dct_adst, ADST_DCT, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 90| 0| assign_itx_fn(pfx, w, h, dct_flipadst, FLIPADST_DCT, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 91| 0| assign_itx_fn(pfx, w, h, dct_identity, H_DCT, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 92| 0| assign_itx_fn(pfx, w, h, adst_dct, DCT_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 93| 0| assign_itx_fn(pfx, w, h, adst_adst, ADST_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 94| 0| assign_itx_fn(pfx, w, h, adst_flipadst, FLIPADST_ADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 95| 0| assign_itx_fn(pfx, w, h, flipadst_dct, DCT_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 96| 0| assign_itx_fn(pfx, w, h, flipadst_adst, ADST_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 97| 0| assign_itx_fn(pfx, w, h, flipadst_flipadst, FLIPADST_FLIPADST, ext); \ | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ | | 98| 0| assign_itx_fn(pfx, w, h, identity_dct, V_DCT, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 279| 0| assign_itx2_fn (R, 16, 32, avx512icl); ------------------ | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 280| 0| assign_itx1_fn (R, 16, 64, avx512icl); ------------------ | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 281| 0| assign_itx2_fn (R, 32, 8, avx512icl); ------------------ | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 282| 0| assign_itx2_fn (R, 32, 16, avx512icl); ------------------ | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 283| 0| assign_itx2_fn ( , 32, 32, avx512icl); ------------------ | | 84| 0| assign_itx1_fn(pfx, w, h, ext); \ | | ------------------ | | | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | | | ------------------ | | | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 85| 0| assign_itx_fn(pfx, w, h, identity_identity, IDTX, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 284| 0| assign_itx1_fn (R, 32, 64, avx512icl); ------------------ | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 285| 0| assign_itx1_fn (R, 64, 16, avx512icl); ------------------ | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 286| 0| assign_itx1_fn (R, 64, 32, avx512icl); ------------------ | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 287| 0| assign_itx1_fn ( , 64, 64, avx512icl); ------------------ | | 81| 0| assign_itx_fn(pfx, w, h, dct_dct, DCT_DCT, ext) | | ------------------ | | | | 77| 0| c->itxfm_add[pfx##TX_##w##X##h][type_enum] = \ | | | | 78| 0| BF(dav1d_inv_txfm_add_##type##_##w##x##h, ext) | | | | ------------------ | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | ------------------ | | ------------------ ------------------ 288| |#else 289| | if (bpc == 10) { 290| | assign_itx16_bpc_fn( , 8, 8, 10, avx512icl); 291| | assign_itx16_bpc_fn(R, 8, 16, 10, avx512icl); 292| | assign_itx2_bpc_fn (R, 8, 32, 10, avx512icl); 293| | assign_itx16_bpc_fn(R, 16, 8, 10, avx512icl); 294| | assign_itx12_bpc_fn( , 16, 16, 10, avx512icl); 295| | assign_itx2_bpc_fn (R, 16, 32, 10, avx512icl); 296| | assign_itx2_bpc_fn (R, 32, 8, 10, avx512icl); 297| | assign_itx2_bpc_fn (R, 32, 16, 10, avx512icl); 298| | assign_itx2_bpc_fn ( , 32, 32, 10, avx512icl); 299| | assign_itx1_bpc_fn (R, 16, 64, 10, avx512icl); 300| | assign_itx1_bpc_fn (R, 32, 64, 10, avx512icl); 301| | assign_itx1_bpc_fn (R, 64, 16, 10, avx512icl); 302| | assign_itx1_bpc_fn (R, 64, 32, 10, avx512icl); 303| | assign_itx1_bpc_fn ( , 64, 64, 10, avx512icl); 304| | } 305| |#endif 306| 0|#endif 307| 0|} loopfilter_tmpl.c:loop_filter_dsp_init_x86: 41| 8.57k|static ALWAYS_INLINE void loop_filter_dsp_init_x86(Dav1dLoopFilterDSPContext *const c) { 42| 8.57k| const unsigned flags = dav1d_get_cpu_flags(); 43| | 44| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (44:9): [True: 0, False: 8.57k] ------------------ 45| | 46| 8.57k| c->loop_filter_sb[0][0] = BF(dav1d_lpf_h_sb_y, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 47| 8.57k| c->loop_filter_sb[0][1] = BF(dav1d_lpf_v_sb_y, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 48| 8.57k| c->loop_filter_sb[1][0] = BF(dav1d_lpf_h_sb_uv, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 49| 8.57k| c->loop_filter_sb[1][1] = BF(dav1d_lpf_v_sb_uv, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 50| | 51| 8.57k|#if ARCH_X86_64 52| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (52:9): [True: 0, False: 8.57k] ------------------ 53| | 54| 8.57k| c->loop_filter_sb[0][0] = BF(dav1d_lpf_h_sb_y, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 55| 8.57k| c->loop_filter_sb[0][1] = BF(dav1d_lpf_v_sb_y, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 56| 8.57k| c->loop_filter_sb[1][0] = BF(dav1d_lpf_h_sb_uv, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 57| 8.57k| c->loop_filter_sb[1][1] = BF(dav1d_lpf_v_sb_uv, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 58| | 59| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (59:9): [True: 8.57k, False: 0] ------------------ 60| | 61| 0| c->loop_filter_sb[0][1] = BF(dav1d_lpf_v_sb_y, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 62| 0| c->loop_filter_sb[1][1] = BF(dav1d_lpf_v_sb_uv, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 63| | 64| 0| if (!(flags & DAV1D_X86_CPU_FLAG_SLOW_GATHER)) { ------------------ | Branch (64:9): [True: 0, False: 0] ------------------ 65| 0| c->loop_filter_sb[0][0] = BF(dav1d_lpf_h_sb_y, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 66| 0| c->loop_filter_sb[1][0] = BF(dav1d_lpf_h_sb_uv, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 67| 0| } 68| 0|#endif 69| 0|} looprestoration_tmpl.c:loop_restoration_dsp_init_x86: 50| 8.57k|static ALWAYS_INLINE void loop_restoration_dsp_init_x86(Dav1dLoopRestorationDSPContext *const c, const int bpc) { 51| 8.57k| const unsigned flags = dav1d_get_cpu_flags(); 52| | 53| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_SSE2)) return; ------------------ | Branch (53:9): [True: 0, False: 8.57k] ------------------ 54| 8.57k|#if BITDEPTH == 8 55| 8.57k| c->wiener[0] = BF(dav1d_wiener_filter7, sse2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 56| 8.57k| c->wiener[1] = BF(dav1d_wiener_filter5, sse2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 57| 8.57k|#endif 58| | 59| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (59:9): [True: 0, False: 8.57k] ------------------ 60| 8.57k| c->wiener[0] = BF(dav1d_wiener_filter7, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 61| 8.57k| c->wiener[1] = BF(dav1d_wiener_filter5, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 62| 8.57k| if (BITDEPTH == 8 || bpc == 10) { ------------------ | Branch (62:9): [True: 3.46k, Folded] | Branch (62:26): [True: 2.11k, False: 2.99k] ------------------ 63| 5.58k| c->sgr[0] = BF(dav1d_sgr_filter_5x5, ssse3); ------------------ | | 52| 5.58k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 64| 5.58k| c->sgr[1] = BF(dav1d_sgr_filter_3x3, ssse3); ------------------ | | 52| 5.58k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 65| 5.58k| c->sgr[2] = BF(dav1d_sgr_filter_mix, ssse3); ------------------ | | 52| 5.58k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 66| 5.58k| } 67| | 68| 8.57k|#if ARCH_X86_64 69| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (69:9): [True: 0, False: 8.57k] ------------------ 70| | 71| 8.57k| c->wiener[0] = BF(dav1d_wiener_filter7, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 72| 8.57k| c->wiener[1] = BF(dav1d_wiener_filter5, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 73| 8.57k| if (BITDEPTH == 8 || bpc == 10) { ------------------ | Branch (73:9): [True: 3.46k, Folded] | Branch (73:26): [True: 2.11k, False: 2.99k] ------------------ 74| 5.58k| c->sgr[0] = BF(dav1d_sgr_filter_5x5, avx2); ------------------ | | 52| 5.58k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 75| 5.58k| c->sgr[1] = BF(dav1d_sgr_filter_3x3, avx2); ------------------ | | 52| 5.58k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 76| 5.58k| c->sgr[2] = BF(dav1d_sgr_filter_mix, avx2); ------------------ | | 52| 5.58k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 77| 5.58k| } 78| | 79| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (79:9): [True: 8.57k, False: 0] ------------------ 80| | 81| 0| c->wiener[0] = BF(dav1d_wiener_filter7, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 82| 0|#if BITDEPTH == 8 83| | /* With VNNI we don't need a 5-tap version. */ 84| 0| c->wiener[1] = c->wiener[0]; 85| |#else 86| | c->wiener[1] = BF(dav1d_wiener_filter5, avx512icl); 87| |#endif 88| 0| if (BITDEPTH == 8 || bpc == 10) { ------------------ | Branch (88:9): [True: 0, Folded] | Branch (88:26): [True: 0, False: 0] ------------------ 89| 0| c->sgr[0] = BF(dav1d_sgr_filter_5x5, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 90| 0| c->sgr[1] = BF(dav1d_sgr_filter_3x3, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 91| 0| c->sgr[2] = BF(dav1d_sgr_filter_mix, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 92| 0| } 93| 0|#endif 94| 0|} mc_tmpl.c:mc_dsp_init_x86: 92| 8.57k|static ALWAYS_INLINE void mc_dsp_init_x86(Dav1dMCDSPContext *const c) { 93| 8.57k| const unsigned flags = dav1d_get_cpu_flags(); 94| | 95| 8.57k| if(!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) ------------------ | Branch (95:8): [True: 0, False: 8.57k] ------------------ 96| 0| return; 97| | 98| 8.57k| init_8tap_fns(ssse3); ------------------ | | 143| 8.57k| init_8tap_gen(mc, opt); \ | | ------------------ | | | | 132| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 133| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 134| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 135| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 136| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 137| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 138| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 139| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 140| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, opt) | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 144| 8.57k| init_8tap_gen(mct, opt) | | ------------------ | | | | 132| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 133| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 134| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 135| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 136| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 137| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 138| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 139| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 140| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, opt) | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ ------------------ 99| | 100| 8.57k| init_mc_fn(FILTER_2D_BILINEAR, bilin, ssse3); ------------------ | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 101| 8.57k| init_mct_fn(FILTER_2D_BILINEAR, bilin, ssse3); ------------------ | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 102| | 103| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_REGULAR, 8tap_scaled_regular, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 104| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_scaled_regular_smooth, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 105| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_scaled_regular_sharp, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 106| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_scaled_smooth_regular, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 107| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SMOOTH, 8tap_scaled_smooth, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 108| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_scaled_smooth_sharp, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 109| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_scaled_sharp_regular, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 110| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_scaled_sharp_smooth, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 111| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SHARP, 8tap_scaled_sharp, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 112| 8.57k| init_mc_scaled_fn(FILTER_2D_BILINEAR, bilin_scaled, ssse3); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 113| | 114| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_REGULAR, 8tap_scaled_regular, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 115| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_scaled_regular_smooth, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 116| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_scaled_regular_sharp, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 117| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_scaled_smooth_regular, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 118| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SMOOTH, 8tap_scaled_smooth, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 119| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_scaled_smooth_sharp, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 120| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_scaled_sharp_regular, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 121| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_scaled_sharp_smooth, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 122| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SHARP, 8tap_scaled_sharp, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 123| 8.57k| init_mct_scaled_fn(FILTER_2D_BILINEAR, bilin_scaled, ssse3); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 124| | 125| 8.57k| c->avg = BF(dav1d_avg, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 126| 8.57k| c->w_avg = BF(dav1d_w_avg, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 127| 8.57k| c->mask = BF(dav1d_mask, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 128| 8.57k| c->w_mask[0] = BF(dav1d_w_mask_444, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 129| 8.57k| c->w_mask[1] = BF(dav1d_w_mask_422, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 130| 8.57k| c->w_mask[2] = BF(dav1d_w_mask_420, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 131| 8.57k| c->blend = BF(dav1d_blend, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 132| 8.57k| c->blend_v = BF(dav1d_blend_v, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 133| 8.57k| c->blend_h = BF(dav1d_blend_h, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 134| 8.57k| c->warp8x8 = BF(dav1d_warp_affine_8x8, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 135| 8.57k| c->warp8x8t = BF(dav1d_warp_affine_8x8t, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 136| 8.57k| c->emu_edge = BF(dav1d_emu_edge, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 137| 8.57k| c->resize = BF(dav1d_resize, ssse3); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 138| | 139| 8.57k| if(!(flags & DAV1D_X86_CPU_FLAG_SSE41)) ------------------ | Branch (139:8): [True: 0, False: 8.57k] ------------------ 140| 0| return; 141| | 142| 8.57k|#if BITDEPTH == 8 143| 8.57k| c->warp8x8 = BF(dav1d_warp_affine_8x8, sse4); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 144| 8.57k| c->warp8x8t = BF(dav1d_warp_affine_8x8t, sse4); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 145| 8.57k|#endif 146| | 147| 8.57k|#if ARCH_X86_64 148| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) ------------------ | Branch (148:9): [True: 0, False: 8.57k] ------------------ 149| 0| return; 150| | 151| 8.57k| init_8tap_fns(avx2); ------------------ | | 143| 8.57k| init_8tap_gen(mc, opt); \ | | ------------------ | | | | 132| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 133| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 134| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 135| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 136| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 137| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 138| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 139| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 140| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, opt) | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 144| 8.57k| init_8tap_gen(mct, opt) | | ------------------ | | | | 132| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 133| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 134| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 135| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 136| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 137| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 138| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 139| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, opt); \ | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 140| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, opt) | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ ------------------ 152| | 153| 8.57k| init_mc_fn(FILTER_2D_BILINEAR, bilin, avx2); ------------------ | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 154| 8.57k| init_mct_fn(FILTER_2D_BILINEAR, bilin, avx2); ------------------ | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 155| | 156| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_REGULAR, 8tap_scaled_regular, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 157| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_scaled_regular_smooth, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 158| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_scaled_regular_sharp, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 159| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_scaled_smooth_regular, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 160| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SMOOTH, 8tap_scaled_smooth, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 161| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_scaled_smooth_sharp, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 162| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_scaled_sharp_regular, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 163| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_scaled_sharp_smooth, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 164| 8.57k| init_mc_scaled_fn(FILTER_2D_8TAP_SHARP, 8tap_scaled_sharp, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 165| 8.57k| init_mc_scaled_fn(FILTER_2D_BILINEAR, bilin_scaled, avx2); ------------------ | | 40| 8.57k| c->mc_scaled[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 166| | 167| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_REGULAR, 8tap_scaled_regular, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 168| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_scaled_regular_smooth, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 169| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_scaled_regular_sharp, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 170| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_scaled_smooth_regular, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 171| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SMOOTH, 8tap_scaled_smooth, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 172| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_scaled_smooth_sharp, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 173| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_scaled_sharp_regular, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 174| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_scaled_sharp_smooth, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 175| 8.57k| init_mct_scaled_fn(FILTER_2D_8TAP_SHARP, 8tap_scaled_sharp, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 176| 8.57k| init_mct_scaled_fn(FILTER_2D_BILINEAR, bilin_scaled, avx2); ------------------ | | 42| 8.57k| c->mct_scaled[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 177| | 178| 8.57k| c->avg = BF(dav1d_avg, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 179| 8.57k| c->w_avg = BF(dav1d_w_avg, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 180| 8.57k| c->mask = BF(dav1d_mask, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 181| 8.57k| c->w_mask[0] = BF(dav1d_w_mask_444, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 182| 8.57k| c->w_mask[1] = BF(dav1d_w_mask_422, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 183| 8.57k| c->w_mask[2] = BF(dav1d_w_mask_420, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 184| 8.57k| c->blend = BF(dav1d_blend, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 185| 8.57k| c->blend_v = BF(dav1d_blend_v, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 186| 8.57k| c->blend_h = BF(dav1d_blend_h, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 187| 8.57k| c->warp8x8 = BF(dav1d_warp_affine_8x8, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 188| 8.57k| c->warp8x8t = BF(dav1d_warp_affine_8x8t, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 189| 8.57k| c->emu_edge = BF(dav1d_emu_edge, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 190| 8.57k| c->resize = BF(dav1d_resize, avx2); ------------------ | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 191| | 192| 8.57k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) ------------------ | Branch (192:9): [True: 8.57k, False: 0] ------------------ 193| 8.57k| return; 194| | 195| 8.57k| init_8tap_fns(avx512icl); ------------------ | | 143| 0| init_8tap_gen(mc, opt); \ | | ------------------ | | | | 132| 0| init_##name##_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, opt); \ | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 133| 0| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, opt); \ | | | | ------------------ | | | | | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 134| 0| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, opt); \ | | | | ------------------ | | | | | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 135| 0| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, opt); \ | | | | ------------------ | | | | | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 136| 0| init_##name##_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, opt); \ | | | | ------------------ | | | | | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 137| 0| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, opt); \ | | | | ------------------ | | | | | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 138| 0| init_##name##_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, opt); \ | | | | ------------------ | | | | | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 139| 0| init_##name##_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, opt); \ | | | | ------------------ | | | | | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 140| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, opt) | | | | ------------------ | | | | | | 36| 8.57k| c->mc[type] = BF(dav1d_put_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ | | 144| 0| init_8tap_gen(mct, opt) | | ------------------ | | | | 132| 0| init_##name##_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 133| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 134| 8.57k| init_##name##_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 135| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 136| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 137| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 138| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 139| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, opt); \ | | | | ------------------ | | | | | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | | | 140| 8.57k| init_##name##_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, opt) | | | | ------------------ | | | | | | 38| 8.57k| c->mct[type] = BF(dav1d_prep_##name, suffix) | | | | | | ------------------ | | | | | | | | 52| 8.57k|#define BF(x, suffix) x##_8bpc_##suffix | | | | | | ------------------ | | | | ------------------ | | ------------------ ------------------ 196| | 197| 0| init_mc_fn (FILTER_2D_BILINEAR, bilin, avx512icl); ------------------ | | 36| 0| c->mc[type] = BF(dav1d_put_##name, suffix) | | ------------------ | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 198| 0| init_mct_fn(FILTER_2D_BILINEAR, bilin, avx512icl); ------------------ | | 38| 0| c->mct[type] = BF(dav1d_prep_##name, suffix) | | ------------------ | | | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix | | ------------------ ------------------ 199| | 200| 0| c->avg = BF(dav1d_avg, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 201| 0| c->w_avg = BF(dav1d_w_avg, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 202| 0| c->mask = BF(dav1d_mask, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 203| 0| c->w_mask[0] = BF(dav1d_w_mask_444, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 204| 0| c->w_mask[1] = BF(dav1d_w_mask_422, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 205| 0| c->w_mask[2] = BF(dav1d_w_mask_420, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 206| 0| c->blend = BF(dav1d_blend, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 207| 0| c->blend_v = BF(dav1d_blend_v, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 208| 0| c->blend_h = BF(dav1d_blend_h, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 209| | 210| 0| if (!(flags & DAV1D_X86_CPU_FLAG_SLOW_GATHER)) { ------------------ | Branch (210:9): [True: 0, False: 0] ------------------ 211| 0| c->resize = BF(dav1d_resize, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 212| 0| c->warp8x8 = BF(dav1d_warp_affine_8x8, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 213| 0| c->warp8x8t = BF(dav1d_warp_affine_8x8t, avx512icl); ------------------ | | 52| 0|#define BF(x, suffix) x##_8bpc_##suffix ------------------ 214| 0| } 215| 0|#endif 216| 0|} msac.c:msac_init_x86: 59| 315k|static ALWAYS_INLINE void msac_init_x86(MsacContext *const s) { 60| 315k| const unsigned flags = dav1d_get_cpu_flags(); 61| | 62| 315k| if (flags & DAV1D_X86_CPU_FLAG_SSE2) { ------------------ | Branch (62:9): [True: 315k, False: 18.4E] ------------------ 63| 315k| s->symbol_adapt16 = dav1d_msac_decode_symbol_adapt16_sse2; 64| 315k| } 65| | 66| 315k| if (flags & DAV1D_X86_CPU_FLAG_AVX2) { ------------------ | Branch (66:9): [True: 315k, False: 18.4E] ------------------ 67| 315k| s->symbol_adapt16 = dav1d_msac_decode_symbol_adapt16_avx2; 68| 315k| } 69| 315k|} pal.c:pal_dsp_init_x86: 34| 9.41k|static ALWAYS_INLINE void pal_dsp_init_x86(Dav1dPalDSPContext *const c) { 35| 9.41k| const unsigned flags = dav1d_get_cpu_flags(); 36| | 37| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (37:9): [True: 0, False: 9.41k] ------------------ 38| | 39| 9.41k| c->pal_idx_finish = dav1d_pal_idx_finish_ssse3; 40| | 41| 9.41k|#if ARCH_X86_64 42| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (42:9): [True: 0, False: 9.41k] ------------------ 43| | 44| 9.41k| c->pal_idx_finish = dav1d_pal_idx_finish_avx2; 45| | 46| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (46:9): [True: 9.41k, False: 0] ------------------ 47| | 48| 0| c->pal_idx_finish = dav1d_pal_idx_finish_avx512icl; 49| 0|#endif 50| 0|} refmvs.c:refmvs_dsp_init_x86: 41| 9.41k|static ALWAYS_INLINE void refmvs_dsp_init_x86(Dav1dRefmvsDSPContext *const c) { 42| 9.41k| const unsigned flags = dav1d_get_cpu_flags(); 43| | 44| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_SSE2)) return; ------------------ | Branch (44:9): [True: 0, False: 9.41k] ------------------ 45| | 46| 9.41k| c->splat_mv = dav1d_splat_mv_sse2; 47| | 48| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return; ------------------ | Branch (48:9): [True: 0, False: 9.41k] ------------------ 49| | 50| 9.41k| c->save_tmvs = dav1d_save_tmvs_ssse3; 51| | 52| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_SSE41)) return; ------------------ | Branch (52:9): [True: 0, False: 9.41k] ------------------ 53| 9.41k|#if ARCH_X86_64 54| 9.41k| c->load_tmvs = dav1d_load_tmvs_sse4; 55| | 56| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return; ------------------ | Branch (56:9): [True: 0, False: 9.41k] ------------------ 57| | 58| 9.41k| c->save_tmvs = dav1d_save_tmvs_avx2; 59| 9.41k| c->splat_mv = dav1d_splat_mv_avx2; 60| | 61| 9.41k| if (!(flags & DAV1D_X86_CPU_FLAG_AVX512ICL)) return; ------------------ | Branch (61:9): [True: 9.41k, False: 0] ------------------ 62| | 63| 0| c->save_tmvs = dav1d_save_tmvs_avx512icl; 64| 0| c->splat_mv = dav1d_splat_mv_avx512icl; 65| 0|#endif 66| 0|} LLVMFuzzerInitialize: 59| 2|int LLVMFuzzerInitialize(int *argc, char ***argv) { 60| 2| int i = 1; 61| 11| for (; i < *argc; i++) { ------------------ | Branch (61:12): [True: 9, False: 2] ------------------ 62| 9| if (!strcmp((*argv)[i], "--cpumask")) { ------------------ | Branch (62:13): [True: 0, False: 9] ------------------ 63| 0| const char * cpumask = (*argv)[i+1]; 64| 0| if (cpumask) { ------------------ | Branch (64:17): [True: 0, False: 0] ------------------ 65| 0| char *end; 66| 0| unsigned res; 67| 0| if (!strncmp(cpumask, "0x", 2)) { ------------------ | Branch (67:21): [True: 0, False: 0] ------------------ 68| 0| cpumask += 2; 69| 0| res = (unsigned) strtoul(cpumask, &end, 16); 70| 0| } else { 71| 0| res = (unsigned) strtoul(cpumask, &end, 0); 72| 0| } 73| 0| if (end != cpumask && !end[0]) { ------------------ | Branch (73:21): [True: 0, False: 0] | Branch (73:39): [True: 0, False: 0] ------------------ 74| 0| dav1d_set_cpu_flags_mask(res); 75| 0| } 76| 0| } 77| 0| break; 78| 0| } 79| 9| } 80| | 81| 2| for (; i < *argc - 2; i++) { ------------------ | Branch (81:12): [True: 0, False: 2] ------------------ 82| 0| (*argv)[i] = (*argv)[i + 2]; 83| 0| } 84| | 85| 2| *argc = i; 86| | 87| 2| return 0; 88| 2|} LLVMFuzzerTestOneInput: 94| 9.42k|{ 95| 9.42k| Dav1dSettings settings = { 0 }; 96| 9.42k| Dav1dContext * ctx = NULL; 97| 9.42k| Dav1dPicture pic; 98| 9.42k| const uint8_t *ptr = data; 99| 9.42k| int have_seq_hdr = 0; 100| 9.42k| int err; 101| | 102| 9.42k| dav1d_version(); 103| | 104| 9.42k| if (size < 32) goto end; ------------------ | Branch (104:9): [True: 8, False: 9.41k] ------------------ 105| |#ifdef DAV1D_ALLOC_FAIL 106| | unsigned h = djb_xor(ptr, 32); 107| | unsigned seed = h; 108| | unsigned probability = h > (RAND_MAX >> 5) ? RAND_MAX >> 5 : h; 109| | int max_frame_delay = (h & 0xf) + 1; 110| | int n_threads = ((h >> 4) & 0x7) + 1; 111| | if (max_frame_delay > 5) max_frame_delay = 1; 112| | if (n_threads > 3) n_threads = 1; 113| |#endif 114| 9.41k| ptr += 32; // skip ivf header 115| | 116| 9.41k| dav1d_default_settings(&settings); 117| | 118| 9.41k|#ifdef DAV1D_MT_FUZZING 119| 9.41k| settings.max_frame_delay = settings.n_threads = 4; 120| |#elif defined(DAV1D_ALLOC_FAIL) 121| | settings.max_frame_delay = max_frame_delay; 122| | settings.n_threads = n_threads; 123| | dav1d_setup_alloc_fail(seed, probability); 124| |#else 125| | settings.max_frame_delay = settings.n_threads = 1; 126| |#endif 127| 9.41k|#if defined(DAV1D_FUZZ_MAX_SIZE) 128| 9.41k| settings.frame_size_limit = DAV1D_FUZZ_MAX_SIZE; ------------------ | | 56| 9.41k|#define DAV1D_FUZZ_MAX_SIZE 4096 * 4096 ------------------ 129| 9.41k|#endif 130| | 131| 9.41k| err = dav1d_open(&ctx, &settings); 132| 9.41k| if (err < 0) goto end; ------------------ | Branch (132:9): [True: 0, False: 9.41k] ------------------ 133| | 134| 429k| while (ptr <= data + size - 12) { ------------------ | Branch (134:12): [True: 423k, False: 6.09k] ------------------ 135| 423k| Dav1dData buf; 136| 423k| uint8_t *p; 137| | 138| 423k| size_t frame_size = r32le(ptr); 139| 423k| ptr += 12; 140| | 141| 423k| if (frame_size > size || ptr > data + size - frame_size) ------------------ | Branch (141:13): [True: 2.69k, False: 420k] | Branch (141:34): [True: 626, False: 419k] ------------------ 142| 3.32k| break; 143| | 144| 419k| if (!frame_size) continue; ------------------ | Branch (144:13): [True: 1.83k, False: 417k] ------------------ 145| | 146| 417k| if (!have_seq_hdr) { ------------------ | Branch (146:13): [True: 15.3k, False: 402k] ------------------ 147| 15.3k| Dav1dSequenceHeader seq; 148| 15.3k| int err = dav1d_parse_sequence_header(&seq, ptr, frame_size); 149| | // skip frames until we see a sequence header 150| 15.3k| if (err != 0) { ------------------ | Branch (150:18): [True: 6.21k, False: 9.13k] ------------------ 151| 6.21k| ptr += frame_size; 152| 6.21k| continue; 153| 6.21k| } 154| 9.13k| have_seq_hdr = 1; 155| 9.13k| } 156| | 157| | // copy frame data to a new buffer to catch reads past the end of input 158| 411k| p = dav1d_data_create(&buf, frame_size); 159| 411k| if (!p) goto cleanup; ------------------ | Branch (159:13): [True: 0, False: 411k] ------------------ 160| 411k| memcpy(p, ptr, frame_size); 161| 411k| ptr += frame_size; 162| | 163| 430k| do { 164| 430k| if ((err = dav1d_send_data(ctx, &buf)) < 0) { ------------------ | Branch (164:17): [True: 59.8k, False: 370k] ------------------ 165| 59.8k| if (err != DAV1D_ERR(EAGAIN)) ------------------ | | 58| 59.8k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (165:21): [True: 39.8k, False: 19.9k] ------------------ 166| 39.8k| break; 167| 59.8k| } 168| 390k| memset(&pic, 0, sizeof(pic)); 169| 390k| err = dav1d_get_picture(ctx, &pic); 170| 390k| if (err == 0) { ------------------ | Branch (170:17): [True: 167k, False: 223k] ------------------ 171| 167k| dav1d_picture_unref(&pic); 172| 223k| } else if (err != DAV1D_ERR(EAGAIN)) { ------------------ | | 58| 223k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (172:24): [True: 176k, False: 46.8k] ------------------ 173| 176k| break; 174| 176k| } 175| 390k| } while (buf.sz > 0); ------------------ | Branch (175:18): [True: 18.5k, False: 195k] ------------------ 176| | 177| 411k| if (buf.sz > 0) ------------------ | Branch (177:13): [True: 41.2k, False: 370k] ------------------ 178| 41.2k| dav1d_data_unref(&buf); 179| 411k| } 180| | 181| 9.41k| memset(&pic, 0, sizeof(pic)); 182| 9.41k| if ((err = dav1d_get_picture(ctx, &pic)) == 0) { ------------------ | Branch (182:9): [True: 5.16k, False: 4.25k] ------------------ 183| | /* Test calling dav1d_picture_unref() after dav1d_close() */ 184| 11.7k| do { 185| 11.7k| Dav1dPicture pic2 = { 0 }; 186| 11.7k| if ((err = dav1d_get_picture(ctx, &pic2)) == 0) ------------------ | Branch (186:17): [True: 4.25k, False: 7.45k] ------------------ 187| 4.25k| dav1d_picture_unref(&pic2); 188| 11.7k| } while (err != DAV1D_ERR(EAGAIN)); ------------------ | | 58| 11.7k|#define DAV1D_ERR(e) (-(e)) ///< Negate POSIX error code. ------------------ | Branch (188:18): [True: 6.54k, False: 5.16k] ------------------ 189| | 190| 5.16k| dav1d_close(&ctx); 191| 5.16k| dav1d_picture_unref(&pic); 192| 5.16k| return 0; 193| 5.16k| } 194| | 195| 4.25k|cleanup: 196| 4.25k| dav1d_close(&ctx); 197| 4.26k|end: 198| 4.26k| return 0; 199| 4.25k|} dav1d_fuzzer.c:r32le: 52| 423k|static unsigned r32le(const uint8_t *const p) { 53| 423k| return ((uint32_t)p[3] << 24U) | (p[2] << 16U) | (p[1] << 8U) | p[0]; 54| 423k|}