/src/aom/aom_dsp/x86/convolve.h
Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | * Copyright (c) 2016, Alliance for Open Media. All rights reserved. |
3 | | * |
4 | | * This source code is subject to the terms of the BSD 2 Clause License and |
5 | | * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License |
6 | | * was not distributed with this source code in the LICENSE file, you can |
7 | | * obtain it at www.aomedia.org/license/software. If the Alliance for Open |
8 | | * Media Patent License 1.0 was not distributed with this source code in the |
9 | | * PATENTS file, you can obtain it at www.aomedia.org/license/patent. |
10 | | */ |
11 | | #ifndef AOM_AOM_DSP_X86_CONVOLVE_H_ |
12 | | #define AOM_AOM_DSP_X86_CONVOLVE_H_ |
13 | | |
14 | | #include <assert.h> |
15 | | |
16 | | #include "config/aom_config.h" |
17 | | #include "config/aom_dsp_rtcd.h" |
18 | | |
19 | | #include "aom/aom_integer.h" |
20 | | #include "aom_ports/mem.h" |
21 | | |
22 | | typedef void filter8_1dfunction(const uint8_t *src_ptr, ptrdiff_t src_pitch, |
23 | | uint8_t *output_ptr, ptrdiff_t out_pitch, |
24 | | uint32_t output_height, const int16_t *filter); |
25 | | |
26 | | #define FUN_CONV_1D(name, step_q4, filter, dir, src_start, avg, opt) \ |
27 | | void aom_convolve8_##name##_##opt( \ |
28 | | const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, \ |
29 | | ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, \ |
30 | 0 | const int16_t *filter_y, int y_step_q4, int w, int h) { \ |
31 | 0 | (void)filter_x; \ |
32 | 0 | (void)x_step_q4; \ |
33 | 0 | (void)filter_y; \ |
34 | 0 | (void)y_step_q4; \ |
35 | 0 | assert((-128 <= filter[3]) && (filter[3] <= 127)); \ |
36 | 0 | assert(step_q4 == 16); \ |
37 | 0 | if (((filter[0] | filter[1] | filter[6] | filter[7]) == 0) && \ |
38 | 0 | (filter[2] | filter[5])) { \ |
39 | 0 | while (w >= 16) { \ |
40 | 0 | aom_filter_block1d16_##dir##4_##avg##opt(src_start, src_stride, dst, \ |
41 | 0 | dst_stride, h, filter); \ |
42 | 0 | src += 16; \ |
43 | 0 | dst += 16; \ |
44 | 0 | w -= 16; \ |
45 | 0 | } \ |
46 | 0 | while (w >= 8) { \ |
47 | 0 | aom_filter_block1d8_##dir##4_##avg##opt(src_start, src_stride, dst, \ |
48 | 0 | dst_stride, h, filter); \ |
49 | 0 | src += 8; \ |
50 | 0 | dst += 8; \ |
51 | 0 | w -= 8; \ |
52 | 0 | } \ |
53 | 0 | while (w >= 4) { \ |
54 | 0 | aom_filter_block1d4_##dir##4_##avg##opt(src_start, src_stride, dst, \ |
55 | 0 | dst_stride, h, filter); \ |
56 | 0 | src += 4; \ |
57 | 0 | dst += 4; \ |
58 | 0 | w -= 4; \ |
59 | 0 | } \ |
60 | 0 | } else if (filter[0] | filter[1] | filter[2]) { \ |
61 | 0 | while (w >= 16) { \ |
62 | 0 | aom_filter_block1d16_##dir##8_##avg##opt(src_start, src_stride, dst, \ |
63 | 0 | dst_stride, h, filter); \ |
64 | 0 | src += 16; \ |
65 | 0 | dst += 16; \ |
66 | 0 | w -= 16; \ |
67 | 0 | } \ |
68 | 0 | while (w >= 8) { \ |
69 | 0 | aom_filter_block1d8_##dir##8_##avg##opt(src_start, src_stride, dst, \ |
70 | 0 | dst_stride, h, filter); \ |
71 | 0 | src += 8; \ |
72 | 0 | dst += 8; \ |
73 | 0 | w -= 8; \ |
74 | 0 | } \ |
75 | 0 | while (w >= 4) { \ |
76 | 0 | aom_filter_block1d4_##dir##8_##avg##opt(src_start, src_stride, dst, \ |
77 | 0 | dst_stride, h, filter); \ |
78 | 0 | src += 4; \ |
79 | 0 | dst += 4; \ |
80 | 0 | w -= 4; \ |
81 | 0 | } \ |
82 | 0 | } else { \ |
83 | 0 | while (w >= 16) { \ |
84 | 0 | aom_filter_block1d16_##dir##2_##avg##opt(src, src_stride, dst, \ |
85 | 0 | dst_stride, h, filter); \ |
86 | 0 | src += 16; \ |
87 | 0 | dst += 16; \ |
88 | 0 | w -= 16; \ |
89 | 0 | } \ |
90 | 0 | while (w >= 8) { \ |
91 | 0 | aom_filter_block1d8_##dir##2_##avg##opt(src, src_stride, dst, \ |
92 | 0 | dst_stride, h, filter); \ |
93 | 0 | src += 8; \ |
94 | 0 | dst += 8; \ |
95 | 0 | w -= 8; \ |
96 | 0 | } \ |
97 | 0 | while (w >= 4) { \ |
98 | 0 | aom_filter_block1d4_##dir##2_##avg##opt(src, src_stride, dst, \ |
99 | 0 | dst_stride, h, filter); \ |
100 | 0 | src += 4; \ |
101 | 0 | dst += 4; \ |
102 | 0 | w -= 4; \ |
103 | 0 | } \ |
104 | 0 | } \ |
105 | 0 | if (w) { \ |
106 | 0 | aom_convolve8_##name##_c(src, src_stride, dst, dst_stride, filter_x, \ |
107 | 0 | x_step_q4, filter_y, y_step_q4, w, h); \ |
108 | 0 | } \ |
109 | 0 | } Unexecuted instantiation: aom_convolve8_horiz_ssse3 Unexecuted instantiation: aom_convolve8_vert_ssse3 Unexecuted instantiation: aom_convolve8_horiz_avx2 Unexecuted instantiation: aom_convolve8_vert_avx2 |
110 | | |
111 | | #if CONFIG_AV1_HIGHBITDEPTH |
112 | | typedef void highbd_filter8_1dfunction(const uint16_t *src_ptr, |
113 | | const ptrdiff_t src_pitch, |
114 | | uint16_t *output_ptr, |
115 | | ptrdiff_t out_pitch, |
116 | | unsigned int output_height, |
117 | | const int16_t *filter, int bd); |
118 | | |
119 | | #define HIGH_FUN_CONV_1D(name, step_q4, filter, dir, src_start, avg, opt) \ |
120 | | void aom_highbd_convolve8_##name##_##opt( \ |
121 | | const uint8_t *src8, ptrdiff_t src_stride, uint8_t *dst8, \ |
122 | | ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, \ |
123 | 0 | const int16_t *filter_y, int y_step_q4, int w, int h, int bd) { \ |
124 | 0 | uint16_t *src = CONVERT_TO_SHORTPTR(src8); \ |
125 | 0 | uint16_t *dst = CONVERT_TO_SHORTPTR(dst8); \ |
126 | 0 | if (step_q4 == 16 && filter[3] != 128) { \ |
127 | 0 | if (((filter[0] | filter[1] | filter[6] | filter[7]) == 0) && \ |
128 | 0 | (filter[2] | filter[5])) { \ |
129 | 0 | while (w >= 16) { \ |
130 | 0 | aom_highbd_filter_block1d16_##dir##4_##avg##opt( \ |
131 | 0 | src_start, src_stride, dst, dst_stride, h, filter, bd); \ |
132 | 0 | src += 16; \ |
133 | 0 | dst += 16; \ |
134 | 0 | w -= 16; \ |
135 | 0 | } \ |
136 | 0 | while (w >= 8) { \ |
137 | 0 | aom_highbd_filter_block1d8_##dir##4_##avg##opt( \ |
138 | 0 | src_start, src_stride, dst, dst_stride, h, filter, bd); \ |
139 | 0 | src += 8; \ |
140 | 0 | dst += 8; \ |
141 | 0 | w -= 8; \ |
142 | 0 | } \ |
143 | 0 | while (w >= 4) { \ |
144 | 0 | aom_highbd_filter_block1d4_##dir##4_##avg##opt( \ |
145 | 0 | src_start, src_stride, dst, dst_stride, h, filter, bd); \ |
146 | 0 | src += 4; \ |
147 | 0 | dst += 4; \ |
148 | 0 | w -= 4; \ |
149 | 0 | } \ |
150 | 0 | } else if (filter[0] | filter[1] | filter[2]) { \ |
151 | 0 | while (w >= 16) { \ |
152 | 0 | aom_highbd_filter_block1d16_##dir##8_##avg##opt( \ |
153 | 0 | src_start, src_stride, dst, dst_stride, h, filter, bd); \ |
154 | 0 | src += 16; \ |
155 | 0 | dst += 16; \ |
156 | 0 | w -= 16; \ |
157 | 0 | } \ |
158 | 0 | while (w >= 8) { \ |
159 | 0 | aom_highbd_filter_block1d8_##dir##8_##avg##opt( \ |
160 | 0 | src_start, src_stride, dst, dst_stride, h, filter, bd); \ |
161 | 0 | src += 8; \ |
162 | 0 | dst += 8; \ |
163 | 0 | w -= 8; \ |
164 | 0 | } \ |
165 | 0 | while (w >= 4) { \ |
166 | 0 | aom_highbd_filter_block1d4_##dir##8_##avg##opt( \ |
167 | 0 | src_start, src_stride, dst, dst_stride, h, filter, bd); \ |
168 | 0 | src += 4; \ |
169 | 0 | dst += 4; \ |
170 | 0 | w -= 4; \ |
171 | 0 | } \ |
172 | 0 | } else { \ |
173 | 0 | while (w >= 16) { \ |
174 | 0 | aom_highbd_filter_block1d16_##dir##2_##avg##opt( \ |
175 | 0 | src, src_stride, dst, dst_stride, h, filter, bd); \ |
176 | 0 | src += 16; \ |
177 | 0 | dst += 16; \ |
178 | 0 | w -= 16; \ |
179 | 0 | } \ |
180 | 0 | while (w >= 8) { \ |
181 | 0 | aom_highbd_filter_block1d8_##dir##2_##avg##opt( \ |
182 | 0 | src, src_stride, dst, dst_stride, h, filter, bd); \ |
183 | 0 | src += 8; \ |
184 | 0 | dst += 8; \ |
185 | 0 | w -= 8; \ |
186 | 0 | } \ |
187 | 0 | while (w >= 4) { \ |
188 | 0 | aom_highbd_filter_block1d4_##dir##2_##avg##opt( \ |
189 | 0 | src, src_stride, dst, dst_stride, h, filter, bd); \ |
190 | 0 | src += 4; \ |
191 | 0 | dst += 4; \ |
192 | 0 | w -= 4; \ |
193 | 0 | } \ |
194 | 0 | } \ |
195 | 0 | } \ |
196 | 0 | if (w) { \ |
197 | 0 | aom_highbd_convolve8_##name##_c( \ |
198 | 0 | CONVERT_TO_BYTEPTR(src), src_stride, CONVERT_TO_BYTEPTR(dst), \ |
199 | 0 | dst_stride, filter_x, x_step_q4, filter_y, y_step_q4, w, h, bd); \ |
200 | 0 | } \ |
201 | 0 | } Unexecuted instantiation: aom_highbd_convolve8_horiz_sse2 Unexecuted instantiation: aom_highbd_convolve8_vert_sse2 Unexecuted instantiation: aom_highbd_convolve8_horiz_avx2 Unexecuted instantiation: aom_highbd_convolve8_vert_avx2 |
202 | | #endif // CONFIG_AV1_HIGHBITDEPTH |
203 | | |
204 | | #endif // AOM_AOM_DSP_X86_CONVOLVE_H_ |