Coverage Report

Created: 2026-11-12 08:20

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/aom/aom_dsp/x86/blend_sse4.h
Line
Count
Source
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
12
#ifndef AOM_AOM_DSP_X86_BLEND_SSE4_H_
13
#define AOM_AOM_DSP_X86_BLEND_SSE4_H_
14
15
#include "aom_dsp/blend.h"
16
#include "aom_dsp/x86/synonyms.h"
17
static const uint8_t g_blend_a64_mask_shuffle[32] = {
18
  0, 2, 4, 6, 8, 10, 12, 14, 1, 3, 5, 7, 9, 11, 13, 15,
19
  0, 2, 4, 6, 8, 10, 12, 14, 1, 3, 5, 7, 9, 11, 13, 15,
20
};
21
22
//////////////////////////////////////////////////////////////////////////////
23
// Common kernels
24
//////////////////////////////////////////////////////////////////////////////
25
26
static inline __m128i blend_4(const uint8_t *src0, const uint8_t *src1,
27
197k
                              const __m128i *v_m0_w, const __m128i *v_m1_w) {
28
197k
  const __m128i v_s0_b = xx_loadl_32(src0);
29
197k
  const __m128i v_s1_b = xx_loadl_32(src1);
30
197k
  const __m128i v_s0_w = _mm_cvtepu8_epi16(v_s0_b);
31
197k
  const __m128i v_s1_w = _mm_cvtepu8_epi16(v_s1_b);
32
33
197k
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, *v_m0_w);
34
197k
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, *v_m1_w);
35
197k
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
36
197k
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
37
38
197k
  return v_res_w;
39
197k
}
Unexecuted instantiation: blend_a64_mask_sse4.c:blend_4
blend_a64_vmask_sse4.c:blend_4
Line
Count
Source
27
197k
                              const __m128i *v_m0_w, const __m128i *v_m1_w) {
28
197k
  const __m128i v_s0_b = xx_loadl_32(src0);
29
197k
  const __m128i v_s1_b = xx_loadl_32(src1);
30
197k
  const __m128i v_s0_w = _mm_cvtepu8_epi16(v_s0_b);
31
197k
  const __m128i v_s1_w = _mm_cvtepu8_epi16(v_s1_b);
32
33
197k
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, *v_m0_w);
34
197k
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, *v_m1_w);
35
197k
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
36
197k
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
37
38
197k
  return v_res_w;
39
197k
}
Unexecuted instantiation: blend_a64_mask_avx2.c:blend_4
40
41
static inline __m128i blend_8(const uint8_t *src0, const uint8_t *src1,
42
3.00M
                              const __m128i *v_m0_w, const __m128i *v_m1_w) {
43
3.00M
  const __m128i v_s0_b = xx_loadl_64(src0);
44
3.00M
  const __m128i v_s1_b = xx_loadl_64(src1);
45
3.00M
  const __m128i v_s0_w = _mm_cvtepu8_epi16(v_s0_b);
46
3.00M
  const __m128i v_s1_w = _mm_cvtepu8_epi16(v_s1_b);
47
48
3.00M
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, *v_m0_w);
49
3.00M
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, *v_m1_w);
50
51
3.00M
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
52
53
3.00M
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
54
55
3.00M
  return v_res_w;
56
3.00M
}
Unexecuted instantiation: blend_a64_mask_sse4.c:blend_8
blend_a64_vmask_sse4.c:blend_8
Line
Count
Source
42
3.00M
                              const __m128i *v_m0_w, const __m128i *v_m1_w) {
43
3.00M
  const __m128i v_s0_b = xx_loadl_64(src0);
44
3.00M
  const __m128i v_s1_b = xx_loadl_64(src1);
45
3.00M
  const __m128i v_s0_w = _mm_cvtepu8_epi16(v_s0_b);
46
3.00M
  const __m128i v_s1_w = _mm_cvtepu8_epi16(v_s1_b);
47
48
3.00M
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, *v_m0_w);
49
3.00M
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, *v_m1_w);
50
51
3.00M
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
52
53
3.00M
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
54
55
3.00M
  return v_res_w;
56
3.00M
}
Unexecuted instantiation: blend_a64_mask_avx2.c:blend_8
57
58
static inline __m128i blend_4_u8(const uint8_t *src0, const uint8_t *src1,
59
                                 const __m128i *v_m0_b, const __m128i *v_m1_b,
60
2.88M
                                 const __m128i *rounding) {
61
2.88M
  const __m128i v_s0_b = xx_loadl_32(src0);
62
2.88M
  const __m128i v_s1_b = xx_loadl_32(src1);
63
64
2.88M
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
65
2.88M
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
66
67
2.88M
  const __m128i v_res_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
68
2.88M
  const __m128i v_res = _mm_packus_epi16(v_res_w, v_res_w);
69
2.88M
  return v_res;
70
2.88M
}
blend_a64_mask_sse4.c:blend_4_u8
Line
Count
Source
60
1.99M
                                 const __m128i *rounding) {
61
1.99M
  const __m128i v_s0_b = xx_loadl_32(src0);
62
1.99M
  const __m128i v_s1_b = xx_loadl_32(src1);
63
64
1.99M
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
65
1.99M
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
66
67
1.99M
  const __m128i v_res_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
68
1.99M
  const __m128i v_res = _mm_packus_epi16(v_res_w, v_res_w);
69
1.99M
  return v_res;
70
1.99M
}
Unexecuted instantiation: blend_a64_vmask_sse4.c:blend_4_u8
blend_a64_mask_avx2.c:blend_4_u8
Line
Count
Source
60
890k
                                 const __m128i *rounding) {
61
890k
  const __m128i v_s0_b = xx_loadl_32(src0);
62
890k
  const __m128i v_s1_b = xx_loadl_32(src1);
63
64
890k
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
65
890k
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
66
67
890k
  const __m128i v_res_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
68
890k
  const __m128i v_res = _mm_packus_epi16(v_res_w, v_res_w);
69
890k
  return v_res;
70
890k
}
71
72
static inline __m128i blend_8_u8(const uint8_t *src0, const uint8_t *src1,
73
                                 const __m128i *v_m0_b, const __m128i *v_m1_b,
74
3.60M
                                 const __m128i *rounding) {
75
3.60M
  const __m128i v_s0_b = xx_loadl_64(src0);
76
3.60M
  const __m128i v_s1_b = xx_loadl_64(src1);
77
78
3.60M
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
79
3.60M
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
80
81
3.60M
  const __m128i v_res_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
82
3.60M
  const __m128i v_res = _mm_packus_epi16(v_res_w, v_res_w);
83
3.60M
  return v_res;
84
3.60M
}
blend_a64_mask_sse4.c:blend_8_u8
Line
Count
Source
74
1.25M
                                 const __m128i *rounding) {
75
1.25M
  const __m128i v_s0_b = xx_loadl_64(src0);
76
1.25M
  const __m128i v_s1_b = xx_loadl_64(src1);
77
78
1.25M
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
79
1.25M
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
80
81
1.25M
  const __m128i v_res_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
82
1.25M
  const __m128i v_res = _mm_packus_epi16(v_res_w, v_res_w);
83
1.25M
  return v_res;
84
1.25M
}
Unexecuted instantiation: blend_a64_vmask_sse4.c:blend_8_u8
blend_a64_mask_avx2.c:blend_8_u8
Line
Count
Source
74
2.34M
                                 const __m128i *rounding) {
75
2.34M
  const __m128i v_s0_b = xx_loadl_64(src0);
76
2.34M
  const __m128i v_s1_b = xx_loadl_64(src1);
77
78
2.34M
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
79
2.34M
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
80
81
2.34M
  const __m128i v_res_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
82
2.34M
  const __m128i v_res = _mm_packus_epi16(v_res_w, v_res_w);
83
2.34M
  return v_res;
84
2.34M
}
85
86
static inline __m128i blend_16_u8(const uint8_t *src0, const uint8_t *src1,
87
                                  const __m128i *v_m0_b, const __m128i *v_m1_b,
88
2.28M
                                  const __m128i *rounding) {
89
2.28M
  const __m128i v_s0_b = xx_loadu_128(src0);
90
2.28M
  const __m128i v_s1_b = xx_loadu_128(src1);
91
92
2.28M
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
93
2.28M
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
94
2.28M
  const __m128i v_p1_w = _mm_maddubs_epi16(_mm_unpackhi_epi8(v_s0_b, v_s1_b),
95
2.28M
                                           _mm_unpackhi_epi8(*v_m0_b, *v_m1_b));
96
97
2.28M
  const __m128i v_res0_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
98
2.28M
  const __m128i v_res1_w = _mm_mulhrs_epi16(v_p1_w, *rounding);
99
2.28M
  const __m128i v_res = _mm_packus_epi16(v_res0_w, v_res1_w);
100
2.28M
  return v_res;
101
2.28M
}
blend_a64_mask_sse4.c:blend_16_u8
Line
Count
Source
88
603k
                                  const __m128i *rounding) {
89
603k
  const __m128i v_s0_b = xx_loadu_128(src0);
90
603k
  const __m128i v_s1_b = xx_loadu_128(src1);
91
92
603k
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
93
603k
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
94
603k
  const __m128i v_p1_w = _mm_maddubs_epi16(_mm_unpackhi_epi8(v_s0_b, v_s1_b),
95
603k
                                           _mm_unpackhi_epi8(*v_m0_b, *v_m1_b));
96
97
603k
  const __m128i v_res0_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
98
603k
  const __m128i v_res1_w = _mm_mulhrs_epi16(v_p1_w, *rounding);
99
603k
  const __m128i v_res = _mm_packus_epi16(v_res0_w, v_res1_w);
100
603k
  return v_res;
101
603k
}
Unexecuted instantiation: blend_a64_vmask_sse4.c:blend_16_u8
blend_a64_mask_avx2.c:blend_16_u8
Line
Count
Source
88
1.68M
                                  const __m128i *rounding) {
89
1.68M
  const __m128i v_s0_b = xx_loadu_128(src0);
90
1.68M
  const __m128i v_s1_b = xx_loadu_128(src1);
91
92
1.68M
  const __m128i v_p0_w = _mm_maddubs_epi16(_mm_unpacklo_epi8(v_s0_b, v_s1_b),
93
1.68M
                                           _mm_unpacklo_epi8(*v_m0_b, *v_m1_b));
94
1.68M
  const __m128i v_p1_w = _mm_maddubs_epi16(_mm_unpackhi_epi8(v_s0_b, v_s1_b),
95
1.68M
                                           _mm_unpackhi_epi8(*v_m0_b, *v_m1_b));
96
97
1.68M
  const __m128i v_res0_w = _mm_mulhrs_epi16(v_p0_w, *rounding);
98
1.68M
  const __m128i v_res1_w = _mm_mulhrs_epi16(v_p1_w, *rounding);
99
1.68M
  const __m128i v_res = _mm_packus_epi16(v_res0_w, v_res1_w);
100
1.68M
  return v_res;
101
1.68M
}
102
103
typedef __m128i (*blend_unit_fn)(const uint16_t *src0, const uint16_t *src1,
104
                                 const __m128i v_m0_w, const __m128i v_m1_w);
105
106
static inline __m128i blend_4_b10(const uint16_t *src0, const uint16_t *src1,
107
2.55M
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
108
2.55M
  const __m128i v_s0_w = xx_loadl_64(src0);
109
2.55M
  const __m128i v_s1_w = xx_loadl_64(src1);
110
111
2.55M
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, v_m0_w);
112
2.55M
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, v_m1_w);
113
114
2.55M
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
115
116
2.55M
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
117
118
2.55M
  return v_res_w;
119
2.55M
}
blend_a64_mask_sse4.c:blend_4_b10
Line
Count
Source
107
2.36M
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
108
2.36M
  const __m128i v_s0_w = xx_loadl_64(src0);
109
2.36M
  const __m128i v_s1_w = xx_loadl_64(src1);
110
111
2.36M
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, v_m0_w);
112
2.36M
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, v_m1_w);
113
114
2.36M
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
115
116
2.36M
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
117
118
2.36M
  return v_res_w;
119
2.36M
}
blend_a64_vmask_sse4.c:blend_4_b10
Line
Count
Source
107
191k
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
108
191k
  const __m128i v_s0_w = xx_loadl_64(src0);
109
191k
  const __m128i v_s1_w = xx_loadl_64(src1);
110
111
191k
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, v_m0_w);
112
191k
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, v_m1_w);
113
114
191k
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
115
116
191k
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
117
118
191k
  return v_res_w;
119
191k
}
Unexecuted instantiation: blend_a64_mask_avx2.c:blend_4_b10
120
121
static inline __m128i blend_8_b10(const uint16_t *src0, const uint16_t *src1,
122
9.85M
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
123
9.85M
  const __m128i v_s0_w = xx_loadu_128(src0);
124
9.85M
  const __m128i v_s1_w = xx_loadu_128(src1);
125
126
9.85M
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, v_m0_w);
127
9.85M
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, v_m1_w);
128
129
9.85M
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
130
131
9.85M
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
132
133
9.85M
  return v_res_w;
134
9.85M
}
blend_a64_mask_sse4.c:blend_8_b10
Line
Count
Source
122
6.80M
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
123
6.80M
  const __m128i v_s0_w = xx_loadu_128(src0);
124
6.80M
  const __m128i v_s1_w = xx_loadu_128(src1);
125
126
6.80M
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, v_m0_w);
127
6.80M
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, v_m1_w);
128
129
6.80M
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
130
131
6.80M
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
132
133
6.80M
  return v_res_w;
134
6.80M
}
blend_a64_vmask_sse4.c:blend_8_b10
Line
Count
Source
122
3.04M
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
123
3.04M
  const __m128i v_s0_w = xx_loadu_128(src0);
124
3.04M
  const __m128i v_s1_w = xx_loadu_128(src1);
125
126
3.04M
  const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, v_m0_w);
127
3.04M
  const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, v_m1_w);
128
129
3.04M
  const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w);
130
131
3.04M
  const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS);
132
133
3.04M
  return v_res_w;
134
3.04M
}
Unexecuted instantiation: blend_a64_mask_avx2.c:blend_8_b10
135
136
static inline __m128i blend_4_b12(const uint16_t *src0, const uint16_t *src1,
137
148k
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
138
148k
  const __m128i v_s0_w = xx_loadl_64(src0);
139
148k
  const __m128i v_s1_w = xx_loadl_64(src1);
140
141
  // Interleave
142
148k
  const __m128i v_m01_w = _mm_unpacklo_epi16(v_m0_w, v_m1_w);
143
148k
  const __m128i v_s01_w = _mm_unpacklo_epi16(v_s0_w, v_s1_w);
144
145
  // Multiply-Add
146
148k
  const __m128i v_sum_d = _mm_madd_epi16(v_s01_w, v_m01_w);
147
148
  // Scale
149
148k
  const __m128i v_ssum_d =
150
148k
      _mm_srli_epi32(v_sum_d, AOM_BLEND_A64_ROUND_BITS - 1);
151
152
  // Pack
153
148k
  const __m128i v_pssum_d = _mm_packs_epi32(v_ssum_d, v_ssum_d);
154
155
  // Round
156
148k
  const __m128i v_res_w = xx_round_epu16(v_pssum_d);
157
158
148k
  return v_res_w;
159
148k
}
blend_a64_mask_sse4.c:blend_4_b12
Line
Count
Source
137
141k
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
138
141k
  const __m128i v_s0_w = xx_loadl_64(src0);
139
141k
  const __m128i v_s1_w = xx_loadl_64(src1);
140
141
  // Interleave
142
141k
  const __m128i v_m01_w = _mm_unpacklo_epi16(v_m0_w, v_m1_w);
143
141k
  const __m128i v_s01_w = _mm_unpacklo_epi16(v_s0_w, v_s1_w);
144
145
  // Multiply-Add
146
141k
  const __m128i v_sum_d = _mm_madd_epi16(v_s01_w, v_m01_w);
147
148
  // Scale
149
141k
  const __m128i v_ssum_d =
150
141k
      _mm_srli_epi32(v_sum_d, AOM_BLEND_A64_ROUND_BITS - 1);
151
152
  // Pack
153
141k
  const __m128i v_pssum_d = _mm_packs_epi32(v_ssum_d, v_ssum_d);
154
155
  // Round
156
141k
  const __m128i v_res_w = xx_round_epu16(v_pssum_d);
157
158
141k
  return v_res_w;
159
141k
}
blend_a64_vmask_sse4.c:blend_4_b12
Line
Count
Source
137
6.80k
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
138
6.80k
  const __m128i v_s0_w = xx_loadl_64(src0);
139
6.80k
  const __m128i v_s1_w = xx_loadl_64(src1);
140
141
  // Interleave
142
6.80k
  const __m128i v_m01_w = _mm_unpacklo_epi16(v_m0_w, v_m1_w);
143
6.80k
  const __m128i v_s01_w = _mm_unpacklo_epi16(v_s0_w, v_s1_w);
144
145
  // Multiply-Add
146
6.80k
  const __m128i v_sum_d = _mm_madd_epi16(v_s01_w, v_m01_w);
147
148
  // Scale
149
6.80k
  const __m128i v_ssum_d =
150
6.80k
      _mm_srli_epi32(v_sum_d, AOM_BLEND_A64_ROUND_BITS - 1);
151
152
  // Pack
153
6.80k
  const __m128i v_pssum_d = _mm_packs_epi32(v_ssum_d, v_ssum_d);
154
155
  // Round
156
6.80k
  const __m128i v_res_w = xx_round_epu16(v_pssum_d);
157
158
6.80k
  return v_res_w;
159
6.80k
}
Unexecuted instantiation: blend_a64_mask_avx2.c:blend_4_b12
160
161
static inline __m128i blend_8_b12(const uint16_t *src0, const uint16_t *src1,
162
978k
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
163
978k
  const __m128i v_s0_w = xx_loadu_128(src0);
164
978k
  const __m128i v_s1_w = xx_loadu_128(src1);
165
166
  // Interleave
167
978k
  const __m128i v_m01l_w = _mm_unpacklo_epi16(v_m0_w, v_m1_w);
168
978k
  const __m128i v_m01h_w = _mm_unpackhi_epi16(v_m0_w, v_m1_w);
169
978k
  const __m128i v_s01l_w = _mm_unpacklo_epi16(v_s0_w, v_s1_w);
170
978k
  const __m128i v_s01h_w = _mm_unpackhi_epi16(v_s0_w, v_s1_w);
171
172
  // Multiply-Add
173
978k
  const __m128i v_suml_d = _mm_madd_epi16(v_s01l_w, v_m01l_w);
174
978k
  const __m128i v_sumh_d = _mm_madd_epi16(v_s01h_w, v_m01h_w);
175
176
  // Scale
177
978k
  const __m128i v_ssuml_d =
178
978k
      _mm_srli_epi32(v_suml_d, AOM_BLEND_A64_ROUND_BITS - 1);
179
978k
  const __m128i v_ssumh_d =
180
978k
      _mm_srli_epi32(v_sumh_d, AOM_BLEND_A64_ROUND_BITS - 1);
181
182
  // Pack
183
978k
  const __m128i v_pssum_d = _mm_packs_epi32(v_ssuml_d, v_ssumh_d);
184
185
  // Round
186
978k
  const __m128i v_res_w = xx_round_epu16(v_pssum_d);
187
188
978k
  return v_res_w;
189
978k
}
blend_a64_mask_sse4.c:blend_8_b12
Line
Count
Source
162
675k
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
163
675k
  const __m128i v_s0_w = xx_loadu_128(src0);
164
675k
  const __m128i v_s1_w = xx_loadu_128(src1);
165
166
  // Interleave
167
675k
  const __m128i v_m01l_w = _mm_unpacklo_epi16(v_m0_w, v_m1_w);
168
675k
  const __m128i v_m01h_w = _mm_unpackhi_epi16(v_m0_w, v_m1_w);
169
675k
  const __m128i v_s01l_w = _mm_unpacklo_epi16(v_s0_w, v_s1_w);
170
675k
  const __m128i v_s01h_w = _mm_unpackhi_epi16(v_s0_w, v_s1_w);
171
172
  // Multiply-Add
173
675k
  const __m128i v_suml_d = _mm_madd_epi16(v_s01l_w, v_m01l_w);
174
675k
  const __m128i v_sumh_d = _mm_madd_epi16(v_s01h_w, v_m01h_w);
175
176
  // Scale
177
675k
  const __m128i v_ssuml_d =
178
675k
      _mm_srli_epi32(v_suml_d, AOM_BLEND_A64_ROUND_BITS - 1);
179
675k
  const __m128i v_ssumh_d =
180
675k
      _mm_srli_epi32(v_sumh_d, AOM_BLEND_A64_ROUND_BITS - 1);
181
182
  // Pack
183
675k
  const __m128i v_pssum_d = _mm_packs_epi32(v_ssuml_d, v_ssumh_d);
184
185
  // Round
186
675k
  const __m128i v_res_w = xx_round_epu16(v_pssum_d);
187
188
675k
  return v_res_w;
189
675k
}
blend_a64_vmask_sse4.c:blend_8_b12
Line
Count
Source
162
302k
                                  const __m128i v_m0_w, const __m128i v_m1_w) {
163
302k
  const __m128i v_s0_w = xx_loadu_128(src0);
164
302k
  const __m128i v_s1_w = xx_loadu_128(src1);
165
166
  // Interleave
167
302k
  const __m128i v_m01l_w = _mm_unpacklo_epi16(v_m0_w, v_m1_w);
168
302k
  const __m128i v_m01h_w = _mm_unpackhi_epi16(v_m0_w, v_m1_w);
169
302k
  const __m128i v_s01l_w = _mm_unpacklo_epi16(v_s0_w, v_s1_w);
170
302k
  const __m128i v_s01h_w = _mm_unpackhi_epi16(v_s0_w, v_s1_w);
171
172
  // Multiply-Add
173
302k
  const __m128i v_suml_d = _mm_madd_epi16(v_s01l_w, v_m01l_w);
174
302k
  const __m128i v_sumh_d = _mm_madd_epi16(v_s01h_w, v_m01h_w);
175
176
  // Scale
177
302k
  const __m128i v_ssuml_d =
178
302k
      _mm_srli_epi32(v_suml_d, AOM_BLEND_A64_ROUND_BITS - 1);
179
302k
  const __m128i v_ssumh_d =
180
302k
      _mm_srli_epi32(v_sumh_d, AOM_BLEND_A64_ROUND_BITS - 1);
181
182
  // Pack
183
302k
  const __m128i v_pssum_d = _mm_packs_epi32(v_ssuml_d, v_ssumh_d);
184
185
  // Round
186
302k
  const __m128i v_res_w = xx_round_epu16(v_pssum_d);
187
188
302k
  return v_res_w;
189
302k
}
Unexecuted instantiation: blend_a64_mask_avx2.c:blend_8_b12
190
191
#endif  // AOM_AOM_DSP_X86_BLEND_SSE4_H_