1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
|
/*
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include <immintrin.h>
#include "./aom_dsp_rtcd.h"
#include "aom_ports/mem.h"
#define FSAD64_H(h) \
unsigned int aom_sad64x##h##_avx2(const uint8_t *src_ptr, int src_stride, \
const uint8_t *ref_ptr, int ref_stride) { \
int i, res; \
__m256i sad1_reg, sad2_reg, ref1_reg, ref2_reg; \
__m256i sum_sad = _mm256_setzero_si256(); \
__m256i sum_sad_h; \
__m128i sum_sad128; \
for (i = 0; i < h; i++) { \
ref1_reg = _mm256_loadu_si256((__m256i const *)ref_ptr); \
ref2_reg = _mm256_loadu_si256((__m256i const *)(ref_ptr + 32)); \
sad1_reg = _mm256_sad_epu8( \
ref1_reg, _mm256_loadu_si256((__m256i const *)src_ptr)); \
sad2_reg = _mm256_sad_epu8( \
ref2_reg, _mm256_loadu_si256((__m256i const *)(src_ptr + 32))); \
sum_sad = \
_mm256_add_epi32(sum_sad, _mm256_add_epi32(sad1_reg, sad2_reg)); \
ref_ptr += ref_stride; \
src_ptr += src_stride; \
} \
sum_sad_h = _mm256_srli_si256(sum_sad, 8); \
sum_sad = _mm256_add_epi32(sum_sad, sum_sad_h); \
sum_sad128 = _mm256_extracti128_si256(sum_sad, 1); \
sum_sad128 = _mm_add_epi32(_mm256_castsi256_si128(sum_sad), sum_sad128); \
res = _mm_cvtsi128_si32(sum_sad128); \
_mm256_zeroupper(); \
return res; \
}
#define FSAD32_H(h) \
unsigned int aom_sad32x##h##_avx2(const uint8_t *src_ptr, int src_stride, \
const uint8_t *ref_ptr, int ref_stride) { \
int i, res; \
__m256i sad1_reg, sad2_reg, ref1_reg, ref2_reg; \
__m256i sum_sad = _mm256_setzero_si256(); \
__m256i sum_sad_h; \
__m128i sum_sad128; \
int ref2_stride = ref_stride << 1; \
int src2_stride = src_stride << 1; \
int max = h >> 1; \
for (i = 0; i < max; i++) { \
ref1_reg = _mm256_loadu_si256((__m256i const *)ref_ptr); \
ref2_reg = _mm256_loadu_si256((__m256i const *)(ref_ptr + ref_stride)); \
sad1_reg = _mm256_sad_epu8( \
ref1_reg, _mm256_loadu_si256((__m256i const *)src_ptr)); \
sad2_reg = _mm256_sad_epu8( \
ref2_reg, \
_mm256_loadu_si256((__m256i const *)(src_ptr + src_stride))); \
sum_sad = \
_mm256_add_epi32(sum_sad, _mm256_add_epi32(sad1_reg, sad2_reg)); \
ref_ptr += ref2_stride; \
src_ptr += src2_stride; \
} \
sum_sad_h = _mm256_srli_si256(sum_sad, 8); \
sum_sad = _mm256_add_epi32(sum_sad, sum_sad_h); \
sum_sad128 = _mm256_extracti128_si256(sum_sad, 1); \
sum_sad128 = _mm_add_epi32(_mm256_castsi256_si128(sum_sad), sum_sad128); \
res = _mm_cvtsi128_si32(sum_sad128); \
_mm256_zeroupper(); \
return res; \
}
#define FSAD64 \
FSAD64_H(64); \
FSAD64_H(32);
#define FSAD32 \
FSAD32_H(64); \
FSAD32_H(32); \
FSAD32_H(16);
/* clang-format off */
FSAD64
FSAD32
/* clang-format on */
#undef FSAD64
#undef FSAD32
#undef FSAD64_H
#undef FSAD32_H
#define FSADAVG64_H(h) \
unsigned int aom_sad64x##h##_avg_avx2( \
const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, \
int ref_stride, const uint8_t *second_pred) { \
int i, res; \
__m256i sad1_reg, sad2_reg, ref1_reg, ref2_reg; \
__m256i sum_sad = _mm256_setzero_si256(); \
__m256i sum_sad_h; \
__m128i sum_sad128; \
for (i = 0; i < h; i++) { \
ref1_reg = _mm256_loadu_si256((__m256i const *)ref_ptr); \
ref2_reg = _mm256_loadu_si256((__m256i const *)(ref_ptr + 32)); \
ref1_reg = _mm256_avg_epu8( \
ref1_reg, _mm256_loadu_si256((__m256i const *)second_pred)); \
ref2_reg = _mm256_avg_epu8( \
ref2_reg, _mm256_loadu_si256((__m256i const *)(second_pred + 32))); \
sad1_reg = _mm256_sad_epu8( \
ref1_reg, _mm256_loadu_si256((__m256i const *)src_ptr)); \
sad2_reg = _mm256_sad_epu8( \
ref2_reg, _mm256_loadu_si256((__m256i const *)(src_ptr + 32))); \
sum_sad = \
_mm256_add_epi32(sum_sad, _mm256_add_epi32(sad1_reg, sad2_reg)); \
ref_ptr += ref_stride; \
src_ptr += src_stride; \
second_pred += 64; \
} \
sum_sad_h = _mm256_srli_si256(sum_sad, 8); \
sum_sad = _mm256_add_epi32(sum_sad, sum_sad_h); \
sum_sad128 = _mm256_extracti128_si256(sum_sad, 1); \
sum_sad128 = _mm_add_epi32(_mm256_castsi256_si128(sum_sad), sum_sad128); \
res = _mm_cvtsi128_si32(sum_sad128); \
_mm256_zeroupper(); \
return res; \
}
#define FSADAVG32_H(h) \
unsigned int aom_sad32x##h##_avg_avx2( \
const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, \
int ref_stride, const uint8_t *second_pred) { \
int i, res; \
__m256i sad1_reg, sad2_reg, ref1_reg, ref2_reg; \
__m256i sum_sad = _mm256_setzero_si256(); \
__m256i sum_sad_h; \
__m128i sum_sad128; \
int ref2_stride = ref_stride << 1; \
int src2_stride = src_stride << 1; \
int max = h >> 1; \
for (i = 0; i < max; i++) { \
ref1_reg = _mm256_loadu_si256((__m256i const *)ref_ptr); \
ref2_reg = _mm256_loadu_si256((__m256i const *)(ref_ptr + ref_stride)); \
ref1_reg = _mm256_avg_epu8( \
ref1_reg, _mm256_loadu_si256((__m256i const *)second_pred)); \
ref2_reg = _mm256_avg_epu8( \
ref2_reg, _mm256_loadu_si256((__m256i const *)(second_pred + 32))); \
sad1_reg = _mm256_sad_epu8( \
ref1_reg, _mm256_loadu_si256((__m256i const *)src_ptr)); \
sad2_reg = _mm256_sad_epu8( \
ref2_reg, \
_mm256_loadu_si256((__m256i const *)(src_ptr + src_stride))); \
sum_sad = \
_mm256_add_epi32(sum_sad, _mm256_add_epi32(sad1_reg, sad2_reg)); \
ref_ptr += ref2_stride; \
src_ptr += src2_stride; \
second_pred += 64; \
} \
sum_sad_h = _mm256_srli_si256(sum_sad, 8); \
sum_sad = _mm256_add_epi32(sum_sad, sum_sad_h); \
sum_sad128 = _mm256_extracti128_si256(sum_sad, 1); \
sum_sad128 = _mm_add_epi32(_mm256_castsi256_si128(sum_sad), sum_sad128); \
res = _mm_cvtsi128_si32(sum_sad128); \
_mm256_zeroupper(); \
return res; \
}
#define FSADAVG64 \
FSADAVG64_H(64); \
FSADAVG64_H(32);
#define FSADAVG32 \
FSADAVG32_H(64); \
FSADAVG32_H(32); \
FSADAVG32_H(16);
/* clang-format off */
FSADAVG64
FSADAVG32
/* clang-format on */
#undef FSADAVG64
#undef FSADAVG32
#undef FSADAVG64_H
#undef FSADAVG32_H
|
