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
|
/*
* Copyright (c) 2012 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "./vpx_dsp_rtcd.h"
typedef void (*get_var_avx2)(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride,
unsigned int *sse, int *sum);
void vpx_get32x32var_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride, unsigned int *sse,
int *sum);
static void variance_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride, int w, int h,
unsigned int *sse, int *sum, get_var_avx2 var_fn,
int block_size) {
int i, j;
*sse = 0;
*sum = 0;
for (i = 0; i < h; i += 16) {
for (j = 0; j < w; j += block_size) {
unsigned int sse0;
int sum0;
var_fn(&src[src_stride * i + j], src_stride, &ref[ref_stride * i + j],
ref_stride, &sse0, &sum0);
*sse += sse0;
*sum += sum0;
}
}
}
unsigned int vpx_variance16x16_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride,
unsigned int *sse) {
int sum;
variance_avx2(src, src_stride, ref, ref_stride, 16, 16, sse, &sum,
vpx_get16x16var_avx2, 16);
return *sse - (((uint32_t)((int64_t)sum * sum)) >> 8);
}
unsigned int vpx_mse16x16_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride,
unsigned int *sse) {
int sum;
vpx_get16x16var_avx2(src, src_stride, ref, ref_stride, sse, &sum);
return *sse;
}
unsigned int vpx_variance32x16_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride,
unsigned int *sse) {
int sum;
variance_avx2(src, src_stride, ref, ref_stride, 32, 16, sse, &sum,
vpx_get32x32var_avx2, 32);
return *sse - (uint32_t)(((int64_t)sum * sum) >> 9);
}
unsigned int vpx_variance32x32_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride,
unsigned int *sse) {
int sum;
variance_avx2(src, src_stride, ref, ref_stride, 32, 32, sse, &sum,
vpx_get32x32var_avx2, 32);
return *sse - (uint32_t)(((int64_t)sum * sum) >> 10);
}
unsigned int vpx_variance64x64_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride,
unsigned int *sse) {
int sum;
variance_avx2(src, src_stride, ref, ref_stride, 64, 64, sse, &sum,
vpx_get32x32var_avx2, 32);
return *sse - (uint32_t)(((int64_t)sum * sum) >> 12);
}
unsigned int vpx_variance64x32_avx2(const uint8_t *src, int src_stride,
const uint8_t *ref, int ref_stride,
unsigned int *sse) {
int sum;
variance_avx2(src, src_stride, ref, ref_stride, 64, 32, sse, &sum,
vpx_get32x32var_avx2, 32);
return *sse - (uint32_t)(((int64_t)sum * sum) >> 11);
}
unsigned int vpx_sub_pixel_variance32xh_avx2(const uint8_t *src, int src_stride,
int x_offset, int y_offset,
const uint8_t *dst, int dst_stride,
int height, unsigned int *sse);
unsigned int vpx_sub_pixel_avg_variance32xh_avx2(
const uint8_t *src, int src_stride, int x_offset, int y_offset,
const uint8_t *dst, int dst_stride, const uint8_t *sec, int sec_stride,
int height, unsigned int *sseptr);
unsigned int vpx_sub_pixel_variance64x64_avx2(const uint8_t *src,
int src_stride, int x_offset,
int y_offset, const uint8_t *dst,
int dst_stride,
unsigned int *sse) {
unsigned int sse1;
const int se1 = vpx_sub_pixel_variance32xh_avx2(
src, src_stride, x_offset, y_offset, dst, dst_stride, 64, &sse1);
unsigned int sse2;
const int se2 =
vpx_sub_pixel_variance32xh_avx2(src + 32, src_stride, x_offset, y_offset,
dst + 32, dst_stride, 64, &sse2);
const int se = se1 + se2;
*sse = sse1 + sse2;
return *sse - (uint32_t)(((int64_t)se * se) >> 12);
}
unsigned int vpx_sub_pixel_variance32x32_avx2(const uint8_t *src,
int src_stride, int x_offset,
int y_offset, const uint8_t *dst,
int dst_stride,
unsigned int *sse) {
const int se = vpx_sub_pixel_variance32xh_avx2(
src, src_stride, x_offset, y_offset, dst, dst_stride, 32, sse);
return *sse - (uint32_t)(((int64_t)se * se) >> 10);
}
unsigned int vpx_sub_pixel_avg_variance64x64_avx2(
const uint8_t *src, int src_stride, int x_offset, int y_offset,
const uint8_t *dst, int dst_stride, unsigned int *sse, const uint8_t *sec) {
unsigned int sse1;
const int se1 = vpx_sub_pixel_avg_variance32xh_avx2(
src, src_stride, x_offset, y_offset, dst, dst_stride, sec, 64, 64, &sse1);
unsigned int sse2;
const int se2 = vpx_sub_pixel_avg_variance32xh_avx2(
src + 32, src_stride, x_offset, y_offset, dst + 32, dst_stride, sec + 32,
64, 64, &sse2);
const int se = se1 + se2;
*sse = sse1 + sse2;
return *sse - (uint32_t)(((int64_t)se * se) >> 12);
}
unsigned int vpx_sub_pixel_avg_variance32x32_avx2(
const uint8_t *src, int src_stride, int x_offset, int y_offset,
const uint8_t *dst, int dst_stride, unsigned int *sse, const uint8_t *sec) {
// Process 32 elements in parallel.
const int se = vpx_sub_pixel_avg_variance32xh_avx2(
src, src_stride, x_offset, y_offset, dst, dst_stride, sec, 32, 32, sse);
return *sse - (uint32_t)(((int64_t)se * se) >> 10);
}
|
