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
|
/*
* 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 <arm_neon.h>
#include "./aom_dsp_rtcd.h"
#include "./aom_config.h"
#include "aom_ports/mem.h"
#include "aom/aom_integer.h"
#include "aom_dsp/variance.h"
static const uint8_t bilinear_filters[8][2] = {
{ 128, 0 }, { 112, 16 }, { 96, 32 }, { 80, 48 },
{ 64, 64 }, { 48, 80 }, { 32, 96 }, { 16, 112 },
};
static void var_filter_block2d_bil_w8(const uint8_t *src_ptr,
uint8_t *output_ptr,
unsigned int src_pixels_per_line,
int pixel_step,
unsigned int output_height,
unsigned int output_width,
const uint8_t *filter) {
const uint8x8_t f0 = vmov_n_u8(filter[0]);
const uint8x8_t f1 = vmov_n_u8(filter[1]);
unsigned int i;
for (i = 0; i < output_height; ++i) {
const uint8x8_t src_0 = vld1_u8(&src_ptr[0]);
const uint8x8_t src_1 = vld1_u8(&src_ptr[pixel_step]);
const uint16x8_t a = vmull_u8(src_0, f0);
const uint16x8_t b = vmlal_u8(a, src_1, f1);
const uint8x8_t out = vrshrn_n_u16(b, FILTER_BITS);
vst1_u8(&output_ptr[0], out);
// Next row...
src_ptr += src_pixels_per_line;
output_ptr += output_width;
}
}
static void var_filter_block2d_bil_w16(const uint8_t *src_ptr,
uint8_t *output_ptr,
unsigned int src_pixels_per_line,
int pixel_step,
unsigned int output_height,
unsigned int output_width,
const uint8_t *filter) {
const uint8x8_t f0 = vmov_n_u8(filter[0]);
const uint8x8_t f1 = vmov_n_u8(filter[1]);
unsigned int i, j;
for (i = 0; i < output_height; ++i) {
for (j = 0; j < output_width; j += 16) {
const uint8x16_t src_0 = vld1q_u8(&src_ptr[j]);
const uint8x16_t src_1 = vld1q_u8(&src_ptr[j + pixel_step]);
const uint16x8_t a = vmull_u8(vget_low_u8(src_0), f0);
const uint16x8_t b = vmlal_u8(a, vget_low_u8(src_1), f1);
const uint8x8_t out_lo = vrshrn_n_u16(b, FILTER_BITS);
const uint16x8_t c = vmull_u8(vget_high_u8(src_0), f0);
const uint16x8_t d = vmlal_u8(c, vget_high_u8(src_1), f1);
const uint8x8_t out_hi = vrshrn_n_u16(d, FILTER_BITS);
vst1q_u8(&output_ptr[j], vcombine_u8(out_lo, out_hi));
}
// Next row...
src_ptr += src_pixels_per_line;
output_ptr += output_width;
}
}
unsigned int aom_sub_pixel_variance8x8_neon(const uint8_t *src, int src_stride,
int xoffset, int yoffset,
const uint8_t *dst, int dst_stride,
unsigned int *sse) {
DECLARE_ALIGNED(16, uint8_t, temp2[8 * 8]);
DECLARE_ALIGNED(16, uint8_t, fdata3[9 * 8]);
var_filter_block2d_bil_w8(src, fdata3, src_stride, 1, 9, 8,
bilinear_filters[xoffset]);
var_filter_block2d_bil_w8(fdata3, temp2, 8, 8, 8, 8,
bilinear_filters[yoffset]);
return aom_variance8x8_neon(temp2, 8, dst, dst_stride, sse);
}
unsigned int aom_sub_pixel_variance16x16_neon(const uint8_t *src,
int src_stride, int xoffset,
int yoffset, const uint8_t *dst,
int dst_stride,
unsigned int *sse) {
DECLARE_ALIGNED(16, uint8_t, temp2[16 * 16]);
DECLARE_ALIGNED(16, uint8_t, fdata3[17 * 16]);
var_filter_block2d_bil_w16(src, fdata3, src_stride, 1, 17, 16,
bilinear_filters[xoffset]);
var_filter_block2d_bil_w16(fdata3, temp2, 16, 16, 16, 16,
bilinear_filters[yoffset]);
return aom_variance16x16_neon(temp2, 16, dst, dst_stride, sse);
}
unsigned int aom_sub_pixel_variance32x32_neon(const uint8_t *src,
int src_stride, int xoffset,
int yoffset, const uint8_t *dst,
int dst_stride,
unsigned int *sse) {
DECLARE_ALIGNED(16, uint8_t, temp2[32 * 32]);
DECLARE_ALIGNED(16, uint8_t, fdata3[33 * 32]);
var_filter_block2d_bil_w16(src, fdata3, src_stride, 1, 33, 32,
bilinear_filters[xoffset]);
var_filter_block2d_bil_w16(fdata3, temp2, 32, 32, 32, 32,
bilinear_filters[yoffset]);
return aom_variance32x32_neon(temp2, 32, dst, dst_stride, sse);
}
unsigned int aom_sub_pixel_variance64x64_neon(const uint8_t *src,
int src_stride, int xoffset,
int yoffset, const uint8_t *dst,
int dst_stride,
unsigned int *sse) {
DECLARE_ALIGNED(16, uint8_t, temp2[64 * 64]);
DECLARE_ALIGNED(16, uint8_t, fdata3[65 * 64]);
var_filter_block2d_bil_w16(src, fdata3, src_stride, 1, 65, 64,
bilinear_filters[xoffset]);
var_filter_block2d_bil_w16(fdata3, temp2, 64, 64, 64, 64,
bilinear_filters[yoffset]);
return aom_variance64x64_neon(temp2, 64, dst, dst_stride, sse);
}
|
