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
188
189
190
|
/*
* Copyright (c) 2023 The WebM project authors. All Rights Reserved.
* Copyright (c) 2023, 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 <assert.h>
#include "config/aom_config.h"
#include "config/aom_dsp_rtcd.h"
#include "aom_dsp/arm/blend_neon.h"
#include "aom_dsp/arm/dist_wtd_avg_neon.h"
#include "aom_dsp/arm/mem_neon.h"
#include "aom_dsp/blend.h"
void aom_highbd_comp_avg_pred_neon(uint8_t *comp_pred8, const uint8_t *pred8,
int width, int height, const uint8_t *ref8,
int ref_stride) {
const uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
const uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
uint16_t *comp_pred = CONVERT_TO_SHORTPTR(comp_pred8);
int i = height;
if (width > 8) {
do {
int j = 0;
do {
const uint16x8_t p = vld1q_u16(pred + j);
const uint16x8_t r = vld1q_u16(ref + j);
uint16x8_t avg = vrhaddq_u16(p, r);
vst1q_u16(comp_pred + j, avg);
j += 8;
} while (j < width);
comp_pred += width;
pred += width;
ref += ref_stride;
} while (--i != 0);
} else if (width == 8) {
do {
const uint16x8_t p = vld1q_u16(pred);
const uint16x8_t r = vld1q_u16(ref);
uint16x8_t avg = vrhaddq_u16(p, r);
vst1q_u16(comp_pred, avg);
comp_pred += width;
pred += width;
ref += ref_stride;
} while (--i != 0);
} else {
assert(width == 4);
do {
const uint16x4_t p = vld1_u16(pred);
const uint16x4_t r = vld1_u16(ref);
uint16x4_t avg = vrhadd_u16(p, r);
vst1_u16(comp_pred, avg);
comp_pred += width;
pred += width;
ref += ref_stride;
} while (--i != 0);
}
}
void aom_highbd_comp_mask_pred_neon(uint8_t *comp_pred8, const uint8_t *pred8,
int width, int height, const uint8_t *ref8,
int ref_stride, const uint8_t *mask,
int mask_stride, int invert_mask) {
uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
uint16_t *comp_pred = CONVERT_TO_SHORTPTR(comp_pred8);
const uint16_t *src0 = invert_mask ? pred : ref;
const uint16_t *src1 = invert_mask ? ref : pred;
const int src_stride0 = invert_mask ? width : ref_stride;
const int src_stride1 = invert_mask ? ref_stride : width;
if (width >= 8) {
do {
int j = 0;
do {
const uint16x8_t s0 = vld1q_u16(src0 + j);
const uint16x8_t s1 = vld1q_u16(src1 + j);
const uint16x8_t m0 = vmovl_u8(vld1_u8(mask + j));
uint16x8_t blend_u16 = alpha_blend_a64_u16x8(m0, s0, s1);
vst1q_u16(comp_pred + j, blend_u16);
j += 8;
} while (j < width);
src0 += src_stride0;
src1 += src_stride1;
mask += mask_stride;
comp_pred += width;
} while (--height != 0);
} else {
assert(width == 4);
do {
const uint16x4_t s0 = vld1_u16(src0);
const uint16x4_t s1 = vld1_u16(src1);
const uint16x4_t m0 = vget_low_u16(vmovl_u8(load_unaligned_u8_4x1(mask)));
uint16x4_t blend_u16 = alpha_blend_a64_u16x4(m0, s0, s1);
vst1_u16(comp_pred, blend_u16);
src0 += src_stride0;
src1 += src_stride1;
mask += mask_stride;
comp_pred += 4;
} while (--height != 0);
}
}
void aom_highbd_dist_wtd_comp_avg_pred_neon(
uint8_t *comp_pred8, const uint8_t *pred8, int width, int height,
const uint8_t *ref8, int ref_stride,
const DIST_WTD_COMP_PARAMS *jcp_param) {
const uint16x8_t fwd_offset_u16 = vdupq_n_u16(jcp_param->fwd_offset);
const uint16x8_t bck_offset_u16 = vdupq_n_u16(jcp_param->bck_offset);
const uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
const uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
uint16_t *comp_pred = CONVERT_TO_SHORTPTR(comp_pred8);
if (width > 8) {
do {
int j = 0;
do {
const uint16x8_t p = vld1q_u16(pred + j);
const uint16x8_t r = vld1q_u16(ref + j);
const uint16x8_t avg =
dist_wtd_avg_u16x8(r, p, fwd_offset_u16, bck_offset_u16);
vst1q_u16(comp_pred + j, avg);
j += 8;
} while (j < width);
comp_pred += width;
pred += width;
ref += ref_stride;
} while (--height != 0);
} else if (width == 8) {
do {
const uint16x8_t p = vld1q_u16(pred);
const uint16x8_t r = vld1q_u16(ref);
const uint16x8_t avg =
dist_wtd_avg_u16x8(r, p, fwd_offset_u16, bck_offset_u16);
vst1q_u16(comp_pred, avg);
comp_pred += width;
pred += width;
ref += ref_stride;
} while (--height != 0);
} else {
assert(width == 4);
do {
const uint16x4_t p = vld1_u16(pred);
const uint16x4_t r = vld1_u16(ref);
const uint16x4_t avg = dist_wtd_avg_u16x4(
r, p, vget_low_u16(fwd_offset_u16), vget_low_u16(bck_offset_u16));
vst1_u16(comp_pred, avg);
comp_pred += width;
pred += width;
ref += ref_stride;
} while (--height != 0);
}
}
|
