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
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
|
/*
* 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 <smmintrin.h> // SSE4.1
#include <assert.h>
#include "aom/aom_integer.h"
#include "aom_ports/mem.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/blend.h"
#include "aom_dsp/x86/synonyms.h"
#include "aom_dsp/x86/blend_sse4.h"
#include "config/aom_dsp_rtcd.h"
//////////////////////////////////////////////////////////////////////////////
// Implementation - No sub-sampling
//////////////////////////////////////////////////////////////////////////////
static void blend_a64_vmask_w4_sse4_1(uint8_t *dst, uint32_t dst_stride,
const uint8_t *src0, uint32_t src0_stride,
const uint8_t *src1, uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
const __m128i v_maxval_w = _mm_set1_epi16(AOM_BLEND_A64_MAX_ALPHA);
(void)w;
do {
const __m128i v_m0_w = _mm_set1_epi16(*mask);
const __m128i v_m1_w = _mm_sub_epi16(v_maxval_w, v_m0_w);
const __m128i v_res_w = blend_4(src0, src1, &v_m0_w, &v_m1_w);
const __m128i v_res_b = _mm_packus_epi16(v_res_w, v_res_w);
xx_storel_32(dst, v_res_b);
dst += dst_stride;
src0 += src0_stride;
src1 += src1_stride;
mask += 1;
} while (--h);
}
static void blend_a64_vmask_w8_sse4_1(uint8_t *dst, uint32_t dst_stride,
const uint8_t *src0, uint32_t src0_stride,
const uint8_t *src1, uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
const __m128i v_maxval_w = _mm_set1_epi16(AOM_BLEND_A64_MAX_ALPHA);
(void)w;
do {
const __m128i v_m0_w = _mm_set1_epi16(*mask);
const __m128i v_m1_w = _mm_sub_epi16(v_maxval_w, v_m0_w);
const __m128i v_res_w = blend_8(src0, src1, &v_m0_w, &v_m1_w);
const __m128i v_res_b = _mm_packus_epi16(v_res_w, v_res_w);
xx_storel_64(dst, v_res_b);
dst += dst_stride;
src0 += src0_stride;
src1 += src1_stride;
mask += 1;
} while (--h);
}
static void blend_a64_vmask_w16n_sse4_1(uint8_t *dst, uint32_t dst_stride,
const uint8_t *src0,
uint32_t src0_stride,
const uint8_t *src1,
uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
const __m128i v_maxval_w = _mm_set1_epi16(AOM_BLEND_A64_MAX_ALPHA);
do {
int c;
const __m128i v_m0_w = _mm_set1_epi16(*mask);
const __m128i v_m1_w = _mm_sub_epi16(v_maxval_w, v_m0_w);
for (c = 0; c < w; c += 16) {
const __m128i v_resl_w = blend_8(src0 + c, src1 + c, &v_m0_w, &v_m1_w);
const __m128i v_resh_w =
blend_8(src0 + c + 8, src1 + c + 8, &v_m0_w, &v_m1_w);
const __m128i v_res_b = _mm_packus_epi16(v_resl_w, v_resh_w);
xx_storeu_128(dst + c, v_res_b);
}
dst += dst_stride;
src0 += src0_stride;
src1 += src1_stride;
mask += 1;
} while (--h);
}
//////////////////////////////////////////////////////////////////////////////
// Dispatch
//////////////////////////////////////////////////////////////////////////////
void aom_blend_a64_vmask_sse4_1(uint8_t *dst, uint32_t dst_stride,
const uint8_t *src0, uint32_t src0_stride,
const uint8_t *src1, uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
typedef void (*blend_fn)(uint8_t * dst, uint32_t dst_stride,
const uint8_t *src0, uint32_t src0_stride,
const uint8_t *src1, uint32_t src1_stride,
const uint8_t *mask, int w, int h);
// Dimension: width_index
static const blend_fn blend[9] = {
blend_a64_vmask_w16n_sse4_1, // w % 16 == 0
aom_blend_a64_vmask_c, // w == 1
aom_blend_a64_vmask_c, // w == 2
NULL, // INVALID
blend_a64_vmask_w4_sse4_1, // w == 4
NULL, // INVALID
NULL, // INVALID
NULL, // INVALID
blend_a64_vmask_w8_sse4_1, // w == 8
};
assert(IMPLIES(src0 == dst, src0_stride == dst_stride));
assert(IMPLIES(src1 == dst, src1_stride == dst_stride));
assert(h >= 1);
assert(w >= 1);
assert(IS_POWER_OF_TWO(h));
assert(IS_POWER_OF_TWO(w));
blend[w & 0xf](dst, dst_stride, src0, src0_stride, src1, src1_stride, mask, w,
h);
}
#if CONFIG_AV1_HIGHBITDEPTH
//////////////////////////////////////////////////////////////////////////////
// Implementation - No sub-sampling
//////////////////////////////////////////////////////////////////////////////
static INLINE void blend_a64_vmask_bn_w4_sse4_1(
uint16_t *dst, uint32_t dst_stride, const uint16_t *src0,
uint32_t src0_stride, const uint16_t *src1, uint32_t src1_stride,
const uint8_t *mask, int h, blend_unit_fn blend) {
const __m128i v_maxval_w = _mm_set1_epi16(AOM_BLEND_A64_MAX_ALPHA);
do {
const __m128i v_m0_w = _mm_set1_epi16(*mask);
const __m128i v_m1_w = _mm_sub_epi16(v_maxval_w, v_m0_w);
const __m128i v_res_w = blend(src0, src1, v_m0_w, v_m1_w);
xx_storel_64(dst, v_res_w);
dst += dst_stride;
src0 += src0_stride;
src1 += src1_stride;
mask += 1;
} while (--h);
}
static void blend_a64_vmask_b10_w4_sse4_1(uint16_t *dst, uint32_t dst_stride,
const uint16_t *src0,
uint32_t src0_stride,
const uint16_t *src1,
uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
(void)w;
blend_a64_vmask_bn_w4_sse4_1(dst, dst_stride, src0, src0_stride, src1,
src1_stride, mask, h, blend_4_b10);
}
static void blend_a64_vmask_b12_w4_sse4_1(uint16_t *dst, uint32_t dst_stride,
const uint16_t *src0,
uint32_t src0_stride,
const uint16_t *src1,
uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
(void)w;
blend_a64_vmask_bn_w4_sse4_1(dst, dst_stride, src0, src0_stride, src1,
src1_stride, mask, h, blend_4_b12);
}
static INLINE void blend_a64_vmask_bn_w8n_sse4_1(
uint16_t *dst, uint32_t dst_stride, const uint16_t *src0,
uint32_t src0_stride, const uint16_t *src1, uint32_t src1_stride,
const uint8_t *mask, int w, int h, blend_unit_fn blend) {
const __m128i v_maxval_w = _mm_set1_epi16(AOM_BLEND_A64_MAX_ALPHA);
do {
int c;
const __m128i v_m0_w = _mm_set1_epi16(*mask);
const __m128i v_m1_w = _mm_sub_epi16(v_maxval_w, v_m0_w);
for (c = 0; c < w; c += 8) {
const __m128i v_res_w = blend(src0 + c, src1 + c, v_m0_w, v_m1_w);
xx_storeu_128(dst + c, v_res_w);
}
dst += dst_stride;
src0 += src0_stride;
src1 += src1_stride;
mask += 1;
} while (--h);
}
static void blend_a64_vmask_b10_w8n_sse4_1(uint16_t *dst, uint32_t dst_stride,
const uint16_t *src0,
uint32_t src0_stride,
const uint16_t *src1,
uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
blend_a64_vmask_bn_w8n_sse4_1(dst, dst_stride, src0, src0_stride, src1,
src1_stride, mask, w, h, blend_8_b10);
}
static void blend_a64_vmask_b12_w8n_sse4_1(uint16_t *dst, uint32_t dst_stride,
const uint16_t *src0,
uint32_t src0_stride,
const uint16_t *src1,
uint32_t src1_stride,
const uint8_t *mask, int w, int h) {
blend_a64_vmask_bn_w8n_sse4_1(dst, dst_stride, src0, src0_stride, src1,
src1_stride, mask, w, h, blend_8_b12);
}
//////////////////////////////////////////////////////////////////////////////
// Dispatch
//////////////////////////////////////////////////////////////////////////////
void aom_highbd_blend_a64_vmask_sse4_1(
uint8_t *dst_8, uint32_t dst_stride, const uint8_t *src0_8,
uint32_t src0_stride, const uint8_t *src1_8, uint32_t src1_stride,
const uint8_t *mask, int w, int h, int bd) {
typedef void (*blend_fn)(uint16_t * dst, uint32_t dst_stride,
const uint16_t *src0, uint32_t src0_stride,
const uint16_t *src1, uint32_t src1_stride,
const uint8_t *mask, int w, int h);
// Dimensions are: bd_index X width_index
static const blend_fn blend[2][2] = {
{
// bd == 8 or 10
blend_a64_vmask_b10_w8n_sse4_1, // w % 8 == 0
blend_a64_vmask_b10_w4_sse4_1, // w == 4
},
{
// bd == 12
blend_a64_vmask_b12_w8n_sse4_1, // w % 8 == 0
blend_a64_vmask_b12_w4_sse4_1, // w == 4
}
};
assert(IMPLIES(src0_8 == dst_8, src0_stride == dst_stride));
assert(IMPLIES(src1_8 == dst_8, src1_stride == dst_stride));
assert(h >= 1);
assert(w >= 1);
assert(IS_POWER_OF_TWO(h));
assert(IS_POWER_OF_TWO(w));
assert(bd == 8 || bd == 10 || bd == 12);
if (UNLIKELY((h | w) & 3)) { // if (w <= 2 || h <= 2)
aom_highbd_blend_a64_vmask_c(dst_8, dst_stride, src0_8, src0_stride, src1_8,
src1_stride, mask, w, h, bd);
} else {
uint16_t *const dst = CONVERT_TO_SHORTPTR(dst_8);
const uint16_t *const src0 = CONVERT_TO_SHORTPTR(src0_8);
const uint16_t *const src1 = CONVERT_TO_SHORTPTR(src1_8);
blend[bd == 12][(w >> 2) & 1](dst, dst_stride, src0, src0_stride, src1,
src1_stride, mask, w, h);
}
}
#endif // CONFIG_AV1_HIGHBITDEPTH
|