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
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
|
/*
* Copyright (c) 2013 Seppo Tomperi
* Copyright (c) 2013 - 2014 Pierre-Edouard Lepere
*
* This file is part of Libav.
*
* Libav is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* Libav is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "config.h"
#include "libavutil/cpu.h"
#include "libavutil/x86/cpu.h"
#include "libavcodec/hevcdsp.h"
#define LFC_FUNC(DIR, DEPTH, OPT) \
void ff_hevc_ ## DIR ## _loop_filter_chroma_ ## DEPTH ## _ ## OPT(uint8_t *pix, ptrdiff_t stride, int *tc, uint8_t *no_p, uint8_t *no_q);
#define LFL_FUNC(DIR, DEPTH, OPT) \
void ff_hevc_ ## DIR ## _loop_filter_luma_ ## DEPTH ## _ ## OPT(uint8_t *pix, ptrdiff_t stride, int beta, int *tc, uint8_t *no_p, uint8_t *no_q);
#define LFC_FUNCS(type, depth) \
LFC_FUNC(h, depth, sse2) \
LFC_FUNC(v, depth, sse2)
#define LFL_FUNCS(type, depth) \
LFL_FUNC(h, depth, ssse3) \
LFL_FUNC(v, depth, ssse3)
LFC_FUNCS(uint8_t, 8)
LFC_FUNCS(uint8_t, 10)
LFL_FUNCS(uint8_t, 8)
LFL_FUNCS(uint8_t, 10)
#define idct_dc_proto(size, bitd, opt) \
void ff_hevc_idct_ ## size ## _dc_add_ ## bitd ## _ ## opt(uint8_t *dst, int16_t *coeffs, ptrdiff_t stride)
idct_dc_proto(4, 8,mmxext);
idct_dc_proto(8, 8,mmxext);
idct_dc_proto(16,8, sse2);
idct_dc_proto(32,8, sse2);
idct_dc_proto(32,8, avx2);
idct_dc_proto(4, 10,mmxext);
idct_dc_proto(8, 10, sse2);
idct_dc_proto(16,10, sse2);
idct_dc_proto(32,10, sse2);
idct_dc_proto(8, 10, avx);
idct_dc_proto(16,10, avx);
idct_dc_proto(32,10, avx);
idct_dc_proto(16,10, avx2);
idct_dc_proto(32,10, avx2);
#define IDCT_DC_FUNCS(W, opt) \
void ff_hevc_idct_ ## W ## _dc_8_ ## opt(int16_t *coeffs); \
void ff_hevc_idct_ ## W ## _dc_10_ ## opt(int16_t *coeffs)
IDCT_DC_FUNCS(4x4, mmxext);
IDCT_DC_FUNCS(8x8, mmxext);
IDCT_DC_FUNCS(8x8, sse2);
IDCT_DC_FUNCS(16x16, sse2);
IDCT_DC_FUNCS(32x32, sse2);
IDCT_DC_FUNCS(16x16, avx2);
IDCT_DC_FUNCS(32x32, avx2);
#define IDCT_FUNCS(opt) \
void ff_hevc_idct_4x4_8_ ## opt(int16_t *coeffs, int col_limit); \
void ff_hevc_idct_4x4_10_ ## opt(int16_t *coeffs, int col_limit); \
void ff_hevc_idct_8x8_8_ ## opt(int16_t *coeffs, int col_limit); \
void ff_hevc_idct_8x8_10_ ## opt(int16_t *coeffs, int col_limit); \
void ff_hevc_idct_16x16_8_ ## opt(int16_t *coeffs, int col_limit); \
void ff_hevc_idct_16x16_10_ ## opt(int16_t *coeffs, int col_limit); \
void ff_hevc_idct_32x32_8_ ## opt(int16_t *coeffs, int col_limit); \
void ff_hevc_idct_32x32_10_ ## opt(int16_t *coeffs, int col_limit);
IDCT_FUNCS(sse2)
IDCT_FUNCS(avx)
void ff_hevc_add_residual_4_8_mmxext(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_8_8_sse2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_16_8_sse2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_32_8_sse2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_8_8_avx(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_16_8_avx(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_32_8_avx(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_32_8_avx2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_4_10_mmxext(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_8_10_sse2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_16_10_sse2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_32_10_sse2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_16_10_avx2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
void ff_hevc_add_residual_32_10_avx2(uint8_t *dst, int16_t *res, ptrdiff_t stride);
#define GET_PIXELS(width, depth, cf) \
void ff_hevc_get_pixels_ ## width ## _ ## depth ## _ ## cf(int16_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer);
GET_PIXELS(4, 8, sse2)
GET_PIXELS(8, 8, sse2)
GET_PIXELS(12, 8, sse2)
GET_PIXELS(16, 8, sse2)
GET_PIXELS(24, 8, sse2)
GET_PIXELS(32, 8, sse2)
GET_PIXELS(48, 8, sse2)
GET_PIXELS(64, 8, sse2)
GET_PIXELS(4, 10, sse2)
GET_PIXELS(8, 10, sse2)
GET_PIXELS(12, 10, sse2)
GET_PIXELS(16, 10, sse2)
GET_PIXELS(24, 10, sse2)
GET_PIXELS(32, 10, sse2)
GET_PIXELS(48, 10, sse2)
GET_PIXELS(64, 10, sse2)
/* those are independent of the bit depth, so declared separately */
#define INTERP_HV_FUNC(width, cf) \
void ff_hevc_qpel_hv_ ## width ## _ ## cf(int16_t *dst, ptrdiff_t dststride, \
int16_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer); \
void ff_hevc_epel_hv_ ## width ## _ ## cf(int16_t *dst, ptrdiff_t dststride, \
int16_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer);
INTERP_HV_FUNC(4, avx)
INTERP_HV_FUNC(8, avx)
INTERP_HV_FUNC(12, avx)
INTERP_HV_FUNC(16, avx)
INTERP_HV_FUNC(24, avx)
INTERP_HV_FUNC(32, avx)
INTERP_HV_FUNC(48, avx)
INTERP_HV_FUNC(64, avx)
#if ARCH_X86_64 && HAVE_AVX_EXTERNAL
#define QPEL_FUNC_HV(width, depth, cf_h, cf_v, cf_hv) \
static void hevc_qpel_hv_ ## width ## _ ## depth ## _ ## cf_hv(int16_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer) \
{ \
const ptrdiff_t stride = FFALIGN(width + 7, 8); \
ff_hevc_qpel_h_ ## width ## _ ## depth ## _ ## cf_h(mcbuffer, 2 * stride, src - 3 * srcstride, srcstride, \
height + 7, mx, my, mcbuffer); \
ff_hevc_qpel_hv_ ## width ## _ ## cf_hv(dst, dststride, mcbuffer + 3 * stride, 2 * stride, \
height, mx, my, mcbuffer); \
}
#else
#define QPEL_FUNC_HV(width, depth, cf_h, cf_v, cf_hv)
#endif /* ARCH_X86_64 && HAVE_AVX_EXTERNAL */
#define QPEL_FUNCS(width, depth, cf_h, cf_v, cf_hv) \
void ff_hevc_qpel_h_ ## width ## _ ## depth ## _ ## cf_h(int16_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer); \
void ff_hevc_qpel_v_ ## width ## _ ## depth ## _ ## cf_v(int16_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer); \
QPEL_FUNC_HV(width, depth, cf_h, cf_v, cf_hv)
QPEL_FUNCS(4, 8, ssse3, ssse3, avx)
QPEL_FUNCS(8, 8, ssse3, ssse3, avx)
QPEL_FUNCS(12, 8, ssse3, ssse3, avx)
QPEL_FUNCS(16, 8, ssse3, ssse3, avx)
QPEL_FUNCS(24, 8, ssse3, ssse3, avx)
QPEL_FUNCS(32, 8, ssse3, ssse3, avx)
QPEL_FUNCS(48, 8, ssse3, ssse3, avx)
QPEL_FUNCS(64, 8, ssse3, ssse3, avx)
QPEL_FUNCS(4, 10, avx, avx, avx)
QPEL_FUNCS(8, 10, avx, avx, avx)
QPEL_FUNCS(12, 10, avx, avx, avx)
QPEL_FUNCS(16, 10, avx, avx, avx)
QPEL_FUNCS(24, 10, avx, avx, avx)
QPEL_FUNCS(32, 10, avx, avx, avx)
QPEL_FUNCS(48, 10, avx, avx, avx)
QPEL_FUNCS(64, 10, avx, avx, avx)
#if ARCH_X86_64 && HAVE_AVX_EXTERNAL
#define EPEL_FUNC_HV(width, depth, cf_h, cf_v, cf_hv) \
static void hevc_epel_hv_ ## width ## _ ## depth ## _ ## cf_hv(int16_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer) \
{ \
const ptrdiff_t stride = FFALIGN(width + 3, 8); \
ff_hevc_epel_h_ ## width ## _ ## depth ## _ ## cf_h(mcbuffer, 2 * stride, src - srcstride, srcstride, \
height + 3, mx, my, mcbuffer); \
ff_hevc_epel_hv_ ## width ## _ ## cf_hv(dst, dststride, mcbuffer + stride, 2 * stride, \
height, mx, my, mcbuffer); \
}
#else
#define EPEL_FUNC_HV(width, depth, cf_h, cf_v, cf_hv)
#endif /* ARCH_X86_64 && HAVE_AVX_EXTERNAL */
#define EPEL_FUNCS(width, depth, cf_h, cf_v, cf_hv) \
void ff_hevc_epel_h_ ## width ## _ ## depth ## _ ## cf_h(int16_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer); \
void ff_hevc_epel_v_ ## width ## _ ## depth ## _ ## cf_v(int16_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int height, int mx, int my, int16_t *mcbuffer); \
EPEL_FUNC_HV(width, depth, cf_h, cf_v, cf_hv)
EPEL_FUNCS(4, 8, ssse3, ssse3, avx)
EPEL_FUNCS(8, 8, ssse3, ssse3, avx)
EPEL_FUNCS(12, 8, ssse3, ssse3, avx)
EPEL_FUNCS(16, 8, ssse3, ssse3, avx)
EPEL_FUNCS(24, 8, ssse3, ssse3, avx)
EPEL_FUNCS(32, 8, ssse3, ssse3, avx)
EPEL_FUNCS(4, 10, avx, avx, avx)
EPEL_FUNCS(8, 10, avx, avx, avx)
EPEL_FUNCS(12, 10, avx, avx, avx)
EPEL_FUNCS(16, 10, avx, avx, avx)
EPEL_FUNCS(24, 10, avx, avx, avx)
EPEL_FUNCS(32, 10, avx, avx, avx)
#define PUT_PRED(width, depth, cf_uw, cf_w) \
void ff_hevc_put_unweighted_pred_ ## width ## _ ## depth ## _ ## cf_uw(uint8_t *dst, ptrdiff_t dststride, \
int16_t *src, ptrdiff_t srcstride, \
int height); \
void ff_hevc_put_unweighted_pred_avg_ ## width ## _ ## depth ## _ ## cf_uw(uint8_t *dst, ptrdiff_t dststride, \
int16_t *src1, int16_t *src2, \
ptrdiff_t srcstride, int height); \
void ff_hevc_put_weighted_pred_ ## width ## _ ## depth ## _ ## cf_w(uint8_t denom, int16_t weight, int16_t offset, \
uint8_t *dst, ptrdiff_t dststride, \
int16_t *src, ptrdiff_t srcstride, \
int height); \
void ff_hevc_put_weighted_pred_avg_ ## width ## _ ## depth ## _ ## cf_w(uint8_t denom, int16_t weight0, int16_t weight1, \
int16_t offset0, int16_t offset1, \
uint8_t *dst, ptrdiff_t dststride, \
int16_t *src0, int16_t *src1, ptrdiff_t srcstride, \
int height);
PUT_PRED(4, 8, sse2, sse4)
PUT_PRED(8, 8, sse2, sse4)
PUT_PRED(12, 8, sse2, sse4)
PUT_PRED(16, 8, sse2, sse4)
PUT_PRED(24, 8, sse2, sse4)
PUT_PRED(32, 8, sse2, sse4)
PUT_PRED(48, 8, sse2, sse4)
PUT_PRED(64, 8, sse2, sse4)
PUT_PRED(4, 10, sse2, sse4)
PUT_PRED(8, 10, sse2, sse4)
PUT_PRED(12, 10, sse2, sse4)
PUT_PRED(16, 10, sse2, sse4)
PUT_PRED(24, 10, sse2, sse4)
PUT_PRED(32, 10, sse2, sse4)
PUT_PRED(48, 10, sse2, sse4)
PUT_PRED(64, 10, sse2, sse4)
void ff_hevc_dsp_init_x86(HEVCDSPContext *c, const int bit_depth)
{
int cpu_flags = av_get_cpu_flags();
#define SET_LUMA_FUNCS(tabname, funcname, depth, cf) \
c->tabname[0] = funcname ## _4_ ## depth ## _ ## cf; \
c->tabname[1] = funcname ## _8_ ## depth ## _ ## cf; \
c->tabname[2] = funcname ## _12_ ## depth ## _ ## cf; \
c->tabname[3] = funcname ## _16_ ## depth ## _ ## cf; \
c->tabname[4] = funcname ## _24_ ## depth ## _ ## cf; \
c->tabname[5] = funcname ## _32_ ## depth ## _ ## cf; \
c->tabname[6] = funcname ## _48_ ## depth ## _ ## cf; \
c->tabname[7] = funcname ## _64_ ## depth ## _ ## cf;
#define SET_CHROMA_FUNCS(tabname, funcname, depth, cf) \
c->tabname[1] = funcname ## _4_ ## depth ## _ ## cf; \
c->tabname[3] = funcname ## _8_ ## depth ## _ ## cf; \
c->tabname[4] = funcname ## _12_ ## depth ## _ ## cf; \
c->tabname[5] = funcname ## _16_ ## depth ## _ ## cf; \
c->tabname[6] = funcname ## _24_ ## depth ## _ ## cf; \
c->tabname[7] = funcname ## _32_ ## depth ## _ ## cf;
#define SET_QPEL_FUNCS(v, h, depth, cf, name) SET_LUMA_FUNCS (put_hevc_qpel[v][h], name, depth, cf)
#define SET_EPEL_FUNCS(v, h, depth, cf, name) SET_CHROMA_FUNCS(put_hevc_epel[v][h], name, depth, cf)
if (bit_depth == 8) {
if (EXTERNAL_MMXEXT(cpu_flags)) {
c->idct_dc[0] = ff_hevc_idct_4x4_dc_8_mmxext;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_8_mmxext;
c->add_residual[0] = ff_hevc_add_residual_4_8_mmxext;
}
if (EXTERNAL_SSE2(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_8_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_8_sse2;
c->add_residual[1] = ff_hevc_add_residual_8_8_sse2;
c->add_residual[2] = ff_hevc_add_residual_16_8_sse2;
c->add_residual[3] = ff_hevc_add_residual_32_8_sse2;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_8_sse2;
c->idct_dc[2] = ff_hevc_idct_16x16_dc_8_sse2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_8_sse2;
c->idct[0] = ff_hevc_idct_4x4_8_sse2;
c->idct[1] = ff_hevc_idct_8x8_8_sse2;
SET_QPEL_FUNCS(0, 0, 8, sse2, ff_hevc_get_pixels);
SET_EPEL_FUNCS(0, 0, 8, sse2, ff_hevc_get_pixels);
SET_LUMA_FUNCS(put_unweighted_pred, ff_hevc_put_unweighted_pred, 8, sse2);
SET_LUMA_FUNCS(put_unweighted_pred_avg, ff_hevc_put_unweighted_pred_avg, 8, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_chroma, ff_hevc_put_unweighted_pred, 8, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_avg_chroma, ff_hevc_put_unweighted_pred_avg, 8, sse2);
}
if (EXTERNAL_SSSE3(cpu_flags)) {
SET_QPEL_FUNCS(0, 1, 8, ssse3, ff_hevc_qpel_h);
SET_QPEL_FUNCS(1, 0, 8, ssse3, ff_hevc_qpel_v);
SET_EPEL_FUNCS(0, 1, 8, ssse3, ff_hevc_epel_h);
SET_EPEL_FUNCS(1, 0, 8, ssse3, ff_hevc_epel_v);
}
if (EXTERNAL_AVX(cpu_flags)) {
c->idct[0] = ff_hevc_idct_4x4_8_avx;
c->idct[1] = ff_hevc_idct_8x8_8_avx;
c->add_residual[1] = ff_hevc_add_residual_8_8_avx;
c->add_residual[2] = ff_hevc_add_residual_16_8_avx;
c->add_residual[3] = ff_hevc_add_residual_32_8_avx;
}
if (EXTERNAL_AVX2(cpu_flags)) {
c->add_residual[3] = ff_hevc_add_residual_32_8_avx2;
}
} else if (bit_depth == 10) {
if (EXTERNAL_MMXEXT(cpu_flags)) {
c->idct_dc[0] = ff_hevc_idct_4x4_dc_10_mmxext;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_10_mmxext;
c->add_residual[0] = ff_hevc_add_residual_4_10_mmxext;
}
if (EXTERNAL_SSE2(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_10_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_10_sse2;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_10_sse2;
c->idct_dc[2] = ff_hevc_idct_16x16_dc_10_sse2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_10_sse2;
c->idct[0] = ff_hevc_idct_4x4_10_sse2;
c->idct[1] = ff_hevc_idct_8x8_10_sse2;
SET_QPEL_FUNCS(0, 0, 10, sse2, ff_hevc_get_pixels);
SET_EPEL_FUNCS(0, 0, 10, sse2, ff_hevc_get_pixels);
SET_LUMA_FUNCS(put_unweighted_pred, ff_hevc_put_unweighted_pred, 10, sse2);
SET_LUMA_FUNCS(put_unweighted_pred_avg, ff_hevc_put_unweighted_pred_avg, 10, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_chroma, ff_hevc_put_unweighted_pred, 10, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_avg_chroma, ff_hevc_put_unweighted_pred_avg, 10, sse2);
c->add_residual[1] = ff_hevc_add_residual_8_10_sse2;
c->add_residual[2] = ff_hevc_add_residual_16_10_sse2;
c->add_residual[3] = ff_hevc_add_residual_32_10_sse2;
}
if (EXTERNAL_AVX(cpu_flags)) {
c->idct[0] = ff_hevc_idct_4x4_10_avx;
c->idct[1] = ff_hevc_idct_8x8_10_avx;
}
if (EXTERNAL_AVX2(cpu_flags)) {
c->add_residual[2] = ff_hevc_add_residual_16_10_avx2;
c->add_residual[3] = ff_hevc_add_residual_32_10_avx2;
}
}
#if ARCH_X86_64
if (bit_depth == 8) {
if (EXTERNAL_SSE2(cpu_flags)) {
c->idct[2] = ff_hevc_idct_16x16_8_sse2;
c->idct[3] = ff_hevc_idct_32x32_8_sse2;
}
if (EXTERNAL_SSSE3(cpu_flags)) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_8_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_8_ssse3;
}
if (EXTERNAL_SSE4(cpu_flags)) {
SET_LUMA_FUNCS(weighted_pred, ff_hevc_put_weighted_pred, 8, sse4);
SET_CHROMA_FUNCS(weighted_pred_chroma, ff_hevc_put_weighted_pred, 8, sse4);
SET_LUMA_FUNCS(weighted_pred_avg, ff_hevc_put_weighted_pred_avg, 8, sse4);
SET_CHROMA_FUNCS(weighted_pred_avg_chroma, ff_hevc_put_weighted_pred_avg, 8, sse4);
}
if (EXTERNAL_AVX(cpu_flags)) {
#if HAVE_AVX_EXTERNAL
SET_QPEL_FUNCS(1, 1, 8, avx, hevc_qpel_hv);
SET_EPEL_FUNCS(1, 1, 8, avx, hevc_epel_hv);
#endif /* HAVE_AVX_EXTERNAL */
c->idct[2] = ff_hevc_idct_16x16_8_avx;
c->idct[3] = ff_hevc_idct_32x32_8_avx;
}
if (EXTERNAL_AVX2(cpu_flags)) {
c->idct_dc[2] = ff_hevc_idct_16x16_dc_8_avx2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_8_avx2;
}
} else if (bit_depth == 10) {
if (EXTERNAL_SSE2(cpu_flags)) {
c->idct[2] = ff_hevc_idct_16x16_10_sse2;
c->idct[3] = ff_hevc_idct_32x32_10_sse2;
}
if (EXTERNAL_SSSE3(cpu_flags)) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_10_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_10_ssse3;
}
if (EXTERNAL_SSE4(cpu_flags)) {
SET_LUMA_FUNCS(weighted_pred, ff_hevc_put_weighted_pred, 10, sse4);
SET_CHROMA_FUNCS(weighted_pred_chroma, ff_hevc_put_weighted_pred, 10, sse4);
SET_LUMA_FUNCS(weighted_pred_avg, ff_hevc_put_weighted_pred_avg, 10, sse4);
SET_CHROMA_FUNCS(weighted_pred_avg_chroma, ff_hevc_put_weighted_pred_avg, 10, sse4);
}
if (EXTERNAL_AVX(cpu_flags)) {
#if HAVE_AVX_EXTERNAL
SET_QPEL_FUNCS(0, 1, 10, avx, ff_hevc_qpel_h);
SET_QPEL_FUNCS(1, 0, 10, avx, ff_hevc_qpel_v);
SET_QPEL_FUNCS(1, 1, 10, avx, hevc_qpel_hv);
SET_EPEL_FUNCS(0, 1, 10, avx, ff_hevc_epel_h);
SET_EPEL_FUNCS(1, 0, 10, avx, ff_hevc_epel_v);
SET_EPEL_FUNCS(1, 1, 10, avx, hevc_epel_hv);
#endif /* HAVE_AVX_EXTERNAL */
c->idct[2] = ff_hevc_idct_16x16_10_avx;
c->idct[3] = ff_hevc_idct_32x32_10_avx;
}
if (EXTERNAL_AVX2(cpu_flags)) {
c->idct_dc[2] = ff_hevc_idct_16x16_dc_10_avx2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_10_avx2;
}
}
#endif /* ARCH_X86_64 */
}
|
