FFmpeg
vvcdsp_init.c
Go to the documentation of this file.
1 /*
2  * VVC DSP init for x86
3  *
4  * Copyright (C) 2022-2024 Nuo Mi
5  * Copyright (c) 2023-2024 Wu Jianhua
6  *
7  * This file is part of FFmpeg.
8  *
9  * FFmpeg is free software; you can redistribute it and/or
10  * modify it under the terms of the GNU Lesser General Public
11  * License as published by the Free Software Foundation; either
12  * version 2.1 of the License, or (at your option) any later version.
13  *
14  * FFmpeg is distributed in the hope that it will be useful,
15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17  * Lesser General Public License for more details.
18  *
19  * You should have received a copy of the GNU Lesser General Public
20  * License along with FFmpeg; if not, write to the Free Software
21  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22  */
23 
24 #include "config.h"
25 
26 #include "libavutil/cpu.h"
27 #include "libavutil/x86/cpu.h"
28 #include "libavcodec/vvc/dec.h"
29 #include "libavcodec/vvc/ctu.h"
30 #include "libavcodec/vvc/dsp.h"
31 #include "libavcodec/x86/h26x/h2656dsp.h"
32 
33 #if ARCH_X86_64
34 
35 #define PUT_PROTOTYPE(name, depth, opt) \
36 void ff_vvc_put_ ## name ## _ ## depth ## _##opt(int16_t *dst, const uint8_t *src, ptrdiff_t srcstride, int height, const int8_t *hf, const int8_t *vf, int width);
37 
38 #define PUT_PROTOTYPES(name, bitd, opt) \
39  PUT_PROTOTYPE(name##2, bitd, opt) \
40  PUT_PROTOTYPE(name##4, bitd, opt) \
41  PUT_PROTOTYPE(name##8, bitd, opt) \
42  PUT_PROTOTYPE(name##12, bitd, opt) \
43  PUT_PROTOTYPE(name##16, bitd, opt) \
44  PUT_PROTOTYPE(name##24, bitd, opt) \
45  PUT_PROTOTYPE(name##32, bitd, opt) \
46  PUT_PROTOTYPE(name##48, bitd, opt) \
47  PUT_PROTOTYPE(name##64, bitd, opt) \
48  PUT_PROTOTYPE(name##128, bitd, opt)
49 
50 #define PUT_BPC_PROTOTYPES(name, opt) \
51  PUT_PROTOTYPES(name, 8, opt) \
52  PUT_PROTOTYPES(name, 10, opt) \
53  PUT_PROTOTYPES(name, 12, opt)
54 
55 #define PUT_TAP_PROTOTYPES(n, opt) \
56  PUT_BPC_PROTOTYPES(n##tap_h, opt) \
57  PUT_BPC_PROTOTYPES(n##tap_v, opt) \
58  PUT_BPC_PROTOTYPES(n##tap_hv, opt)
59 
60 PUT_BPC_PROTOTYPES(pixels, sse4)
61 PUT_BPC_PROTOTYPES(pixels, avx2)
62 
63 PUT_TAP_PROTOTYPES(4, sse4)
64 PUT_TAP_PROTOTYPES(8, sse4)
65 PUT_TAP_PROTOTYPES(4, avx2)
66 PUT_TAP_PROTOTYPES(8, avx2)
67 
68 #define bf(fn, bd, opt) fn##_##bd##_##opt
69 #define BF(fn, bpc, opt) fn##_##bpc##bpc_##opt
70 
71 #define AVG_BPC_PROTOTYPES(bpc, opt) \
72 void BF(ff_vvc_avg, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
73  const int16_t *src0, const int16_t *src1, intptr_t width, intptr_t height, intptr_t pixel_max); \
74 void BF(ff_vvc_w_avg, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
75  const int16_t *src0, const int16_t *src1, intptr_t width, intptr_t height, \
76  intptr_t denom, intptr_t w0, intptr_t w1, intptr_t o0, intptr_t o1, intptr_t pixel_max);
77 
78 #define AVG_PROTOTYPES(bd, opt) \
79 void bf(ff_vvc_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
80  const int16_t *src0, const int16_t *src1, int width, int height); \
81 void bf(ff_vvc_w_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
82  const int16_t *src0, const int16_t *src1, int width, int height, \
83  int denom, int w0, int w1, int o0, int o1);
84 
85 AVG_BPC_PROTOTYPES( 8, avx2)
86 AVG_BPC_PROTOTYPES(16, avx2)
87 
88 AVG_PROTOTYPES( 8, avx2)
89 AVG_PROTOTYPES(10, avx2)
90 AVG_PROTOTYPES(12, avx2)
91 
92 
93 #define DMVR_PROTOTYPES(bd, opt) \
94 void ff_vvc_dmvr_##bd##_##opt(int16_t *dst, const uint8_t *src, ptrdiff_t src_stride, \
95  int height, intptr_t mx, intptr_t my, int width); \
96 void ff_vvc_dmvr_h_##bd##_##opt(int16_t *dst, const uint8_t *src, ptrdiff_t src_stride, \
97  int height, intptr_t mx, intptr_t my, int width); \
98 void ff_vvc_dmvr_v_##bd##_##opt(int16_t *dst, const uint8_t *src, ptrdiff_t src_stride, \
99  int height, intptr_t mx, intptr_t my, int width); \
100 void ff_vvc_dmvr_hv_##bd##_##opt(int16_t *dst, const uint8_t *src, ptrdiff_t src_stride, \
101  int height, intptr_t mx, intptr_t my, int width); \
102 
103 DMVR_PROTOTYPES( 8, avx2)
104 DMVR_PROTOTYPES(10, avx2)
105 DMVR_PROTOTYPES(12, avx2)
106 
107 #define OF_PROTOTYPES(bd, opt) \
108 void ff_vvc_apply_bdof_##bd##_##opt(uint8_t *dst, ptrdiff_t dst_stride, \
109  const int16_t *src0, const int16_t *src1, int w, int h); \
110 
111 OF_PROTOTYPES( 8, avx2)
112 OF_PROTOTYPES(10, avx2)
113 OF_PROTOTYPES(12, avx2)
114 
115 #if ARCH_X86_64 && HAVE_AVX2_EXTERNAL
116 void ff_vvc_apply_bdof_avx2(uint8_t *dst, ptrdiff_t dst_stride, \
117  const int16_t *src0, const int16_t *src1, int w, int h, int pixel_max); \
118 
119 #define OF_FUNC(bd, opt) \
120 void ff_vvc_apply_bdof_##bd##_##opt(uint8_t *dst, ptrdiff_t dst_stride, \
121  const int16_t *src0, const int16_t *src1, int w, int h) \
122 { \
123  ff_vvc_apply_bdof##_##opt(dst, dst_stride, src0, src1, w, h, (1 << bd) - 1); \
124 } \
125 
126 OF_FUNC( 8, avx2)
127 OF_FUNC(10, avx2)
128 OF_FUNC(12, avx2)
129 #endif
130 
131 #define ALF_BPC_PROTOTYPES(bpc, opt) \
132 void BF(ff_vvc_alf_filter_luma, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
133  const uint8_t *src, ptrdiff_t src_stride, ptrdiff_t width, ptrdiff_t height, \
134  const int16_t *filter, const int16_t *clip, ptrdiff_t stride, ptrdiff_t vb_pos, ptrdiff_t pixel_max); \
135 void BF(ff_vvc_alf_filter_chroma, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
136  const uint8_t *src, ptrdiff_t src_stride, ptrdiff_t width, ptrdiff_t height, \
137  const int16_t *filter, const int16_t *clip, ptrdiff_t stride, ptrdiff_t vb_pos, ptrdiff_t pixel_max); \
138 void BF(ff_vvc_alf_classify_grad, bpc, opt)(int *gradient_sum, \
139  const uint8_t *src, ptrdiff_t src_stride, intptr_t width, intptr_t height, intptr_t vb_pos); \
140 void BF(ff_vvc_alf_classify, bpc, opt)(int *class_idx, int *transpose_idx, const int *gradient_sum, \
141  intptr_t width, intptr_t height, intptr_t vb_pos, intptr_t bit_depth); \
142 
143 #define ALF_PROTOTYPES(bpc, bd, opt) \
144 void bf(ff_vvc_alf_filter_luma, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, const uint8_t *src, ptrdiff_t src_stride, \
145  int width, int height, const int16_t *filter, const int16_t *clip, const int vb_pos); \
146 void bf(ff_vvc_alf_filter_chroma, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, const uint8_t *src, ptrdiff_t src_stride, \
147  int width, int height, const int16_t *filter, const int16_t *clip, const int vb_pos); \
148 void bf(ff_vvc_alf_classify, bd, opt)(int *class_idx, int *transpose_idx, \
149  const uint8_t *src, ptrdiff_t src_stride, int width, int height, int vb_pos, int *gradient_tmp); \
150 
151 ALF_BPC_PROTOTYPES(8, avx2)
152 ALF_BPC_PROTOTYPES(16, avx2)
153 
154 ALF_PROTOTYPES(8, 8, avx2)
155 ALF_PROTOTYPES(16, 10, avx2)
156 ALF_PROTOTYPES(16, 12, avx2)
157 
158 #if ARCH_X86_64
159 #if HAVE_SSE4_EXTERNAL
160 #define FW_PUT(name, depth, opt) \
161 void ff_vvc_put_ ## name ## _ ## depth ## _##opt(int16_t *dst, const uint8_t *src, ptrdiff_t srcstride, \
162  int height, const int8_t *hf, const int8_t *vf, int width) \
163 { \
164  ff_h2656_put_## name ## _ ## depth ## _##opt(dst, 2 * MAX_PB_SIZE, src, srcstride, height, hf, vf, width); \
165 }
166 
167 #define FW_PUT_TAP(fname, bitd, opt ) \
168  FW_PUT(fname##4, bitd, opt ) \
169  FW_PUT(fname##8, bitd, opt ) \
170  FW_PUT(fname##16, bitd, opt ) \
171  FW_PUT(fname##32, bitd, opt ) \
172  FW_PUT(fname##64, bitd, opt ) \
173  FW_PUT(fname##128, bitd, opt ) \
174 
175 #define FW_PUT_4TAP(fname, bitd, opt) \
176  FW_PUT(fname ## 2, bitd, opt) \
177  FW_PUT_TAP(fname, bitd, opt)
178 
179 #define FW_PUT_4TAP_SSE4(bitd) \
180  FW_PUT_4TAP(pixels, bitd, sse4) \
181  FW_PUT_4TAP(4tap_h, bitd, sse4) \
182  FW_PUT_4TAP(4tap_v, bitd, sse4) \
183  FW_PUT_4TAP(4tap_hv, bitd, sse4)
184 
185 #define FW_PUT_8TAP_SSE4(bitd) \
186  FW_PUT_TAP(8tap_h, bitd, sse4) \
187  FW_PUT_TAP(8tap_v, bitd, sse4) \
188  FW_PUT_TAP(8tap_hv, bitd, sse4)
189 
190 #define FW_PUT_SSE4(bitd) \
191  FW_PUT_4TAP_SSE4(bitd) \
192  FW_PUT_8TAP_SSE4(bitd)
193 
194 FW_PUT_SSE4( 8)
195 FW_PUT_SSE4(10)
196 FW_PUT_SSE4(12)
197 #endif
198 
199 #if HAVE_AVX2_EXTERNAL
200 #define FW_PUT_TAP_AVX2(n, bitd) \
201  FW_PUT(n ## tap_h32, bitd, avx2) \
202  FW_PUT(n ## tap_h64, bitd, avx2) \
203  FW_PUT(n ## tap_h128, bitd, avx2) \
204  FW_PUT(n ## tap_v32, bitd, avx2) \
205  FW_PUT(n ## tap_v64, bitd, avx2) \
206  FW_PUT(n ## tap_v128, bitd, avx2)
207 
208 #define FW_PUT_AVX2(bitd) \
209  FW_PUT(pixels32, bitd, avx2) \
210  FW_PUT(pixels64, bitd, avx2) \
211  FW_PUT(pixels128, bitd, avx2) \
212  FW_PUT_TAP_AVX2(4, bitd) \
213  FW_PUT_TAP_AVX2(8, bitd) \
214 
215 FW_PUT_AVX2( 8)
216 FW_PUT_AVX2(10)
217 FW_PUT_AVX2(12)
218 
219 #define FW_PUT_TAP_16BPC_AVX2(n, bitd) \
220  FW_PUT(n ## tap_h16, bitd, avx2) \
221  FW_PUT(n ## tap_v16, bitd, avx2) \
222  FW_PUT(n ## tap_hv16, bitd, avx2) \
223  FW_PUT(n ## tap_hv32, bitd, avx2) \
224  FW_PUT(n ## tap_hv64, bitd, avx2) \
225  FW_PUT(n ## tap_hv128, bitd, avx2)
226 
227 #define FW_PUT_16BPC_AVX2(bitd) \
228  FW_PUT(pixels16, bitd, avx2) \
229  FW_PUT_TAP_16BPC_AVX2(4, bitd) \
230  FW_PUT_TAP_16BPC_AVX2(8, bitd)
231 
232 FW_PUT_16BPC_AVX2(10)
233 FW_PUT_16BPC_AVX2(12)
234 
235 #define AVG_FUNCS(bpc, bd, opt) \
236 void bf(ff_vvc_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
237  const int16_t *src0, const int16_t *src1, int width, int height) \
238 { \
239  BF(ff_vvc_avg, bpc, opt)(dst, dst_stride, src0, src1, width, height, (1 << bd) - 1); \
240 } \
241 void bf(ff_vvc_w_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \
242  const int16_t *src0, const int16_t *src1, int width, int height, \
243  int denom, int w0, int w1, int o0, int o1) \
244 { \
245  BF(ff_vvc_w_avg, bpc, opt)(dst, dst_stride, src0, src1, width, height, \
246  denom, w0, w1, o0, o1, (1 << bd) - 1); \
247 }
248 
249 AVG_FUNCS(8, 8, avx2)
250 AVG_FUNCS(16, 10, avx2)
251 AVG_FUNCS(16, 12, avx2)
252 
253 #define ALF_FUNCS(bpc, bd, opt) \
254 void bf(ff_vvc_alf_filter_luma, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, const uint8_t *src, ptrdiff_t src_stride, \
255  int width, int height, const int16_t *filter, const int16_t *clip, const int vb_pos) \
256 { \
257  const int param_stride = (width >> 2) * ALF_NUM_COEFF_LUMA; \
258  BF(ff_vvc_alf_filter_luma, bpc, opt)(dst, dst_stride, src, src_stride, width, height, \
259  filter, clip, param_stride, vb_pos, (1 << bd) - 1); \
260 } \
261 void bf(ff_vvc_alf_filter_chroma, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, const uint8_t *src, ptrdiff_t src_stride, \
262  int width, int height, const int16_t *filter, const int16_t *clip, const int vb_pos) \
263 { \
264  BF(ff_vvc_alf_filter_chroma, bpc, opt)(dst, dst_stride, src, src_stride, width, height, \
265  filter, clip, 0, vb_pos,(1 << bd) - 1); \
266 } \
267 void bf(ff_vvc_alf_classify, bd, opt)(int *class_idx, int *transpose_idx, \
268  const uint8_t *src, ptrdiff_t src_stride, int width, int height, int vb_pos, int *gradient_tmp) \
269 { \
270  BF(ff_vvc_alf_classify_grad, bpc, opt)(gradient_tmp, src, src_stride, width, height, vb_pos); \
271  BF(ff_vvc_alf_classify, bpc, opt)(class_idx, transpose_idx, gradient_tmp, width, height, vb_pos, bd); \
272 } \
273 
274 ALF_FUNCS(8, 8, avx2)
275 ALF_FUNCS(16, 10, avx2)
276 ALF_FUNCS(16, 12, avx2)
277 
278 #endif
279 
280 #define PEL_LINK(dst, C, W, idx1, idx2, name, D, opt) \
281  dst[C][W][idx1][idx2] = ff_vvc_put_## name ## _ ## D ## _##opt; \
282  dst ## _uni[C][W][idx1][idx2] = ff_h2656_put_uni_ ## name ## _ ## D ## _##opt; \
283 
284 #define MC_TAP_LINKS(pointer, C, my, mx, fname, bitd, opt ) \
285  PEL_LINK(pointer, C, 1, my , mx , fname##4 , bitd, opt ); \
286  PEL_LINK(pointer, C, 2, my , mx , fname##8 , bitd, opt ); \
287  PEL_LINK(pointer, C, 3, my , mx , fname##16, bitd, opt ); \
288  PEL_LINK(pointer, C, 4, my , mx , fname##32, bitd, opt ); \
289  PEL_LINK(pointer, C, 5, my , mx , fname##64, bitd, opt ); \
290  PEL_LINK(pointer, C, 6, my , mx , fname##128, bitd, opt );
291 
292 #define MC_8TAP_LINKS(pointer, my, mx, fname, bitd, opt) \
293  MC_TAP_LINKS(pointer, LUMA, my, mx, fname, bitd, opt)
294 
295 #define MC_8TAP_LINKS_SSE4(bd) \
296  MC_8TAP_LINKS(c->inter.put, 0, 0, pixels, bd, sse4); \
297  MC_8TAP_LINKS(c->inter.put, 0, 1, 8tap_h, bd, sse4); \
298  MC_8TAP_LINKS(c->inter.put, 1, 0, 8tap_v, bd, sse4); \
299  MC_8TAP_LINKS(c->inter.put, 1, 1, 8tap_hv, bd, sse4)
300 
301 #define MC_4TAP_LINKS(pointer, my, mx, fname, bitd, opt) \
302  PEL_LINK(pointer, CHROMA, 0, my , mx , fname##2 , bitd, opt ); \
303  MC_TAP_LINKS(pointer, CHROMA, my, mx, fname, bitd, opt) \
304 
305 #define MC_4TAP_LINKS_SSE4(bd) \
306  MC_4TAP_LINKS(c->inter.put, 0, 0, pixels, bd, sse4); \
307  MC_4TAP_LINKS(c->inter.put, 0, 1, 4tap_h, bd, sse4); \
308  MC_4TAP_LINKS(c->inter.put, 1, 0, 4tap_v, bd, sse4); \
309  MC_4TAP_LINKS(c->inter.put, 1, 1, 4tap_hv, bd, sse4)
310 
311 #define MC_LINK_SSE4(bd) \
312  MC_4TAP_LINKS_SSE4(bd) \
313  MC_8TAP_LINKS_SSE4(bd)
314 
315 #define MC_TAP_LINKS_AVX2(C,tap,bd) do { \
316  PEL_LINK(c->inter.put, C, 4, 0, 0, pixels32, bd, avx2) \
317  PEL_LINK(c->inter.put, C, 5, 0, 0, pixels64, bd, avx2) \
318  PEL_LINK(c->inter.put, C, 6, 0, 0, pixels128, bd, avx2) \
319  PEL_LINK(c->inter.put, C, 4, 0, 1, tap##tap_h32, bd, avx2) \
320  PEL_LINK(c->inter.put, C, 5, 0, 1, tap##tap_h64, bd, avx2) \
321  PEL_LINK(c->inter.put, C, 6, 0, 1, tap##tap_h128, bd, avx2) \
322  PEL_LINK(c->inter.put, C, 4, 1, 0, tap##tap_v32, bd, avx2) \
323  PEL_LINK(c->inter.put, C, 5, 1, 0, tap##tap_v64, bd, avx2) \
324  PEL_LINK(c->inter.put, C, 6, 1, 0, tap##tap_v128, bd, avx2) \
325  } while (0)
326 
327 #define MC_LINKS_AVX2(bd) \
328  MC_TAP_LINKS_AVX2(LUMA, 8, bd); \
329  MC_TAP_LINKS_AVX2(CHROMA, 4, bd);
330 
331 #define MC_TAP_LINKS_16BPC_AVX2(C, tap, bd) do { \
332  PEL_LINK(c->inter.put, C, 3, 0, 0, pixels16, bd, avx2) \
333  PEL_LINK(c->inter.put, C, 3, 0, 1, tap##tap_h16, bd, avx2) \
334  PEL_LINK(c->inter.put, C, 3, 1, 0, tap##tap_v16, bd, avx2) \
335  PEL_LINK(c->inter.put, C, 3, 1, 1, tap##tap_hv16, bd, avx2) \
336  PEL_LINK(c->inter.put, C, 4, 1, 1, tap##tap_hv32, bd, avx2) \
337  PEL_LINK(c->inter.put, C, 5, 1, 1, tap##tap_hv64, bd, avx2) \
338  PEL_LINK(c->inter.put, C, 6, 1, 1, tap##tap_hv128, bd, avx2) \
339  } while (0)
340 
341 #define MC_LINKS_16BPC_AVX2(bd) \
342  MC_TAP_LINKS_16BPC_AVX2(LUMA, 8, bd); \
343  MC_TAP_LINKS_16BPC_AVX2(CHROMA, 4, bd);
344 
345 #define AVG_INIT(bd, opt) do { \
346  c->inter.avg = bf(ff_vvc_avg, bd, opt); \
347  c->inter.w_avg = bf(ff_vvc_w_avg, bd, opt); \
348 } while (0)
349 
350 #define DMVR_INIT(bd) do { \
351  c->inter.dmvr[0][0] = ff_vvc_dmvr_##bd##_avx2; \
352  c->inter.dmvr[0][1] = ff_vvc_dmvr_h_##bd##_avx2; \
353  c->inter.dmvr[1][0] = ff_vvc_dmvr_v_##bd##_avx2; \
354  c->inter.dmvr[1][1] = ff_vvc_dmvr_hv_##bd##_avx2; \
355 } while (0)
356 
357 #define OF_INIT(bd) do { \
358  c->inter.apply_bdof = ff_vvc_apply_bdof_##bd##_avx2; \
359 } while (0)
360 
361 #define ALF_INIT(bd) do { \
362  c->alf.filter[LUMA] = ff_vvc_alf_filter_luma_##bd##_avx2; \
363  c->alf.filter[CHROMA] = ff_vvc_alf_filter_chroma_##bd##_avx2; \
364  c->alf.classify = ff_vvc_alf_classify_##bd##_avx2; \
365 } while (0)
366 
367 int ff_vvc_sad_avx2(const int16_t *src0, const int16_t *src1, int dx, int dy, int block_w, int block_h);
368 #define SAD_INIT() c->inter.sad = ff_vvc_sad_avx2
369 #endif
370 
371 
372 #endif // ARCH_X86_64
373 
374 void ff_vvc_dsp_init_x86(VVCDSPContext *const c, const int bd)
375 {
376 #if ARCH_X86_64
377  const int cpu_flags = av_get_cpu_flags();
378 
379  switch (bd) {
380  case 8:
381  if (EXTERNAL_SSE4(cpu_flags)) {
382  MC_LINK_SSE4(8);
383  }
384  if (EXTERNAL_AVX2_FAST(cpu_flags)) {
385  ALF_INIT(8);
386  AVG_INIT(8, avx2);
387  MC_LINKS_AVX2(8);
388  OF_INIT(8);
389  DMVR_INIT(8);
390  SAD_INIT();
391  }
392  break;
393  case 10:
394  if (EXTERNAL_SSE4(cpu_flags)) {
395  MC_LINK_SSE4(10);
396  }
397  if (EXTERNAL_AVX2_FAST(cpu_flags)) {
398  ALF_INIT(10);
399  AVG_INIT(10, avx2);
400  MC_LINKS_AVX2(10);
401  MC_LINKS_16BPC_AVX2(10);
402  OF_INIT(10);
403  DMVR_INIT(10);
404  SAD_INIT();
405  }
406  break;
407  case 12:
408  if (EXTERNAL_SSE4(cpu_flags)) {
409  MC_LINK_SSE4(12);
410  }
411  if (EXTERNAL_AVX2_FAST(cpu_flags)) {
412  ALF_INIT(12);
413  AVG_INIT(12, avx2);
414  MC_LINKS_AVX2(12);
415  MC_LINKS_16BPC_AVX2(12);
416  OF_INIT(12);
417  DMVR_INIT(12);
418  SAD_INIT();
419  }
420  break;
421  default:
422  break;
423  }
424 #endif
425 }
cpu.h
AVG_PROTOTYPES
#define AVG_PROTOTYPES(bd, opt)
Definition: vvcdsp_init.c:30
src1
const pixel * src1
Definition: h264pred_template.c:421
EXTERNAL_AVX2_FAST
#define EXTERNAL_AVX2_FAST(flags)
Definition: cpu.h:79
w
uint8_t w
Definition: llviddspenc.c:38
av_get_cpu_flags
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:107
cpu_flags
static atomic_int cpu_flags
Definition: cpu.c:56
dsp.h
c
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
Definition: undefined.txt:32
h2656dsp.h
dst
uint8_t ptrdiff_t const uint8_t ptrdiff_t int intptr_t intptr_t int int16_t * dst
Definition: dsp.h:83
cpu.h
EXTERNAL_SSE4
#define EXTERNAL_SSE4(flags)
Definition: cpu.h:68
src0
const pixel *const src0
Definition: h264pred_template.c:420
ff_vvc_dsp_init_x86
void ff_vvc_dsp_init_x86(VVCDSPContext *const c, const int bd)
Definition: vvcdsp_init.c:374
h
h
Definition: vp9dsp_template.c:2070
ctu.h
dec.h
VVCDSPContext
Definition: dsp.h:169
Generated on Fri Nov 15 2024 19:22:11 for FFmpeg by   doxygen 1.8.17