Mercurial > cgi-bin > hgwebdir.cgi > PR > Applications > VSs > VSs__H264__App

comparison libavcodec/h264_pred.c @ 2:897f711a7157

find changesets by author, revision, files, or words in the commit message
rearrange to work with autoconf
author Nina Engelhardt <nengel@mailbox.tu-berlin.de>
date Tue, 25 Sep 2012 15:55:33 +0200
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:4f28d62ed14c
1 /*
2 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
3 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 /**
23 * @file
24 * H.264 / AVC / MPEG4 part10 prediction functions.
25 * @author Michael Niedermayer <michaelni@gmx.at>
26 */
27
28 #include "avcodec.h"
29 #include "h264_pred.h"
30 //#include "dsputil.h"
31
32 static void pred4x4_vertical_c(uint8_t *src, uint8_t *topright, int stride){
33 (void) topright;
34 const uint32_t a= ((uint32_t*)(src-stride))[0];
35 ((uint32_t*)(src+0*stride))[0]= a;
36 ((uint32_t*)(src+1*stride))[0]= a;
37 ((uint32_t*)(src+2*stride))[0]= a;
38 ((uint32_t*)(src+3*stride))[0]= a;
39 }
40
41 static void pred4x4_horizontal_c(uint8_t *src, uint8_t *topright, int stride){
42 (void) topright;
43 ((uint32_t*)(src+0*stride))[0]= src[-1+0*stride]*0x01010101;
44 ((uint32_t*)(src+1*stride))[0]= src[-1+1*stride]*0x01010101;
45 ((uint32_t*)(src+2*stride))[0]= src[-1+2*stride]*0x01010101;
46 ((uint32_t*)(src+3*stride))[0]= src[-1+3*stride]*0x01010101;
47 }
48
49 static void pred4x4_dc_c(uint8_t *src, uint8_t *topright, int stride){
50 (void) topright;
51 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]
52 + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;
53
54 ((uint32_t*)(src+0*stride))[0]=
55 ((uint32_t*)(src+1*stride))[0]=
56 ((uint32_t*)(src+2*stride))[0]=
57 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
58 }
59
60 static void pred4x4_left_dc_c(uint8_t *src, uint8_t *topright, int stride){
61 (void) topright;
62 const int dc= ( src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;
63
64 ((uint32_t*)(src+0*stride))[0]=
65 ((uint32_t*)(src+1*stride))[0]=
66 ((uint32_t*)(src+2*stride))[0]=
67 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
68 }
69
70 static void pred4x4_top_dc_c(uint8_t *src, uint8_t *topright, int stride){
71 (void) topright;
72 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;
73
74 ((uint32_t*)(src+0*stride))[0]=
75 ((uint32_t*)(src+1*stride))[0]=
76 ((uint32_t*)(src+2*stride))[0]=
77 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
78 }
79
80 static void pred4x4_128_dc_c(uint8_t *src, uint8_t *topright, int stride){
81 (void) topright;
82 ((uint32_t*)(src+0*stride))[0]=
83 ((uint32_t*)(src+1*stride))[0]=
84 ((uint32_t*)(src+2*stride))[0]=
85 ((uint32_t*)(src+3*stride))[0]= 128U*0x01010101U;
86 }
87
88
89 #define LOAD_TOP_RIGHT_EDGE\
90 const int av_unused t4= topright[0];\
91 const int av_unused t5= topright[1];\
92 const int av_unused t6= topright[2];\
93 const int av_unused t7= topright[3];\
94
95 #define LOAD_DOWN_LEFT_EDGE\
96 const int av_unused l4= src[-1+4*stride];\
97 const int av_unused l5= src[-1+5*stride];\
98 const int av_unused l6= src[-1+6*stride];\
99 const int av_unused l7= src[-1+7*stride];\
100
101 #define LOAD_LEFT_EDGE\
102 const int av_unused l0= src[-1+0*stride];\
103 const int av_unused l1= src[-1+1*stride];\
104 const int av_unused l2= src[-1+2*stride];\
105 const int av_unused l3= src[-1+3*stride];\
106
107 #define LOAD_TOP_EDGE\
108 const int av_unused t0= src[ 0-1*stride];\
109 const int av_unused t1= src[ 1-1*stride];\
110 const int av_unused t2= src[ 2-1*stride];\
111 const int av_unused t3= src[ 3-1*stride];\
112
113 static void pred4x4_down_right_c(uint8_t *src, uint8_t *topright, int stride){
114 (void) topright;
115 const int lt= src[-1-1*stride];
116 LOAD_TOP_EDGE
117 LOAD_LEFT_EDGE
118
119 src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2;
120 src[0+2*stride]=
121 src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2;
122 src[0+1*stride]=
123 src[1+2*stride]=
124 src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2;
125 src[0+0*stride]=
126 src[1+1*stride]=
127 src[2+2*stride]=
128 src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
129 src[1+0*stride]=
130 src[2+1*stride]=
131 src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2;
132 src[2+0*stride]=
133 src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
134 src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;
135 }
136
137 static void pred4x4_down_left_c(uint8_t *src, uint8_t *topright, int stride){
138 LOAD_TOP_EDGE
139 LOAD_TOP_RIGHT_EDGE
140 // LOAD_LEFT_EDGE
141
142 src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2;
143 src[1+0*stride]=
144 src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2;
145 src[2+0*stride]=
146 src[1+1*stride]=
147 src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2;
148 src[3+0*stride]=
149 src[2+1*stride]=
150 src[1+2*stride]=
151 src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2;
152 src[3+1*stride]=
153 src[2+2*stride]=
154 src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2;
155 src[3+2*stride]=
156 src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2;
157 src[3+3*stride]=(t6 + 3*t7 + 2)>>2;
158 }
159
160 static void pred4x4_vertical_right_c(uint8_t *src, uint8_t *topright, int stride){
161 (void) topright;
162 const int lt= src[-1-1*stride];
163 LOAD_TOP_EDGE
164 LOAD_LEFT_EDGE
165
166 src[0+0*stride]=
167 src[1+2*stride]=(lt + t0 + 1)>>1;
168 src[1+0*stride]=
169 src[2+2*stride]=(t0 + t1 + 1)>>1;
170 src[2+0*stride]=
171 src[3+2*stride]=(t1 + t2 + 1)>>1;
172 src[3+0*stride]=(t2 + t3 + 1)>>1;
173 src[0+1*stride]=
174 src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
175 src[1+1*stride]=
176 src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;
177 src[2+1*stride]=
178 src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;
179 src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;
180 src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
181 src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
182 }
183
184 static void pred4x4_vertical_left_c(uint8_t *src, uint8_t *topright, int stride){
185 LOAD_TOP_EDGE
186 LOAD_TOP_RIGHT_EDGE
187
188 src[0+0*stride]=(t0 + t1 + 1)>>1;
189 src[1+0*stride]=
190 src[0+2*stride]=(t1 + t2 + 1)>>1;
191 src[2+0*stride]=
192 src[1+2*stride]=(t2 + t3 + 1)>>1;
193 src[3+0*stride]=
194 src[2+2*stride]=(t3 + t4+ 1)>>1;
195 src[3+2*stride]=(t4 + t5+ 1)>>1;
196 src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
197 src[1+1*stride]=
198 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
199 src[2+1*stride]=
200 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
201 src[3+1*stride]=
202 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
203 src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
204 }
205
206 static void pred4x4_horizontal_up_c(uint8_t *src, uint8_t *topright, int stride){
207 (void) topright;
208 LOAD_LEFT_EDGE
209
210 src[0+0*stride]=(l0 + l1 + 1)>>1;
211 src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;
212 src[2+0*stride]=
213 src[0+1*stride]=(l1 + l2 + 1)>>1;
214 src[3+0*stride]=
215 src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;
216 src[2+1*stride]=
217 src[0+2*stride]=(l2 + l3 + 1)>>1;
218 src[3+1*stride]=
219 src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;
220 src[3+2*stride]=
221 src[1+3*stride]=
222 src[0+3*stride]=
223 src[2+2*stride]=
224 src[2+3*stride]=
225 src[3+3*stride]=l3;
226 }
227
228
229 static void pred4x4_horizontal_down_c(uint8_t *src, uint8_t *topright, int stride){
230 (void) topright;
231 const int lt= src[-1-1*stride];
232 LOAD_TOP_EDGE
233 LOAD_LEFT_EDGE
234
235 src[0+0*stride]=
236 src[2+1*stride]=(lt + l0 + 1)>>1;
237 src[1+0*stride]=
238 src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2;
239 src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2;
240 src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2;
241 src[0+1*stride]=
242 src[2+2*stride]=(l0 + l1 + 1)>>1;
243 src[1+1*stride]=
244 src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
245 src[0+2*stride]=
246 src[2+3*stride]=(l1 + l2+ 1)>>1;
247 src[1+2*stride]=
248 src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
249 src[0+3*stride]=(l2 + l3 + 1)>>1;
250 src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
251 }
252
253 static void pred16x16_vertical_c(uint8_t *src, int stride){
254 int i;
255 const uint32_t a= ((uint32_t*)(src-stride))[0];
256 const uint32_t b= ((uint32_t*)(src-stride))[1];
257 const uint32_t c= ((uint32_t*)(src-stride))[2];
258 const uint32_t d= ((uint32_t*)(src-stride))[3];
259
260 for(i=0; i<16; i++){
261 ((uint32_t*)(src+i*stride))[0]= a;
262 ((uint32_t*)(src+i*stride))[1]= b;
263 ((uint32_t*)(src+i*stride))[2]= c;
264 ((uint32_t*)(src+i*stride))[3]= d;
265 }
266 }
267
268 static void pred16x16_horizontal_c(uint8_t *src, int stride){
269 int i;
270
271 for(i=0; i<16; i++){
272 ((uint32_t*)(src+i*stride))[0]=
273 ((uint32_t*)(src+i*stride))[1]=
274 ((uint32_t*)(src+i*stride))[2]=
275 ((uint32_t*)(src+i*stride))[3]= src[-1+i*stride]*0x01010101;
276 }
277 }
278
279 static void pred16x16_dc_c(uint8_t *src, int stride){
280 int i, dc=0;
281
282 for(i=0;i<16; i++){
283 dc+= src[-1+i*stride];
284 }
285
286 for(i=0;i<16; i++){
287 dc+= src[i-stride];
288 }
289
290 dc= 0x01010101*((dc + 16)>>5);
291
292 for(i=0; i<16; i++){
293 ((uint32_t*)(src+i*stride))[0]=
294 ((uint32_t*)(src+i*stride))[1]=
295 ((uint32_t*)(src+i*stride))[2]=
296 ((uint32_t*)(src+i*stride))[3]= dc;
297 }
298 }
299
300 static void pred16x16_left_dc_c(uint8_t *src, int stride){
301 int i, dc=0;
302
303 for(i=0;i<16; i++){
304 dc+= src[-1+i*stride];
305 }
306
307 dc= 0x01010101*((dc + 8)>>4);
308
309 for(i=0; i<16; i++){
310 ((uint32_t*)(src+i*stride))[0]=
311 ((uint32_t*)(src+i*stride))[1]=
312 ((uint32_t*)(src+i*stride))[2]=
313 ((uint32_t*)(src+i*stride))[3]= dc;
314 }
315 }
316
317 static void pred16x16_top_dc_c(uint8_t *src, int stride){
318 int i, dc=0;
319
320 for(i=0;i<16; i++){
321 dc+= src[i-stride];
322 }
323 dc= 0x01010101*((dc + 8)>>4);
324
325 for(i=0; i<16; i++){
326 ((uint32_t*)(src+i*stride))[0]=
327 ((uint32_t*)(src+i*stride))[1]=
328 ((uint32_t*)(src+i*stride))[2]=
329 ((uint32_t*)(src+i*stride))[3]= dc;
330 }
331 }
332
333 static void pred16x16_128_dc_c(uint8_t *src, int stride){
334 int i;
335
336 for(i=0; i<16; i++){
337 ((uint32_t*)(src+i*stride))[0]=
338 ((uint32_t*)(src+i*stride))[1]=
339 ((uint32_t*)(src+i*stride))[2]=
340 ((uint32_t*)(src+i*stride))[3]= 0x01010101U*128U;
341 }
342 }
343
344 static inline void pred16x16_plane_compat_c(uint8_t *src, int stride, const int svq3, const int rv40){
345 int i, j, k;
346 int a;
347 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
348 const uint8_t * const src0 = src+7-stride;
349 const uint8_t *src1 = src+8*stride-1;
350 const uint8_t *src2 = src1-2*stride; // == src+6*stride-1;
351 int H = src0[1] - src0[-1];
352 int V = src1[0] - src2[ 0];
353 for(k=2; k<=8; ++k) {
354 src1 += stride; src2 -= stride;
355 H += k*(src0[k] - src0[-k]);
356 V += k*(src1[0] - src2[ 0]);
357 }
358 if(svq3){
359 H = ( 5*(H/4) ) / 16;
360 V = ( 5*(V/4) ) / 16;
361
362 /* required for 100% accuracy */
363 i = H; H = V; V = i;
364 }else if(rv40){
365 H = ( H + (H>>2) ) >> 4;
366 V = ( V + (V>>2) ) >> 4;
367 }else{
368 H = ( 5*H+32 ) >> 6;
369 V = ( 5*V+32 ) >> 6;
370 }
371
372 a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);
373 for(j=16; j>0; --j) {
374 int b = a;
375 a += V;
376 for(i=-16; i<0; i+=4) {
377 src[16+i] = cm[ (b ) >> 5 ];
378 src[17+i] = cm[ (b+ H) >> 5 ];
379 src[18+i] = cm[ (b+2*H) >> 5 ];
380 src[19+i] = cm[ (b+3*H) >> 5 ];
381 b += 4*H;
382 }
383 src += stride;
384 }
385 }
386
387 static void pred16x16_plane_c(uint8_t *src, int stride){
388 pred16x16_plane_compat_c(src, stride, 0, 0);
389 }
390
391
392 static void pred8x8_vertical_c(uint8_t *src, int stride){
393 int i;
394 const uint32_t a= ((uint32_t*)(src-stride))[0];
395 const uint32_t b= ((uint32_t*)(src-stride))[1];
396
397 for(i=0; i<8; i++){
398 ((uint32_t*)(src+i*stride))[0]= a;
399 ((uint32_t*)(src+i*stride))[1]= b;
400 }
401 }
402
403 static void pred8x8_horizontal_c(uint8_t *src, int stride){
404 int i;
405
406 for(i=0; i<8; i++){
407 ((uint32_t*)(src+i*stride))[0]=
408 ((uint32_t*)(src+i*stride))[1]= src[-1+i*stride]*0x01010101;
409 }
410 }
411
412 static void pred8x8_128_dc_c(uint8_t *src, int stride){
413 int i;
414
415 for(i=0; i<8; i++){
416 ((uint32_t*)(src+i*stride))[0]=
417 ((uint32_t*)(src+i*stride))[1]= 0x01010101U*128U;
418 }
419 }
420
421 static void pred8x8_left_dc_c(uint8_t *src, int stride){
422 int i;
423 int dc0, dc2;
424
425 dc0=dc2=0;
426 for(i=0;i<4; i++){
427 dc0+= src[-1+i*stride];
428 dc2+= src[-1+(i+4)*stride];
429 }
430 dc0= 0x01010101*((dc0 + 2)>>2);
431 dc2= 0x01010101*((dc2 + 2)>>2);
432
433 for(i=0; i<4; i++){
434 ((uint32_t*)(src+i*stride))[0]=
435 ((uint32_t*)(src+i*stride))[1]= dc0;
436 }
437 for(i=4; i<8; i++){
438 ((uint32_t*)(src+i*stride))[0]=
439 ((uint32_t*)(src+i*stride))[1]= dc2;
440 }
441 }
442
443
444 static void pred8x8_top_dc_c(uint8_t *src, int stride){
445 int i;
446 int dc0, dc1;
447
448 dc0=dc1=0;
449 for(i=0;i<4; i++){
450 dc0+= src[i-stride];
451 dc1+= src[4+i-stride];
452 }
453 dc0= 0x01010101*((dc0 + 2)>>2);
454 dc1= 0x01010101*((dc1 + 2)>>2);
455
456 for(i=0; i<4; i++){
457 ((uint32_t*)(src+i*stride))[0]= dc0;
458 ((uint32_t*)(src+i*stride))[1]= dc1;
459 }
460 for(i=4; i<8; i++){
461 ((uint32_t*)(src+i*stride))[0]= dc0;
462 ((uint32_t*)(src+i*stride))[1]= dc1;
463 }
464 }
465
466 static void pred8x8_dc_c(uint8_t *src, int stride){
467 int i;
468 int dc0, dc1, dc2, dc3;
469
470 dc0=dc1=dc2=0;
471 for(i=0;i<4; i++){
472 dc0+= src[-1+i*stride] + src[i-stride];
473 dc1+= src[4+i-stride];
474 dc2+= src[-1+(i+4)*stride];
475 }
476 dc3= 0x01010101*((dc1 + dc2 + 4)>>3);
477 dc0= 0x01010101*((dc0 + 4)>>3);
478 dc1= 0x01010101*((dc1 + 2)>>2);
479 dc2= 0x01010101*((dc2 + 2)>>2);
480
481 for(i=0; i<4; i++){
482 ((uint32_t*)(src+i*stride))[0]= dc0;
483 ((uint32_t*)(src+i*stride))[1]= dc1;
484 }
485 for(i=4; i<8; i++){
486 ((uint32_t*)(src+i*stride))[0]= dc2;
487 ((uint32_t*)(src+i*stride))[1]= dc3;
488 }
489 }
490
491 //the following 4 function should not be optimized!
492 static void pred8x8_mad_cow_dc_l0t(uint8_t *src, int stride){
493 pred8x8_top_dc_c(src, stride);
494 pred4x4_dc_c(src, NULL, stride);
495 }
496
497 static void pred8x8_mad_cow_dc_0lt(uint8_t *src, int stride){
498 pred8x8_dc_c(src, stride);
499 pred4x4_top_dc_c(src, NULL, stride);
500 }
501
502 static void pred8x8_mad_cow_dc_l00(uint8_t *src, int stride){
503 pred8x8_left_dc_c(src, stride);
504 pred4x4_128_dc_c(src + 4*stride , NULL, stride);
505 pred4x4_128_dc_c(src + 4*stride + 4, NULL, stride);
506 }
507
508 static void pred8x8_mad_cow_dc_0l0(uint8_t *src, int stride){
509 pred8x8_left_dc_c(src, stride);
510 pred4x4_128_dc_c(src , NULL, stride);
511 pred4x4_128_dc_c(src + 4, NULL, stride);
512 }
513
514 static void pred8x8_plane_c(uint8_t *src, int stride){
515 int j, k;
516 int a;
517 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
518 const uint8_t * const src0 = src+3-stride;
519 const uint8_t *src1 = src+4*stride-1;
520 const uint8_t *src2 = src1-2*stride; // == src+2*stride-1;
521 int H = src0[1] - src0[-1];
522 int V = src1[0] - src2[ 0];
523 for(k=2; k<=4; ++k) {
524 src1 += stride; src2 -= stride;
525 H += k*(src0[k] - src0[-k]);
526 V += k*(src1[0] - src2[ 0]);
527 }
528 H = ( 17*H+16 ) >> 5;
529 V = ( 17*V+16 ) >> 5;
530
531 a = 16*(src1[0] + src2[8]+1) - 3*(V+H);
532 for(j=8; j>0; --j) {
533 int b = a;
534 a += V;
535 src[0] = cm[ (b ) >> 5 ];
536 src[1] = cm[ (b+ H) >> 5 ];
537 src[2] = cm[ (b+2*H) >> 5 ];
538 src[3] = cm[ (b+3*H) >> 5 ];
539 src[4] = cm[ (b+4*H) >> 5 ];
540 src[5] = cm[ (b+5*H) >> 5 ];
541 src[6] = cm[ (b+6*H) >> 5 ];
542 src[7] = cm[ (b+7*H) >> 5 ];
543 src += stride;
544 }
545 }
546
547 #define SRC(x,y) src[(x)+(y)*stride]
548 #define PL(y) \
549 const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;
550 #define PREDICT_8x8_LOAD_LEFT \
551 const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \
552 + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \
553 PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \
554 const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2
555
556 #define PT(x) \
557 const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
558 #define PREDICT_8x8_LOAD_TOP \
559 const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \
560 + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \
561 PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \
562 const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
563 + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2
564
565 #define PTR(x) \
566 t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
567 #define PREDICT_8x8_LOAD_TOPRIGHT \
568 int t8, t9, t10, t11, t12, t13, t14, t15; \
569 if(has_topright) { \
570 PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \
571 t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \
572 } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);
573
574 #define PREDICT_8x8_LOAD_TOPLEFT \
575 const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2
576
577 #define PREDICT_8x8_DC(v) \
578 int y; \
579 for( y = 0; y < 8; y++ ) { \
580 ((uint32_t*)src)[0] = \
581 ((uint32_t*)src)[1] = v; \
582 src += stride; \
583 }
584
585 static void pred8x8l_128_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride){
586 (void) has_topleft; (void) has_topright;
587 PREDICT_8x8_DC(0x80808080);
588 }
589
590 static void pred8x8l_left_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride){
591 (void) has_topleft; (void) has_topright;
592 PREDICT_8x8_LOAD_LEFT;
593 const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3) * 0x01010101;
594 PREDICT_8x8_DC(dc);
595 }
596
597 static void pred8x8l_top_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride){
598 PREDICT_8x8_LOAD_TOP;
599 const uint32_t dc = ((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3) * 0x01010101;
600 PREDICT_8x8_DC(dc);
601 }
602
603 static void pred8x8l_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride){
604 PREDICT_8x8_LOAD_LEFT;
605 PREDICT_8x8_LOAD_TOP;
606 const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7
607 +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4) * 0x01010101;
608 PREDICT_8x8_DC(dc);
609 }
610
611 static void pred8x8l_horizontal_c(uint8_t *src, int has_topleft, int has_topright, int stride){
612 (void) has_topleft; (void) has_topright;
613 PREDICT_8x8_LOAD_LEFT;
614 #define ROW(y) ((uint32_t*)(src+y*stride))[0] =\
615 ((uint32_t*)(src+y*stride))[1] = 0x01010101 * l##y
616 ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
617 #undef ROW
618 }
619
620 static void pred8x8l_vertical_c(uint8_t *src, int has_topleft, int has_topright, int stride){
621 int y;
622 PREDICT_8x8_LOAD_TOP;
623 src[0] = t0;
624 src[1] = t1;
625 src[2] = t2;
626 src[3] = t3;
627 src[4] = t4;
628 src[5] = t5;
629 src[6] = t6;
630 src[7] = t7;
631 for( y = 1; y < 8; y++ )
632 *(uint64_t*)(src+y*stride) = *(uint64_t*)src;
633 }
634
635 static void pred8x8l_down_left_c(uint8_t *src, int has_topleft, int has_topright, int stride){
636 PREDICT_8x8_LOAD_TOP;
637 PREDICT_8x8_LOAD_TOPRIGHT;
638 SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2;
639 SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2;
640 SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2;
641 SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2;
642 SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2;
643 SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2;
644 SRC(0,6)=SRC(1,5)=SRC(2,4)=SRC(3,3)=SRC(4,2)=SRC(5,1)=SRC(6,0)= (t6 + 2*t7 + t8 + 2) >> 2;
645 SRC(0,7)=SRC(1,6)=SRC(2,5)=SRC(3,4)=SRC(4,3)=SRC(5,2)=SRC(6,1)=SRC(7,0)= (t7 + 2*t8 + t9 + 2) >> 2;
646 SRC(1,7)=SRC(2,6)=SRC(3,5)=SRC(4,4)=SRC(5,3)=SRC(6,2)=SRC(7,1)= (t8 + 2*t9 + t10 + 2) >> 2;
647 SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2;
648 SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2;
649 SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2;
650 SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2;
651 SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;
652 SRC(7,7)= (t14 + 3*t15 + 2) >> 2;
653 }
654
655 static void pred8x8l_down_right_c(uint8_t *src, int has_topleft, int has_topright, int stride){
656 PREDICT_8x8_LOAD_TOP;
657 PREDICT_8x8_LOAD_LEFT;
658 PREDICT_8x8_LOAD_TOPLEFT;
659 SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2;
660 SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2;
661 SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2;
662 SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2;
663 SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2;
664 SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2;
665 SRC(0,1)=SRC(1,2)=SRC(2,3)=SRC(3,4)=SRC(4,5)=SRC(5,6)=SRC(6,7)= (l1 + 2*l0 + lt + 2) >> 2;
666 SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)=SRC(4,4)=SRC(5,5)=SRC(6,6)=SRC(7,7)= (l0 + 2*lt + t0 + 2) >> 2;
667 SRC(1,0)=SRC(2,1)=SRC(3,2)=SRC(4,3)=SRC(5,4)=SRC(6,5)=SRC(7,6)= (lt + 2*t0 + t1 + 2) >> 2;
668 SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2;
669 SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2;
670 SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2;
671 SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2;
672 SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;
673 SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;
674 }
675
676 static void pred8x8l_vertical_right_c(uint8_t *src, int has_topleft, int has_topright, int stride){
677 PREDICT_8x8_LOAD_TOP;
678 PREDICT_8x8_LOAD_LEFT;
679 PREDICT_8x8_LOAD_TOPLEFT;
680 SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2;
681 SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2;
682 SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2;
683 SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2;
684 SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2;
685 SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2;
686 SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2;
687 SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1;
688 SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2;
689 SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1;
690 SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2;
691 SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1;
692 SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2;
693 SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1;
694 SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2;
695 SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1;
696 SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2;
697 SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1;
698 SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2;
699 SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1;
700 SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;
701 SRC(7,0)= (t6 + t7 + 1) >> 1;
702 }
703
704 static void pred8x8l_horizontal_down_c(uint8_t *src, int has_topleft, int has_topright, int stride){
705 PREDICT_8x8_LOAD_TOP;
706 PREDICT_8x8_LOAD_LEFT;
707 PREDICT_8x8_LOAD_TOPLEFT;
708 SRC(0,7)= (l6 + l7 + 1) >> 1;
709 SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2;
710 SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1;
711 SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2;
712 SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1;
713 SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2;
714 SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1;
715 SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2;
716 SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1;
717 SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2;
718 SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1;
719 SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2;
720 SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1;
721 SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2;
722 SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1;
723 SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2;
724 SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2;
725 SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2;
726 SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2;
727 SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2;
728 SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;
729 SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;
730 }
731
732 static void pred8x8l_vertical_left_c(uint8_t *src, int has_topleft, int has_topright, int stride){
733 PREDICT_8x8_LOAD_TOP;
734 PREDICT_8x8_LOAD_TOPRIGHT;
735 SRC(0,0)= (t0 + t1 + 1) >> 1;
736 SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2;
737 SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1;
738 SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2;
739 SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1;
740 SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2;
741 SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1;
742 SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2;
743 SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1;
744 SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2;
745 SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1;
746 SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2;
747 SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1;
748 SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2;
749 SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1;
750 SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2;
751 SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1;
752 SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2;
753 SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1;
754 SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2;
755 SRC(7,6)= (t10 + t11 + 1) >> 1;
756 SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;
757 }
758
759 static void pred8x8l_horizontal_up_c(uint8_t *src, int has_topleft, int has_topright, int stride){
760 (void) has_topleft; (void) has_topright;
761 PREDICT_8x8_LOAD_LEFT;
762 SRC(0,0)= (l0 + l1 + 1) >> 1;
763 SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2;
764 SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1;
765 SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2;
766 SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1;
767 SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2;
768 SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1;
769 SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2;
770 SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1;
771 SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2;
772 SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1;
773 SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2;
774 SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1;
775 SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2;
776 SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)=
777 SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)=
778 SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=
779 SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;
780 }
781 #undef PREDICT_8x8_LOAD_LEFT
782 #undef PREDICT_8x8_LOAD_TOP
783 #undef PREDICT_8x8_LOAD_TOPLEFT
784 #undef PREDICT_8x8_LOAD_TOPRIGHT
785 #undef PREDICT_8x8_DC
786 #undef PTR
787 #undef PT
788 #undef PL
789 #undef SRC
790
791 static void pred4x4_vertical_add_c(uint8_t *pix, const DCTELEM *block, int stride){
792 int i;
793 pix -= stride;
794 for(i=0; i<4; i++){
795 uint8_t v = pix[0];
796 pix[1*stride]= v += block[0];
797 pix[2*stride]= v += block[4];
798 pix[3*stride]= v += block[8];
799 pix[4*stride]= v + block[12];
800 pix++;
801 block++;
802 }
803 }
804
805 static void pred4x4_horizontal_add_c(uint8_t *pix, const DCTELEM *block, int stride){
806 int i;
807 for(i=0; i<4; i++){
808 uint8_t v = pix[-1];
809 pix[0]= v += block[0];
810 pix[1]= v += block[1];
811 pix[2]= v += block[2];
812 pix[3]= v + block[3];
813 pix+= stride;
814 block+= 4;
815 }
816 }
817
818 static void pred8x8l_vertical_add_c(uint8_t *pix, const DCTELEM *block, int stride){
819 int i;
820 pix -= stride;
821 for(i=0; i<8; i++){
822 uint8_t v = pix[0];
823 pix[1*stride]= v += block[0];
824 pix[2*stride]= v += block[8];
825 pix[3*stride]= v += block[16];
826 pix[4*stride]= v += block[24];
827 pix[5*stride]= v += block[32];
828 pix[6*stride]= v += block[40];
829 pix[7*stride]= v += block[48];
830 pix[8*stride]= v + block[56];
831 pix++;
832 block++;
833 }
834 }
835
836 static void pred8x8l_horizontal_add_c(uint8_t *pix, const DCTELEM *block, int stride){
837 int i;
838 for(i=0; i<8; i++){
839 uint8_t v = pix[-1];
840 pix[0]= v += block[0];
841 pix[1]= v += block[1];
842 pix[2]= v += block[2];
843 pix[3]= v += block[3];
844 pix[4]= v += block[4];
845 pix[5]= v += block[5];
846 pix[6]= v += block[6];
847 pix[7]= v + block[7];
848 pix+= stride;
849 block+= 8;
850 }
851 }
852
853 static void pred16x16_vertical_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
854 int i;
855 for(i=0; i<16; i++)
856 pred4x4_vertical_add_c(pix + block_offset[i], block + i*16, stride);
857 }
858
859 static void pred16x16_horizontal_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
860 int i;
861 for(i=0; i<16; i++)
862 pred4x4_horizontal_add_c(pix + block_offset[i], block + i*16, stride);
863 }
864
865 static void pred8x8_vertical_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
866 int i;
867 for(i=0; i<4; i++)
868 pred4x4_vertical_add_c(pix + block_offset[i], block + i*16, stride);
869 }
870
871 static void pred8x8_horizontal_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
872 int i;
873 for(i=0; i<4; i++)
874 pred4x4_horizontal_add_c(pix + block_offset[i], block + i*16, stride);
875 }
876
877
878 /**
879 * Sets the intra prediction function pointers.
880 */
881 void ff_h264_pred_init(H264PredContext *h){
882
883 h->pred4x4[VERT_PRED ]= pred4x4_vertical_c;
884 h->pred4x4[HOR_PRED ]= pred4x4_horizontal_c;
885 h->pred4x4[DC_PRED ]= pred4x4_dc_c;
886 h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_c;
887 h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c;
888 h->pred4x4[VERT_RIGHT_PRED ]= pred4x4_vertical_right_c;
889 h->pred4x4[HOR_DOWN_PRED ]= pred4x4_horizontal_down_c;
890 h->pred4x4[VERT_LEFT_PRED ]= pred4x4_vertical_left_c;
891 h->pred4x4[HOR_UP_PRED ]= pred4x4_horizontal_up_c;
892 h->pred4x4[LEFT_DC_PRED ]= pred4x4_left_dc_c;
893 h->pred4x4[TOP_DC_PRED ]= pred4x4_top_dc_c;
894 h->pred4x4[DC_128_PRED ]= pred4x4_128_dc_c;
895
896 h->pred8x8l[VERT_PRED ]= pred8x8l_vertical_c;
897 h->pred8x8l[HOR_PRED ]= pred8x8l_horizontal_c;
898 h->pred8x8l[DC_PRED ]= pred8x8l_dc_c;
899 h->pred8x8l[DIAG_DOWN_LEFT_PRED ]= pred8x8l_down_left_c;
900 h->pred8x8l[DIAG_DOWN_RIGHT_PRED]= pred8x8l_down_right_c;
901 h->pred8x8l[VERT_RIGHT_PRED ]= pred8x8l_vertical_right_c;
902 h->pred8x8l[HOR_DOWN_PRED ]= pred8x8l_horizontal_down_c;
903 h->pred8x8l[VERT_LEFT_PRED ]= pred8x8l_vertical_left_c;
904 h->pred8x8l[HOR_UP_PRED ]= pred8x8l_horizontal_up_c;
905 h->pred8x8l[LEFT_DC_PRED ]= pred8x8l_left_dc_c;
906 h->pred8x8l[TOP_DC_PRED ]= pred8x8l_top_dc_c;
907 h->pred8x8l[DC_128_PRED ]= pred8x8l_128_dc_c;
908
909 h->pred8x8[VERT_PRED8x8 ]= pred8x8_vertical_c;
910 h->pred8x8[HOR_PRED8x8 ]= pred8x8_horizontal_c;
911 h->pred8x8[PLANE_PRED8x8 ]= pred8x8_plane_c;
912
913 h->pred8x8[DC_PRED8x8 ]= pred8x8_dc_c;
914 h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_c;
915 h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_c;
916 h->pred8x8[ALZHEIMER_DC_L0T_PRED8x8 ]= pred8x8_mad_cow_dc_l0t;
917 h->pred8x8[ALZHEIMER_DC_0LT_PRED8x8 ]= pred8x8_mad_cow_dc_0lt;
918 h->pred8x8[ALZHEIMER_DC_L00_PRED8x8 ]= pred8x8_mad_cow_dc_l00;
919 h->pred8x8[ALZHEIMER_DC_0L0_PRED8x8 ]= pred8x8_mad_cow_dc_0l0;
920
921 h->pred8x8[DC_128_PRED8x8 ]= pred8x8_128_dc_c;
922
923 h->pred16x16[DC_PRED8x8 ]= pred16x16_dc_c;
924 h->pred16x16[VERT_PRED8x8 ]= pred16x16_vertical_c;
925 h->pred16x16[HOR_PRED8x8 ]= pred16x16_horizontal_c;
926 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_c;
927
928 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_c;
929
930 h->pred16x16[LEFT_DC_PRED8x8]= pred16x16_left_dc_c;
931 h->pred16x16[TOP_DC_PRED8x8 ]= pred16x16_top_dc_c;
932 h->pred16x16[DC_128_PRED8x8 ]= pred16x16_128_dc_c;
933
934 //special lossless h/v prediction for h264
935 h->pred4x4_add [VERT_PRED ]= pred4x4_vertical_add_c;
936 h->pred4x4_add [ HOR_PRED ]= pred4x4_horizontal_add_c;
937 h->pred8x8l_add [VERT_PRED ]= pred8x8l_vertical_add_c;
938 h->pred8x8l_add [ HOR_PRED ]= pred8x8l_horizontal_add_c;
939 h->pred8x8_add [VERT_PRED8x8]= pred8x8_vertical_add_c;
940 h->pred8x8_add [ HOR_PRED8x8]= pred8x8_horizontal_add_c;
941 h->pred16x16_add[VERT_PRED8x8]= pred16x16_vertical_add_c;
942 h->pred16x16_add[ HOR_PRED8x8]= pred16x16_horizontal_add_c;
943
944 if (HAVE_NEON) ff_h264_pred_init_arm(h);
945 }