1 /*****************************************************************************
2 * predict.c: intra prediction
3 *****************************************************************************
4 * Copyright (C) 2003-2021 x264 project
5 *
6 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
7 * Loren Merritt <lorenm@u.washington.edu>
8 * Fiona Glaser <fiona@x264.com>
9 * Henrik Gramner <henrik@gramner.com>
10 *
11 * This program is free software; you can redistribute it and/or modify
12 * it under the terms of the GNU General Public License as published by
13 * the Free Software Foundation; either version 2 of the License, or
14 * (at your option) any later version.
15 *
16 * This program is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 * GNU General Public License for more details.
20 *
21 * You should have received a copy of the GNU General Public License
22 * along with this program; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
24 *
25 * This program is also available under a commercial proprietary license.
26 * For more information, contact us at licensing@x264.com.
27 *****************************************************************************/
28
29 /* predict4x4 are inspired from ffmpeg h264 decoder */
30
31
32 #include "common.h"
33
34 #if HAVE_MMX
35 # include "x86/predict.h"
36 #endif
37 #if HAVE_ALTIVEC
38 # include "ppc/predict.h"
39 #endif
40 #if HAVE_ARMV6
41 # include "arm/predict.h"
42 #endif
43 #if HAVE_AARCH64
44 # include "aarch64/predict.h"
45 #endif
46 #if HAVE_MSA
47 # include "mips/predict.h"
48 #endif
49
50 /****************************************************************************
51 * 16x16 prediction for intra luma block
52 ****************************************************************************/
53
54 #define PREDICT_16x16_DC(v)\
55 for( int i = 0; i < 16; i++ )\
56 {\
57 MPIXEL_X4( src+ 0 ) = v;\
58 MPIXEL_X4( src+ 4 ) = v;\
59 MPIXEL_X4( src+ 8 ) = v;\
60 MPIXEL_X4( src+12 ) = v;\
61 src += FDEC_STRIDE;\
62 }
63
x264_predict_16x16_dc_c(pixel * src)64 void x264_predict_16x16_dc_c( pixel *src )
65 {
66 int dc = 0;
67
68 for( int i = 0; i < 16; i++ )
69 {
70 dc += src[-1 + i * FDEC_STRIDE];
71 dc += src[i - FDEC_STRIDE];
72 }
73 pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 16 ) >> 5 );
74
75 PREDICT_16x16_DC( dcsplat );
76 }
predict_16x16_dc_left_c(pixel * src)77 static void predict_16x16_dc_left_c( pixel *src )
78 {
79 int dc = 0;
80
81 for( int i = 0; i < 16; i++ )
82 dc += src[-1 + i * FDEC_STRIDE];
83 pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 );
84
85 PREDICT_16x16_DC( dcsplat );
86 }
predict_16x16_dc_top_c(pixel * src)87 static void predict_16x16_dc_top_c( pixel *src )
88 {
89 int dc = 0;
90
91 for( int i = 0; i < 16; i++ )
92 dc += src[i - FDEC_STRIDE];
93 pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 );
94
95 PREDICT_16x16_DC( dcsplat );
96 }
predict_16x16_dc_128_c(pixel * src)97 static void predict_16x16_dc_128_c( pixel *src )
98 {
99 PREDICT_16x16_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );
100 }
x264_predict_16x16_h_c(pixel * src)101 void x264_predict_16x16_h_c( pixel *src )
102 {
103 for( int i = 0; i < 16; i++ )
104 {
105 const pixel4 v = PIXEL_SPLAT_X4( src[-1] );
106 MPIXEL_X4( src+ 0 ) = v;
107 MPIXEL_X4( src+ 4 ) = v;
108 MPIXEL_X4( src+ 8 ) = v;
109 MPIXEL_X4( src+12 ) = v;
110 src += FDEC_STRIDE;
111 }
112 }
x264_predict_16x16_v_c(pixel * src)113 void x264_predict_16x16_v_c( pixel *src )
114 {
115 pixel4 v0 = MPIXEL_X4( &src[ 0-FDEC_STRIDE] );
116 pixel4 v1 = MPIXEL_X4( &src[ 4-FDEC_STRIDE] );
117 pixel4 v2 = MPIXEL_X4( &src[ 8-FDEC_STRIDE] );
118 pixel4 v3 = MPIXEL_X4( &src[12-FDEC_STRIDE] );
119
120 for( int i = 0; i < 16; i++ )
121 {
122 MPIXEL_X4( src+ 0 ) = v0;
123 MPIXEL_X4( src+ 4 ) = v1;
124 MPIXEL_X4( src+ 8 ) = v2;
125 MPIXEL_X4( src+12 ) = v3;
126 src += FDEC_STRIDE;
127 }
128 }
x264_predict_16x16_p_c(pixel * src)129 void x264_predict_16x16_p_c( pixel *src )
130 {
131 int H = 0, V = 0;
132
133 /* calculate H and V */
134 for( int i = 0; i <= 7; i++ )
135 {
136 H += ( i + 1 ) * ( src[ 8 + i - FDEC_STRIDE ] - src[6 -i -FDEC_STRIDE] );
137 V += ( i + 1 ) * ( src[-1 + (8+i)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );
138 }
139
140 int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[15 - FDEC_STRIDE] );
141 int b = ( 5 * H + 32 ) >> 6;
142 int c = ( 5 * V + 32 ) >> 6;
143
144 int i00 = a - b * 7 - c * 7 + 16;
145
146 for( int y = 0; y < 16; y++ )
147 {
148 int pix = i00;
149 for( int x = 0; x < 16; x++ )
150 {
151 src[x] = x264_clip_pixel( pix>>5 );
152 pix += b;
153 }
154 src += FDEC_STRIDE;
155 i00 += c;
156 }
157 }
158
159
160 /****************************************************************************
161 * 8x8 prediction for intra chroma block (4:2:0)
162 ****************************************************************************/
163
predict_8x8c_dc_128_c(pixel * src)164 static void predict_8x8c_dc_128_c( pixel *src )
165 {
166 for( int y = 0; y < 8; y++ )
167 {
168 MPIXEL_X4( src+0 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
169 MPIXEL_X4( src+4 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
170 src += FDEC_STRIDE;
171 }
172 }
predict_8x8c_dc_left_c(pixel * src)173 static void predict_8x8c_dc_left_c( pixel *src )
174 {
175 int dc0 = 0, dc1 = 0;
176
177 for( int y = 0; y < 4; y++ )
178 {
179 dc0 += src[y * FDEC_STRIDE - 1];
180 dc1 += src[(y+4) * FDEC_STRIDE - 1];
181 }
182 pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );
183 pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );
184
185 for( int y = 0; y < 4; y++ )
186 {
187 MPIXEL_X4( src+0 ) = dc0splat;
188 MPIXEL_X4( src+4 ) = dc0splat;
189 src += FDEC_STRIDE;
190 }
191 for( int y = 0; y < 4; y++ )
192 {
193 MPIXEL_X4( src+0 ) = dc1splat;
194 MPIXEL_X4( src+4 ) = dc1splat;
195 src += FDEC_STRIDE;
196 }
197
198 }
predict_8x8c_dc_top_c(pixel * src)199 static void predict_8x8c_dc_top_c( pixel *src )
200 {
201 int dc0 = 0, dc1 = 0;
202
203 for( int x = 0; x < 4; x++ )
204 {
205 dc0 += src[x - FDEC_STRIDE];
206 dc1 += src[x + 4 - FDEC_STRIDE];
207 }
208 pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );
209 pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );
210
211 for( int y = 0; y < 8; y++ )
212 {
213 MPIXEL_X4( src+0 ) = dc0splat;
214 MPIXEL_X4( src+4 ) = dc1splat;
215 src += FDEC_STRIDE;
216 }
217 }
x264_predict_8x8c_dc_c(pixel * src)218 void x264_predict_8x8c_dc_c( pixel *src )
219 {
220 int s0 = 0, s1 = 0, s2 = 0, s3 = 0;
221
222 /*
223 s0 s1
224 s2
225 s3
226 */
227 for( int i = 0; i < 4; i++ )
228 {
229 s0 += src[i - FDEC_STRIDE];
230 s1 += src[i + 4 - FDEC_STRIDE];
231 s2 += src[-1 + i * FDEC_STRIDE];
232 s3 += src[-1 + (i+4)*FDEC_STRIDE];
233 }
234 /*
235 dc0 dc1
236 dc2 dc3
237 */
238 pixel4 dc0 = PIXEL_SPLAT_X4( ( s0 + s2 + 4 ) >> 3 );
239 pixel4 dc1 = PIXEL_SPLAT_X4( ( s1 + 2 ) >> 2 );
240 pixel4 dc2 = PIXEL_SPLAT_X4( ( s3 + 2 ) >> 2 );
241 pixel4 dc3 = PIXEL_SPLAT_X4( ( s1 + s3 + 4 ) >> 3 );
242
243 for( int y = 0; y < 4; y++ )
244 {
245 MPIXEL_X4( src+0 ) = dc0;
246 MPIXEL_X4( src+4 ) = dc1;
247 src += FDEC_STRIDE;
248 }
249
250 for( int y = 0; y < 4; y++ )
251 {
252 MPIXEL_X4( src+0 ) = dc2;
253 MPIXEL_X4( src+4 ) = dc3;
254 src += FDEC_STRIDE;
255 }
256 }
x264_predict_8x8c_h_c(pixel * src)257 void x264_predict_8x8c_h_c( pixel *src )
258 {
259 for( int i = 0; i < 8; i++ )
260 {
261 pixel4 v = PIXEL_SPLAT_X4( src[-1] );
262 MPIXEL_X4( src+0 ) = v;
263 MPIXEL_X4( src+4 ) = v;
264 src += FDEC_STRIDE;
265 }
266 }
x264_predict_8x8c_v_c(pixel * src)267 void x264_predict_8x8c_v_c( pixel *src )
268 {
269 pixel4 v0 = MPIXEL_X4( src+0-FDEC_STRIDE );
270 pixel4 v1 = MPIXEL_X4( src+4-FDEC_STRIDE );
271
272 for( int i = 0; i < 8; i++ )
273 {
274 MPIXEL_X4( src+0 ) = v0;
275 MPIXEL_X4( src+4 ) = v1;
276 src += FDEC_STRIDE;
277 }
278 }
x264_predict_8x8c_p_c(pixel * src)279 void x264_predict_8x8c_p_c( pixel *src )
280 {
281 int H = 0, V = 0;
282
283 for( int i = 0; i < 4; i++ )
284 {
285 H += ( i + 1 ) * ( src[4+i - FDEC_STRIDE] - src[2 - i -FDEC_STRIDE] );
286 V += ( i + 1 ) * ( src[-1 +(i+4)*FDEC_STRIDE] - src[-1+(2-i)*FDEC_STRIDE] );
287 }
288
289 int a = 16 * ( src[-1+7*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );
290 int b = ( 17 * H + 16 ) >> 5;
291 int c = ( 17 * V + 16 ) >> 5;
292 int i00 = a -3*b -3*c + 16;
293
294 for( int y = 0; y < 8; y++ )
295 {
296 int pix = i00;
297 for( int x = 0; x < 8; x++ )
298 {
299 src[x] = x264_clip_pixel( pix>>5 );
300 pix += b;
301 }
302 src += FDEC_STRIDE;
303 i00 += c;
304 }
305 }
306
307 /****************************************************************************
308 * 8x16 prediction for intra chroma block (4:2:2)
309 ****************************************************************************/
310
predict_8x16c_dc_128_c(pixel * src)311 static void predict_8x16c_dc_128_c( pixel *src )
312 {
313 for( int y = 0; y < 16; y++ )
314 {
315 MPIXEL_X4( src+0 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
316 MPIXEL_X4( src+4 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
317 src += FDEC_STRIDE;
318 }
319 }
predict_8x16c_dc_left_c(pixel * src)320 static void predict_8x16c_dc_left_c( pixel *src )
321 {
322 for( int i = 0; i < 4; i++ )
323 {
324 int dc = 0;
325
326 for( int y = 0; y < 4; y++ )
327 dc += src[y*FDEC_STRIDE - 1];
328
329 pixel4 dcsplat = PIXEL_SPLAT_X4( (dc + 2) >> 2 );
330
331 for( int y = 0; y < 4; y++ )
332 {
333 MPIXEL_X4( src+0 ) = dcsplat;
334 MPIXEL_X4( src+4 ) = dcsplat;
335 src += FDEC_STRIDE;
336 }
337 }
338 }
predict_8x16c_dc_top_c(pixel * src)339 static void predict_8x16c_dc_top_c( pixel *src )
340 {
341 int dc0 = 0, dc1 = 0;
342
343 for( int x = 0; x < 4; x++ )
344 {
345 dc0 += src[x - FDEC_STRIDE];
346 dc1 += src[x + 4 - FDEC_STRIDE];
347 }
348 pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );
349 pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );
350
351 for( int y = 0; y < 16; y++ )
352 {
353 MPIXEL_X4( src+0 ) = dc0splat;
354 MPIXEL_X4( src+4 ) = dc1splat;
355 src += FDEC_STRIDE;
356 }
357 }
x264_predict_8x16c_dc_c(pixel * src)358 void x264_predict_8x16c_dc_c( pixel *src )
359 {
360 int s0 = 0, s1 = 0, s2 = 0, s3 = 0, s4 = 0, s5 = 0;
361
362 /*
363 s0 s1
364 s2
365 s3
366 s4
367 s5
368 */
369 for( int i = 0; i < 4; i++ )
370 {
371 s0 += src[i+0 - FDEC_STRIDE];
372 s1 += src[i+4 - FDEC_STRIDE];
373 s2 += src[-1 + (i+0) * FDEC_STRIDE];
374 s3 += src[-1 + (i+4) * FDEC_STRIDE];
375 s4 += src[-1 + (i+8) * FDEC_STRIDE];
376 s5 += src[-1 + (i+12) * FDEC_STRIDE];
377 }
378 /*
379 dc0 dc1
380 dc2 dc3
381 dc4 dc5
382 dc6 dc7
383 */
384 pixel4 dc0 = PIXEL_SPLAT_X4( ( s0 + s2 + 4 ) >> 3 );
385 pixel4 dc1 = PIXEL_SPLAT_X4( ( s1 + 2 ) >> 2 );
386 pixel4 dc2 = PIXEL_SPLAT_X4( ( s3 + 2 ) >> 2 );
387 pixel4 dc3 = PIXEL_SPLAT_X4( ( s1 + s3 + 4 ) >> 3 );
388 pixel4 dc4 = PIXEL_SPLAT_X4( ( s4 + 2 ) >> 2 );
389 pixel4 dc5 = PIXEL_SPLAT_X4( ( s1 + s4 + 4 ) >> 3 );
390 pixel4 dc6 = PIXEL_SPLAT_X4( ( s5 + 2 ) >> 2 );
391 pixel4 dc7 = PIXEL_SPLAT_X4( ( s1 + s5 + 4 ) >> 3 );
392
393 for( int y = 0; y < 4; y++ )
394 {
395 MPIXEL_X4( src+0 ) = dc0;
396 MPIXEL_X4( src+4 ) = dc1;
397 src += FDEC_STRIDE;
398 }
399 for( int y = 0; y < 4; y++ )
400 {
401 MPIXEL_X4( src+0 ) = dc2;
402 MPIXEL_X4( src+4 ) = dc3;
403 src += FDEC_STRIDE;
404 }
405 for( int y = 0; y < 4; y++ )
406 {
407 MPIXEL_X4( src+0 ) = dc4;
408 MPIXEL_X4( src+4 ) = dc5;
409 src += FDEC_STRIDE;
410 }
411 for( int y = 0; y < 4; y++ )
412 {
413 MPIXEL_X4( src+0 ) = dc6;
414 MPIXEL_X4( src+4 ) = dc7;
415 src += FDEC_STRIDE;
416 }
417 }
x264_predict_8x16c_h_c(pixel * src)418 void x264_predict_8x16c_h_c( pixel *src )
419 {
420 for( int i = 0; i < 16; i++ )
421 {
422 pixel4 v = PIXEL_SPLAT_X4( src[-1] );
423 MPIXEL_X4( src+0 ) = v;
424 MPIXEL_X4( src+4 ) = v;
425 src += FDEC_STRIDE;
426 }
427 }
x264_predict_8x16c_v_c(pixel * src)428 void x264_predict_8x16c_v_c( pixel *src )
429 {
430 pixel4 v0 = MPIXEL_X4( src+0-FDEC_STRIDE );
431 pixel4 v1 = MPIXEL_X4( src+4-FDEC_STRIDE );
432
433 for( int i = 0; i < 16; i++ )
434 {
435 MPIXEL_X4( src+0 ) = v0;
436 MPIXEL_X4( src+4 ) = v1;
437 src += FDEC_STRIDE;
438 }
439 }
x264_predict_8x16c_p_c(pixel * src)440 void x264_predict_8x16c_p_c( pixel *src )
441 {
442 int H = 0;
443 int V = 0;
444
445 for( int i = 0; i < 4; i++ )
446 H += ( i + 1 ) * ( src[4 + i - FDEC_STRIDE] - src[2 - i - FDEC_STRIDE] );
447 for( int i = 0; i < 8; i++ )
448 V += ( i + 1 ) * ( src[-1 + (i+8)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );
449
450 int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );
451 int b = ( 17 * H + 16 ) >> 5;
452 int c = ( 5 * V + 32 ) >> 6;
453 int i00 = a -3*b -7*c + 16;
454
455 for( int y = 0; y < 16; y++ )
456 {
457 int pix = i00;
458 for( int x = 0; x < 8; x++ )
459 {
460 src[x] = x264_clip_pixel( pix>>5 );
461 pix += b;
462 }
463 src += FDEC_STRIDE;
464 i00 += c;
465 }
466 }
467
468 /****************************************************************************
469 * 4x4 prediction for intra luma block
470 ****************************************************************************/
471
472 #define SRC(x,y) src[(x)+(y)*FDEC_STRIDE]
473 #define SRC_X4(x,y) MPIXEL_X4( &SRC(x,y) )
474
475 #define PREDICT_4x4_DC(v)\
476 SRC_X4(0,0) = SRC_X4(0,1) = SRC_X4(0,2) = SRC_X4(0,3) = v;
477
predict_4x4_dc_128_c(pixel * src)478 static void predict_4x4_dc_128_c( pixel *src )
479 {
480 PREDICT_4x4_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );
481 }
predict_4x4_dc_left_c(pixel * src)482 static void predict_4x4_dc_left_c( pixel *src )
483 {
484 pixel4 dc = PIXEL_SPLAT_X4( (SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) + 2) >> 2 );
485 PREDICT_4x4_DC( dc );
486 }
predict_4x4_dc_top_c(pixel * src)487 static void predict_4x4_dc_top_c( pixel *src )
488 {
489 pixel4 dc = PIXEL_SPLAT_X4( (SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 2) >> 2 );
490 PREDICT_4x4_DC( dc );
491 }
x264_predict_4x4_dc_c(pixel * src)492 void x264_predict_4x4_dc_c( pixel *src )
493 {
494 pixel4 dc = PIXEL_SPLAT_X4( (SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) +
495 SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 4) >> 3 );
496 PREDICT_4x4_DC( dc );
497 }
x264_predict_4x4_h_c(pixel * src)498 void x264_predict_4x4_h_c( pixel *src )
499 {
500 SRC_X4(0,0) = PIXEL_SPLAT_X4( SRC(-1,0) );
501 SRC_X4(0,1) = PIXEL_SPLAT_X4( SRC(-1,1) );
502 SRC_X4(0,2) = PIXEL_SPLAT_X4( SRC(-1,2) );
503 SRC_X4(0,3) = PIXEL_SPLAT_X4( SRC(-1,3) );
504 }
x264_predict_4x4_v_c(pixel * src)505 void x264_predict_4x4_v_c( pixel *src )
506 {
507 PREDICT_4x4_DC(SRC_X4(0,-1));
508 }
509
510 #define PREDICT_4x4_LOAD_LEFT\
511 int l0 = SRC(-1,0);\
512 int l1 = SRC(-1,1);\
513 int l2 = SRC(-1,2);\
514 UNUSED int l3 = SRC(-1,3);
515
516 #define PREDICT_4x4_LOAD_TOP\
517 int t0 = SRC(0,-1);\
518 int t1 = SRC(1,-1);\
519 int t2 = SRC(2,-1);\
520 UNUSED int t3 = SRC(3,-1);
521
522 #define PREDICT_4x4_LOAD_TOP_RIGHT\
523 int t4 = SRC(4,-1);\
524 int t5 = SRC(5,-1);\
525 int t6 = SRC(6,-1);\
526 UNUSED int t7 = SRC(7,-1);
527
528 #define F1(a,b) (((a)+(b)+1)>>1)
529 #define F2(a,b,c) (((a)+2*(b)+(c)+2)>>2)
530
predict_4x4_ddl_c(pixel * src)531 static void predict_4x4_ddl_c( pixel *src )
532 {
533 PREDICT_4x4_LOAD_TOP
534 PREDICT_4x4_LOAD_TOP_RIGHT
535 SRC(0,0)= F2(t0,t1,t2);
536 SRC(1,0)=SRC(0,1)= F2(t1,t2,t3);
537 SRC(2,0)=SRC(1,1)=SRC(0,2)= F2(t2,t3,t4);
538 SRC(3,0)=SRC(2,1)=SRC(1,2)=SRC(0,3)= F2(t3,t4,t5);
539 SRC(3,1)=SRC(2,2)=SRC(1,3)= F2(t4,t5,t6);
540 SRC(3,2)=SRC(2,3)= F2(t5,t6,t7);
541 SRC(3,3)= F2(t6,t7,t7);
542 }
predict_4x4_ddr_c(pixel * src)543 static void predict_4x4_ddr_c( pixel *src )
544 {
545 int lt = SRC(-1,-1);
546 PREDICT_4x4_LOAD_LEFT
547 PREDICT_4x4_LOAD_TOP
548 SRC(3,0)= F2(t3,t2,t1);
549 SRC(2,0)=SRC(3,1)= F2(t2,t1,t0);
550 SRC(1,0)=SRC(2,1)=SRC(3,2)= F2(t1,t0,lt);
551 SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)= F2(t0,lt,l0);
552 SRC(0,1)=SRC(1,2)=SRC(2,3)= F2(lt,l0,l1);
553 SRC(0,2)=SRC(1,3)= F2(l0,l1,l2);
554 SRC(0,3)= F2(l1,l2,l3);
555 }
556
predict_4x4_vr_c(pixel * src)557 static void predict_4x4_vr_c( pixel *src )
558 {
559 int lt = SRC(-1,-1);
560 PREDICT_4x4_LOAD_LEFT
561 PREDICT_4x4_LOAD_TOP
562 SRC(0,3)= F2(l2,l1,l0);
563 SRC(0,2)= F2(l1,l0,lt);
564 SRC(0,1)=SRC(1,3)= F2(l0,lt,t0);
565 SRC(0,0)=SRC(1,2)= F1(lt,t0);
566 SRC(1,1)=SRC(2,3)= F2(lt,t0,t1);
567 SRC(1,0)=SRC(2,2)= F1(t0,t1);
568 SRC(2,1)=SRC(3,3)= F2(t0,t1,t2);
569 SRC(2,0)=SRC(3,2)= F1(t1,t2);
570 SRC(3,1)= F2(t1,t2,t3);
571 SRC(3,0)= F1(t2,t3);
572 }
573
predict_4x4_hd_c(pixel * src)574 static void predict_4x4_hd_c( pixel *src )
575 {
576 int lt= SRC(-1,-1);
577 PREDICT_4x4_LOAD_LEFT
578 PREDICT_4x4_LOAD_TOP
579 SRC(0,3)= F1(l2,l3);
580 SRC(1,3)= F2(l1,l2,l3);
581 SRC(0,2)=SRC(2,3)= F1(l1,l2);
582 SRC(1,2)=SRC(3,3)= F2(l0,l1,l2);
583 SRC(0,1)=SRC(2,2)= F1(l0,l1);
584 SRC(1,1)=SRC(3,2)= F2(lt,l0,l1);
585 SRC(0,0)=SRC(2,1)= F1(lt,l0);
586 SRC(1,0)=SRC(3,1)= F2(t0,lt,l0);
587 SRC(2,0)= F2(t1,t0,lt);
588 SRC(3,0)= F2(t2,t1,t0);
589 }
590
predict_4x4_vl_c(pixel * src)591 static void predict_4x4_vl_c( pixel *src )
592 {
593 PREDICT_4x4_LOAD_TOP
594 PREDICT_4x4_LOAD_TOP_RIGHT
595 SRC(0,0)= F1(t0,t1);
596 SRC(0,1)= F2(t0,t1,t2);
597 SRC(1,0)=SRC(0,2)= F1(t1,t2);
598 SRC(1,1)=SRC(0,3)= F2(t1,t2,t3);
599 SRC(2,0)=SRC(1,2)= F1(t2,t3);
600 SRC(2,1)=SRC(1,3)= F2(t2,t3,t4);
601 SRC(3,0)=SRC(2,2)= F1(t3,t4);
602 SRC(3,1)=SRC(2,3)= F2(t3,t4,t5);
603 SRC(3,2)= F1(t4,t5);
604 SRC(3,3)= F2(t4,t5,t6);
605 }
606
predict_4x4_hu_c(pixel * src)607 static void predict_4x4_hu_c( pixel *src )
608 {
609 PREDICT_4x4_LOAD_LEFT
610 SRC(0,0)= F1(l0,l1);
611 SRC(1,0)= F2(l0,l1,l2);
612 SRC(2,0)=SRC(0,1)= F1(l1,l2);
613 SRC(3,0)=SRC(1,1)= F2(l1,l2,l3);
614 SRC(2,1)=SRC(0,2)= F1(l2,l3);
615 SRC(3,1)=SRC(1,2)= F2(l2,l3,l3);
616 SRC(3,2)=SRC(1,3)=SRC(0,3)=
617 SRC(2,2)=SRC(2,3)=SRC(3,3)= l3;
618 }
619
620 /****************************************************************************
621 * 8x8 prediction for intra luma block
622 ****************************************************************************/
623
624 #define PL(y) \
625 edge[14-y] = F2(SRC(-1,y-1), SRC(-1,y), SRC(-1,y+1));
626 #define PT(x) \
627 edge[16+x] = F2(SRC(x-1,-1), SRC(x,-1), SRC(x+1,-1));
628
predict_8x8_filter_c(pixel * src,pixel edge[36],int i_neighbor,int i_filters)629 static void predict_8x8_filter_c( pixel *src, pixel edge[36], int i_neighbor, int i_filters )
630 {
631 /* edge[7..14] = l7..l0
632 * edge[15] = lt
633 * edge[16..31] = t0 .. t15
634 * edge[32] = t15 */
635
636 int have_lt = i_neighbor & MB_TOPLEFT;
637 if( i_filters & MB_LEFT )
638 {
639 edge[15] = (SRC(0,-1) + 2*SRC(-1,-1) + SRC(-1,0) + 2) >> 2;
640 edge[14] = ((have_lt ? SRC(-1,-1) : SRC(-1,0))
641 + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2;
642 PL(1) PL(2) PL(3) PL(4) PL(5) PL(6)
643 edge[6] =
644 edge[7] = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2;
645 }
646
647 if( i_filters & MB_TOP )
648 {
649 int have_tr = i_neighbor & MB_TOPRIGHT;
650 edge[16] = ((have_lt ? SRC(-1,-1) : SRC(0,-1))
651 + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2;
652 PT(1) PT(2) PT(3) PT(4) PT(5) PT(6)
653 edge[23] = (SRC(6,-1) + 2*SRC(7,-1)
654 + (have_tr ? SRC(8,-1) : SRC(7,-1)) + 2) >> 2;
655
656 if( i_filters & MB_TOPRIGHT )
657 {
658 if( have_tr )
659 {
660 PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14)
661 edge[31] =
662 edge[32] = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2;
663 }
664 else
665 {
666 MPIXEL_X4( edge+24 ) = PIXEL_SPLAT_X4( SRC(7,-1) );
667 MPIXEL_X4( edge+28 ) = PIXEL_SPLAT_X4( SRC(7,-1) );
668 edge[32] = SRC(7,-1);
669 }
670 }
671 }
672 }
673
674 #undef PL
675 #undef PT
676
677 #define PL(y) \
678 UNUSED int l##y = edge[14-y];
679 #define PT(x) \
680 UNUSED int t##x = edge[16+x];
681 #define PREDICT_8x8_LOAD_TOPLEFT \
682 int lt = edge[15];
683 #define PREDICT_8x8_LOAD_LEFT \
684 PL(0) PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) PL(7)
685 #define PREDICT_8x8_LOAD_TOP \
686 PT(0) PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) PT(7)
687 #define PREDICT_8x8_LOAD_TOPRIGHT \
688 PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14) PT(15)
689
690 #define PREDICT_8x8_DC(v) \
691 for( int y = 0; y < 8; y++ ) { \
692 MPIXEL_X4( src+0 ) = v; \
693 MPIXEL_X4( src+4 ) = v; \
694 src += FDEC_STRIDE; \
695 }
696
predict_8x8_dc_128_c(pixel * src,pixel edge[36])697 static void predict_8x8_dc_128_c( pixel *src, pixel edge[36] )
698 {
699 PREDICT_8x8_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );
700 }
predict_8x8_dc_left_c(pixel * src,pixel edge[36])701 static void predict_8x8_dc_left_c( pixel *src, pixel edge[36] )
702 {
703 PREDICT_8x8_LOAD_LEFT
704 pixel4 dc = PIXEL_SPLAT_X4( (l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3 );
705 PREDICT_8x8_DC( dc );
706 }
predict_8x8_dc_top_c(pixel * src,pixel edge[36])707 static void predict_8x8_dc_top_c( pixel *src, pixel edge[36] )
708 {
709 PREDICT_8x8_LOAD_TOP
710 pixel4 dc = PIXEL_SPLAT_X4( (t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3 );
711 PREDICT_8x8_DC( dc );
712 }
x264_predict_8x8_dc_c(pixel * src,pixel edge[36])713 void x264_predict_8x8_dc_c( pixel *src, pixel edge[36] )
714 {
715 PREDICT_8x8_LOAD_LEFT
716 PREDICT_8x8_LOAD_TOP
717 pixel4 dc = PIXEL_SPLAT_X4( (l0+l1+l2+l3+l4+l5+l6+l7+t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4 );
718 PREDICT_8x8_DC( dc );
719 }
x264_predict_8x8_h_c(pixel * src,pixel edge[36])720 void x264_predict_8x8_h_c( pixel *src, pixel edge[36] )
721 {
722 PREDICT_8x8_LOAD_LEFT
723 #define ROW(y) MPIXEL_X4( src+y*FDEC_STRIDE+0 ) =\
724 MPIXEL_X4( src+y*FDEC_STRIDE+4 ) = PIXEL_SPLAT_X4( l##y );
725 ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
726 #undef ROW
727 }
x264_predict_8x8_v_c(pixel * src,pixel edge[36])728 void x264_predict_8x8_v_c( pixel *src, pixel edge[36] )
729 {
730 pixel4 top[2] = { MPIXEL_X4( edge+16 ),
731 MPIXEL_X4( edge+20 ) };
732 for( int y = 0; y < 8; y++ )
733 {
734 MPIXEL_X4( src+y*FDEC_STRIDE+0 ) = top[0];
735 MPIXEL_X4( src+y*FDEC_STRIDE+4 ) = top[1];
736 }
737 }
predict_8x8_ddl_c(pixel * src,pixel edge[36])738 static void predict_8x8_ddl_c( pixel *src, pixel edge[36] )
739 {
740 PREDICT_8x8_LOAD_TOP
741 PREDICT_8x8_LOAD_TOPRIGHT
742 SRC(0,0)= F2(t0,t1,t2);
743 SRC(0,1)=SRC(1,0)= F2(t1,t2,t3);
744 SRC(0,2)=SRC(1,1)=SRC(2,0)= F2(t2,t3,t4);
745 SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= F2(t3,t4,t5);
746 SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= F2(t4,t5,t6);
747 SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= F2(t5,t6,t7);
748 SRC(0,6)=SRC(1,5)=SRC(2,4)=SRC(3,3)=SRC(4,2)=SRC(5,1)=SRC(6,0)= F2(t6,t7,t8);
749 SRC(0,7)=SRC(1,6)=SRC(2,5)=SRC(3,4)=SRC(4,3)=SRC(5,2)=SRC(6,1)=SRC(7,0)= F2(t7,t8,t9);
750 SRC(1,7)=SRC(2,6)=SRC(3,5)=SRC(4,4)=SRC(5,3)=SRC(6,2)=SRC(7,1)= F2(t8,t9,t10);
751 SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= F2(t9,t10,t11);
752 SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= F2(t10,t11,t12);
753 SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= F2(t11,t12,t13);
754 SRC(5,7)=SRC(6,6)=SRC(7,5)= F2(t12,t13,t14);
755 SRC(6,7)=SRC(7,6)= F2(t13,t14,t15);
756 SRC(7,7)= F2(t14,t15,t15);
757 }
predict_8x8_ddr_c(pixel * src,pixel edge[36])758 static void predict_8x8_ddr_c( pixel *src, pixel edge[36] )
759 {
760 PREDICT_8x8_LOAD_TOP
761 PREDICT_8x8_LOAD_LEFT
762 PREDICT_8x8_LOAD_TOPLEFT
763 SRC(0,7)= F2(l7,l6,l5);
764 SRC(0,6)=SRC(1,7)= F2(l6,l5,l4);
765 SRC(0,5)=SRC(1,6)=SRC(2,7)= F2(l5,l4,l3);
766 SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= F2(l4,l3,l2);
767 SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= F2(l3,l2,l1);
768 SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= F2(l2,l1,l0);
769 SRC(0,1)=SRC(1,2)=SRC(2,3)=SRC(3,4)=SRC(4,5)=SRC(5,6)=SRC(6,7)= F2(l1,l0,lt);
770 SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)=SRC(4,4)=SRC(5,5)=SRC(6,6)=SRC(7,7)= F2(l0,lt,t0);
771 SRC(1,0)=SRC(2,1)=SRC(3,2)=SRC(4,3)=SRC(5,4)=SRC(6,5)=SRC(7,6)= F2(lt,t0,t1);
772 SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= F2(t0,t1,t2);
773 SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= F2(t1,t2,t3);
774 SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= F2(t2,t3,t4);
775 SRC(5,0)=SRC(6,1)=SRC(7,2)= F2(t3,t4,t5);
776 SRC(6,0)=SRC(7,1)= F2(t4,t5,t6);
777 SRC(7,0)= F2(t5,t6,t7);
778
779 }
predict_8x8_vr_c(pixel * src,pixel edge[36])780 static void predict_8x8_vr_c( pixel *src, pixel edge[36] )
781 {
782 PREDICT_8x8_LOAD_TOP
783 PREDICT_8x8_LOAD_LEFT
784 PREDICT_8x8_LOAD_TOPLEFT
785 SRC(0,6)= F2(l5,l4,l3);
786 SRC(0,7)= F2(l6,l5,l4);
787 SRC(0,4)=SRC(1,6)= F2(l3,l2,l1);
788 SRC(0,5)=SRC(1,7)= F2(l4,l3,l2);
789 SRC(0,2)=SRC(1,4)=SRC(2,6)= F2(l1,l0,lt);
790 SRC(0,3)=SRC(1,5)=SRC(2,7)= F2(l2,l1,l0);
791 SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= F2(l0,lt,t0);
792 SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= F1(lt,t0);
793 SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= F2(lt,t0,t1);
794 SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= F1(t0,t1);
795 SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= F2(t0,t1,t2);
796 SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= F1(t1,t2);
797 SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= F2(t1,t2,t3);
798 SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= F1(t2,t3);
799 SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= F2(t2,t3,t4);
800 SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= F1(t3,t4);
801 SRC(5,1)=SRC(6,3)=SRC(7,5)= F2(t3,t4,t5);
802 SRC(5,0)=SRC(6,2)=SRC(7,4)= F1(t4,t5);
803 SRC(6,1)=SRC(7,3)= F2(t4,t5,t6);
804 SRC(6,0)=SRC(7,2)= F1(t5,t6);
805 SRC(7,1)= F2(t5,t6,t7);
806 SRC(7,0)= F1(t6,t7);
807 }
predict_8x8_hd_c(pixel * src,pixel edge[36])808 static void predict_8x8_hd_c( pixel *src, pixel edge[36] )
809 {
810 PREDICT_8x8_LOAD_TOP
811 PREDICT_8x8_LOAD_LEFT
812 PREDICT_8x8_LOAD_TOPLEFT
813 int p1 = pack_pixel_1to2(F1(l6,l7), F2(l5,l6,l7));
814 int p2 = pack_pixel_1to2(F1(l5,l6), F2(l4,l5,l6));
815 int p3 = pack_pixel_1to2(F1(l4,l5), F2(l3,l4,l5));
816 int p4 = pack_pixel_1to2(F1(l3,l4), F2(l2,l3,l4));
817 int p5 = pack_pixel_1to2(F1(l2,l3), F2(l1,l2,l3));
818 int p6 = pack_pixel_1to2(F1(l1,l2), F2(l0,l1,l2));
819 int p7 = pack_pixel_1to2(F1(l0,l1), F2(lt,l0,l1));
820 int p8 = pack_pixel_1to2(F1(lt,l0), F2(l0,lt,t0));
821 int p9 = pack_pixel_1to2(F2(t1,t0,lt), F2(t2,t1,t0));
822 int p10 = pack_pixel_1to2(F2(t3,t2,t1), F2(t4,t3,t2));
823 int p11 = pack_pixel_1to2(F2(t5,t4,t3), F2(t6,t5,t4));
824 SRC_X4(0,7)= pack_pixel_2to4(p1,p2);
825 SRC_X4(0,6)= pack_pixel_2to4(p2,p3);
826 SRC_X4(4,7)=SRC_X4(0,5)= pack_pixel_2to4(p3,p4);
827 SRC_X4(4,6)=SRC_X4(0,4)= pack_pixel_2to4(p4,p5);
828 SRC_X4(4,5)=SRC_X4(0,3)= pack_pixel_2to4(p5,p6);
829 SRC_X4(4,4)=SRC_X4(0,2)= pack_pixel_2to4(p6,p7);
830 SRC_X4(4,3)=SRC_X4(0,1)= pack_pixel_2to4(p7,p8);
831 SRC_X4(4,2)=SRC_X4(0,0)= pack_pixel_2to4(p8,p9);
832 SRC_X4(4,1)= pack_pixel_2to4(p9,p10);
833 SRC_X4(4,0)= pack_pixel_2to4(p10,p11);
834 }
predict_8x8_vl_c(pixel * src,pixel edge[36])835 static void predict_8x8_vl_c( pixel *src, pixel edge[36] )
836 {
837 PREDICT_8x8_LOAD_TOP
838 PREDICT_8x8_LOAD_TOPRIGHT
839 SRC(0,0)= F1(t0,t1);
840 SRC(0,1)= F2(t0,t1,t2);
841 SRC(0,2)=SRC(1,0)= F1(t1,t2);
842 SRC(0,3)=SRC(1,1)= F2(t1,t2,t3);
843 SRC(0,4)=SRC(1,2)=SRC(2,0)= F1(t2,t3);
844 SRC(0,5)=SRC(1,3)=SRC(2,1)= F2(t2,t3,t4);
845 SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= F1(t3,t4);
846 SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= F2(t3,t4,t5);
847 SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= F1(t4,t5);
848 SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= F2(t4,t5,t6);
849 SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= F1(t5,t6);
850 SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= F2(t5,t6,t7);
851 SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= F1(t6,t7);
852 SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= F2(t6,t7,t8);
853 SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= F1(t7,t8);
854 SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= F2(t7,t8,t9);
855 SRC(5,6)=SRC(6,4)=SRC(7,2)= F1(t8,t9);
856 SRC(5,7)=SRC(6,5)=SRC(7,3)= F2(t8,t9,t10);
857 SRC(6,6)=SRC(7,4)= F1(t9,t10);
858 SRC(6,7)=SRC(7,5)= F2(t9,t10,t11);
859 SRC(7,6)= F1(t10,t11);
860 SRC(7,7)= F2(t10,t11,t12);
861 }
predict_8x8_hu_c(pixel * src,pixel edge[36])862 static void predict_8x8_hu_c( pixel *src, pixel edge[36] )
863 {
864 PREDICT_8x8_LOAD_LEFT
865 int p1 = pack_pixel_1to2(F1(l0,l1), F2(l0,l1,l2));
866 int p2 = pack_pixel_1to2(F1(l1,l2), F2(l1,l2,l3));
867 int p3 = pack_pixel_1to2(F1(l2,l3), F2(l2,l3,l4));
868 int p4 = pack_pixel_1to2(F1(l3,l4), F2(l3,l4,l5));
869 int p5 = pack_pixel_1to2(F1(l4,l5), F2(l4,l5,l6));
870 int p6 = pack_pixel_1to2(F1(l5,l6), F2(l5,l6,l7));
871 int p7 = pack_pixel_1to2(F1(l6,l7), F2(l6,l7,l7));
872 int p8 = pack_pixel_1to2(l7,l7);
873 SRC_X4(0,0)= pack_pixel_2to4(p1,p2);
874 SRC_X4(0,1)= pack_pixel_2to4(p2,p3);
875 SRC_X4(4,0)=SRC_X4(0,2)= pack_pixel_2to4(p3,p4);
876 SRC_X4(4,1)=SRC_X4(0,3)= pack_pixel_2to4(p4,p5);
877 SRC_X4(4,2)=SRC_X4(0,4)= pack_pixel_2to4(p5,p6);
878 SRC_X4(4,3)=SRC_X4(0,5)= pack_pixel_2to4(p6,p7);
879 SRC_X4(4,4)=SRC_X4(0,6)= pack_pixel_2to4(p7,p8);
880 SRC_X4(4,5)=SRC_X4(4,6)= SRC_X4(0,7) = SRC_X4(4,7) = pack_pixel_2to4(p8,p8);
881 }
882
883 /****************************************************************************
884 * Exported functions:
885 ****************************************************************************/
x264_predict_16x16_init(uint32_t cpu,x264_predict_t pf[7])886 void x264_predict_16x16_init( uint32_t cpu, x264_predict_t pf[7] )
887 {
888 pf[I_PRED_16x16_V ] = x264_predict_16x16_v_c;
889 pf[I_PRED_16x16_H ] = x264_predict_16x16_h_c;
890 pf[I_PRED_16x16_DC] = x264_predict_16x16_dc_c;
891 pf[I_PRED_16x16_P ] = x264_predict_16x16_p_c;
892 pf[I_PRED_16x16_DC_LEFT]= predict_16x16_dc_left_c;
893 pf[I_PRED_16x16_DC_TOP ]= predict_16x16_dc_top_c;
894 pf[I_PRED_16x16_DC_128 ]= predict_16x16_dc_128_c;
895
896 #if HAVE_MMX
897 x264_predict_16x16_init_mmx( cpu, pf );
898 #endif
899
900 #if HAVE_ALTIVEC
901 if( cpu&X264_CPU_ALTIVEC )
902 x264_predict_16x16_init_altivec( pf );
903 #endif
904
905 #if HAVE_ARMV6
906 x264_predict_16x16_init_arm( cpu, pf );
907 #endif
908
909 #if HAVE_AARCH64
910 x264_predict_16x16_init_aarch64( cpu, pf );
911 #endif
912
913 #if !HIGH_BIT_DEPTH
914 #if HAVE_MSA
915 if( cpu&X264_CPU_MSA )
916 {
917 pf[I_PRED_16x16_V ] = x264_intra_predict_vert_16x16_msa;
918 pf[I_PRED_16x16_H ] = x264_intra_predict_hor_16x16_msa;
919 pf[I_PRED_16x16_DC] = x264_intra_predict_dc_16x16_msa;
920 pf[I_PRED_16x16_P ] = x264_intra_predict_plane_16x16_msa;
921 pf[I_PRED_16x16_DC_LEFT]= x264_intra_predict_dc_left_16x16_msa;
922 pf[I_PRED_16x16_DC_TOP ]= x264_intra_predict_dc_top_16x16_msa;
923 pf[I_PRED_16x16_DC_128 ]= x264_intra_predict_dc_128_16x16_msa;
924 }
925 #endif
926 #endif
927 }
928
x264_predict_8x8c_init(uint32_t cpu,x264_predict_t pf[7])929 void x264_predict_8x8c_init( uint32_t cpu, x264_predict_t pf[7] )
930 {
931 pf[I_PRED_CHROMA_V ] = x264_predict_8x8c_v_c;
932 pf[I_PRED_CHROMA_H ] = x264_predict_8x8c_h_c;
933 pf[I_PRED_CHROMA_DC] = x264_predict_8x8c_dc_c;
934 pf[I_PRED_CHROMA_P ] = x264_predict_8x8c_p_c;
935 pf[I_PRED_CHROMA_DC_LEFT]= predict_8x8c_dc_left_c;
936 pf[I_PRED_CHROMA_DC_TOP ]= predict_8x8c_dc_top_c;
937 pf[I_PRED_CHROMA_DC_128 ]= predict_8x8c_dc_128_c;
938
939 #if HAVE_MMX
940 x264_predict_8x8c_init_mmx( cpu, pf );
941 #endif
942
943 #if HAVE_ALTIVEC
944 if( cpu&X264_CPU_ALTIVEC )
945 x264_predict_8x8c_init_altivec( pf );
946 #endif
947
948 #if HAVE_ARMV6
949 x264_predict_8x8c_init_arm( cpu, pf );
950 #endif
951
952 #if HAVE_AARCH64
953 x264_predict_8x8c_init_aarch64( cpu, pf );
954 #endif
955
956 #if !HIGH_BIT_DEPTH
957 #if HAVE_MSA
958 if( cpu&X264_CPU_MSA )
959 {
960 pf[I_PRED_CHROMA_P ] = x264_intra_predict_plane_8x8_msa;
961 }
962 #endif
963 #endif
964 }
965
x264_predict_8x16c_init(uint32_t cpu,x264_predict_t pf[7])966 void x264_predict_8x16c_init( uint32_t cpu, x264_predict_t pf[7] )
967 {
968 pf[I_PRED_CHROMA_V ] = x264_predict_8x16c_v_c;
969 pf[I_PRED_CHROMA_H ] = x264_predict_8x16c_h_c;
970 pf[I_PRED_CHROMA_DC] = x264_predict_8x16c_dc_c;
971 pf[I_PRED_CHROMA_P ] = x264_predict_8x16c_p_c;
972 pf[I_PRED_CHROMA_DC_LEFT]= predict_8x16c_dc_left_c;
973 pf[I_PRED_CHROMA_DC_TOP ]= predict_8x16c_dc_top_c;
974 pf[I_PRED_CHROMA_DC_128 ]= predict_8x16c_dc_128_c;
975
976 #if HAVE_MMX
977 x264_predict_8x16c_init_mmx( cpu, pf );
978 #endif
979
980 #if HAVE_ARMV6
981 x264_predict_8x16c_init_arm( cpu, pf );
982 #endif
983
984 #if HAVE_AARCH64
985 x264_predict_8x16c_init_aarch64( cpu, pf );
986 #endif
987 }
988
x264_predict_8x8_init(uint32_t cpu,x264_predict8x8_t pf[12],x264_predict_8x8_filter_t * predict_filter)989 void x264_predict_8x8_init( uint32_t cpu, x264_predict8x8_t pf[12], x264_predict_8x8_filter_t *predict_filter )
990 {
991 pf[I_PRED_8x8_V] = x264_predict_8x8_v_c;
992 pf[I_PRED_8x8_H] = x264_predict_8x8_h_c;
993 pf[I_PRED_8x8_DC] = x264_predict_8x8_dc_c;
994 pf[I_PRED_8x8_DDL] = predict_8x8_ddl_c;
995 pf[I_PRED_8x8_DDR] = predict_8x8_ddr_c;
996 pf[I_PRED_8x8_VR] = predict_8x8_vr_c;
997 pf[I_PRED_8x8_HD] = predict_8x8_hd_c;
998 pf[I_PRED_8x8_VL] = predict_8x8_vl_c;
999 pf[I_PRED_8x8_HU] = predict_8x8_hu_c;
1000 pf[I_PRED_8x8_DC_LEFT]= predict_8x8_dc_left_c;
1001 pf[I_PRED_8x8_DC_TOP] = predict_8x8_dc_top_c;
1002 pf[I_PRED_8x8_DC_128] = predict_8x8_dc_128_c;
1003 *predict_filter = predict_8x8_filter_c;
1004
1005 #if HAVE_MMX
1006 x264_predict_8x8_init_mmx( cpu, pf, predict_filter );
1007 #endif
1008
1009 #if HAVE_ARMV6
1010 x264_predict_8x8_init_arm( cpu, pf, predict_filter );
1011 #endif
1012
1013 #if HAVE_AARCH64
1014 x264_predict_8x8_init_aarch64( cpu, pf, predict_filter );
1015 #endif
1016
1017 #if !HIGH_BIT_DEPTH
1018 #if HAVE_MSA
1019 if( cpu&X264_CPU_MSA )
1020 {
1021 pf[I_PRED_8x8_DDL] = x264_intra_predict_ddl_8x8_msa;
1022 }
1023 #endif
1024 #endif
1025 }
1026
x264_predict_4x4_init(uint32_t cpu,x264_predict_t pf[12])1027 void x264_predict_4x4_init( uint32_t cpu, x264_predict_t pf[12] )
1028 {
1029 pf[I_PRED_4x4_V] = x264_predict_4x4_v_c;
1030 pf[I_PRED_4x4_H] = x264_predict_4x4_h_c;
1031 pf[I_PRED_4x4_DC] = x264_predict_4x4_dc_c;
1032 pf[I_PRED_4x4_DDL] = predict_4x4_ddl_c;
1033 pf[I_PRED_4x4_DDR] = predict_4x4_ddr_c;
1034 pf[I_PRED_4x4_VR] = predict_4x4_vr_c;
1035 pf[I_PRED_4x4_HD] = predict_4x4_hd_c;
1036 pf[I_PRED_4x4_VL] = predict_4x4_vl_c;
1037 pf[I_PRED_4x4_HU] = predict_4x4_hu_c;
1038 pf[I_PRED_4x4_DC_LEFT]= predict_4x4_dc_left_c;
1039 pf[I_PRED_4x4_DC_TOP] = predict_4x4_dc_top_c;
1040 pf[I_PRED_4x4_DC_128] = predict_4x4_dc_128_c;
1041
1042 #if HAVE_MMX
1043 x264_predict_4x4_init_mmx( cpu, pf );
1044 #endif
1045
1046 #if HAVE_ARMV6
1047 x264_predict_4x4_init_arm( cpu, pf );
1048 #endif
1049
1050 #if HAVE_AARCH64
1051 x264_predict_4x4_init_aarch64( cpu, pf );
1052 #endif
1053 }
1054
1055