1 /*****************************************************************************
2 * predict.c: h264 encoder
3 *****************************************************************************
4 * Copyright (C) 2003-2008 x264 project
6 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
7 * Loren Merritt <lorenm@u.washington.edu>
8 * Fiona Glaser <fiona@x264.com>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
23 *****************************************************************************/
25 /* predict4x4 are inspired from ffmpeg h264 decoder */
31 # include "x86/predict.h"
34 # include "ppc/predict.h"
37 # include "arm/predict.h"
40 /****************************************************************************
41 * 16x16 prediction for intra luma block
42 ****************************************************************************/
44 #define PREDICT_16x16_DC(v) \
45 for( int i = 0; i < 16; i++ )\
54 static void predict_16x16_dc( uint8_t *src )
58 for( int i = 0; i < 16; i++ )
60 dc += src[-1 + i * FDEC_STRIDE];
61 dc += src[i - FDEC_STRIDE];
63 dc = (( dc + 16 ) >> 5) * 0x01010101;
67 static void predict_16x16_dc_left( uint8_t *src )
71 for( int i = 0; i < 16; i++ )
72 dc += src[-1 + i * FDEC_STRIDE];
73 dc = (( dc + 8 ) >> 4) * 0x01010101;
77 static void predict_16x16_dc_top( uint8_t *src )
81 for( int i = 0; i < 16; i++ )
82 dc += src[i - FDEC_STRIDE];
83 dc = (( dc + 8 ) >> 4) * 0x01010101;
87 static void predict_16x16_dc_128( uint8_t *src )
89 PREDICT_16x16_DC(0x80808080);
91 static void predict_16x16_h( uint8_t *src )
93 for( int i = 0; i < 16; i++ )
95 const uint32_t v = 0x01010101 * src[-1];
103 static void predict_16x16_v( uint8_t *src )
105 uint32_t v0 = M32( &src[ 0-FDEC_STRIDE] );
106 uint32_t v1 = M32( &src[ 4-FDEC_STRIDE] );
107 uint32_t v2 = M32( &src[ 8-FDEC_STRIDE] );
108 uint32_t v3 = M32( &src[12-FDEC_STRIDE] );
110 for( int i = 0; i < 16; i++ )
119 static void predict_16x16_p( uint8_t *src )
123 /* calculate H and V */
124 for( int i = 0; i <= 7; i++ )
126 H += ( i + 1 ) * ( src[ 8 + i - FDEC_STRIDE ] - src[6 -i -FDEC_STRIDE] );
127 V += ( i + 1 ) * ( src[-1 + (8+i)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );
130 int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[15 - FDEC_STRIDE] );
131 int b = ( 5 * H + 32 ) >> 6;
132 int c = ( 5 * V + 32 ) >> 6;
134 int i00 = a - b * 7 - c * 7 + 16;
136 for( int y = 0; y < 16; y++ )
139 for( int x = 0; x < 16; x++ )
141 src[x] = x264_clip_uint8( pix>>5 );
150 /****************************************************************************
151 * 8x8 prediction for intra chroma block
152 ****************************************************************************/
154 static void predict_8x8c_dc_128( uint8_t *src )
156 for( int y = 0; y < 8; y++ )
158 M32( src+0 ) = 0x80808080;
159 M32( src+4 ) = 0x80808080;
163 static void predict_8x8c_dc_left( uint8_t *src )
165 uint32_t dc0 = 0, dc1 = 0;
167 for( int y = 0; y < 4; y++ )
169 dc0 += src[y * FDEC_STRIDE - 1];
170 dc1 += src[(y+4) * FDEC_STRIDE - 1];
172 dc0 = (( dc0 + 2 ) >> 2)*0x01010101;
173 dc1 = (( dc1 + 2 ) >> 2)*0x01010101;
175 for( int y = 0; y < 4; y++ )
181 for( int y = 0; y < 4; y++ )
189 static void predict_8x8c_dc_top( uint8_t *src )
191 uint32_t dc0 = 0, dc1 = 0;
193 for( int x = 0; x < 4; x++ )
195 dc0 += src[x - FDEC_STRIDE];
196 dc1 += src[x + 4 - FDEC_STRIDE];
198 dc0 = (( dc0 + 2 ) >> 2)*0x01010101;
199 dc1 = (( dc1 + 2 ) >> 2)*0x01010101;
201 for( int y = 0; y < 8; y++ )
208 static void predict_8x8c_dc( uint8_t *src )
210 int s0 = 0, s1 = 0, s2 = 0, s3 = 0;
217 for( int i = 0; i < 4; i++ )
219 s0 += src[i - FDEC_STRIDE];
220 s1 += src[i + 4 - FDEC_STRIDE];
221 s2 += src[-1 + i * FDEC_STRIDE];
222 s3 += src[-1 + (i+4)*FDEC_STRIDE];
228 uint32_t dc0 = (( s0 + s2 + 4 ) >> 3)*0x01010101;
229 uint32_t dc1 = (( s1 + 2 ) >> 2)*0x01010101;
230 uint32_t dc2 = (( s3 + 2 ) >> 2)*0x01010101;
231 uint32_t dc3 = (( s1 + s3 + 4 ) >> 3)*0x01010101;
233 for( int y = 0; y < 4; y++ )
240 for( int y = 0; y < 4; y++ )
247 static void predict_8x8c_h( uint8_t *src )
249 for( int i = 0; i < 8; i++ )
251 uint32_t v = 0x01010101 * src[-1];
257 static void predict_8x8c_v( uint8_t *src )
259 uint32_t v0 = M32( src+0-FDEC_STRIDE );
260 uint32_t v1 = M32( src+4-FDEC_STRIDE );
262 for( int i = 0; i < 8; i++ )
269 static void predict_8x8c_p( uint8_t *src )
273 for( int i = 0; i < 4; i++ )
275 H += ( i + 1 ) * ( src[4+i - FDEC_STRIDE] - src[2 - i -FDEC_STRIDE] );
276 V += ( i + 1 ) * ( src[-1 +(i+4)*FDEC_STRIDE] - src[-1+(2-i)*FDEC_STRIDE] );
279 int a = 16 * ( src[-1+7*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );
280 int b = ( 17 * H + 16 ) >> 5;
281 int c = ( 17 * V + 16 ) >> 5;
282 int i00 = a -3*b -3*c + 16;
284 for( int y = 0; y < 8; y++ )
287 for( int x = 0; x < 8; x++ )
289 src[x] = x264_clip_uint8( pix>>5 );
297 /****************************************************************************
298 * 4x4 prediction for intra luma block
299 ****************************************************************************/
301 #define SRC(x,y) src[(x)+(y)*FDEC_STRIDE]
302 #define SRC32(x,y) M32( &SRC(x,y) )
304 #define PREDICT_4x4_DC(v)\
305 SRC32(0,0) = SRC32(0,1) = SRC32(0,2) = SRC32(0,3) = v;
307 static void predict_4x4_dc_128( uint8_t *src )
309 PREDICT_4x4_DC(0x80808080);
311 static void predict_4x4_dc_left( uint8_t *src )
313 uint32_t dc = ((SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) + 2) >> 2) * 0x01010101;
316 static void predict_4x4_dc_top( uint8_t *src )
318 uint32_t dc = ((SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 2) >> 2) * 0x01010101;
321 static void predict_4x4_dc( uint8_t *src )
323 uint32_t dc = ((SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) +
324 SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 4) >> 3) * 0x01010101;
327 static void predict_4x4_h( uint8_t *src )
329 SRC32(0,0) = SRC(-1,0) * 0x01010101;
330 SRC32(0,1) = SRC(-1,1) * 0x01010101;
331 SRC32(0,2) = SRC(-1,2) * 0x01010101;
332 SRC32(0,3) = SRC(-1,3) * 0x01010101;
334 static void predict_4x4_v( uint8_t *src )
336 PREDICT_4x4_DC(SRC32(0,-1));
339 #define PREDICT_4x4_LOAD_LEFT\
343 UNUSED int l3 = SRC(-1,3);
345 #define PREDICT_4x4_LOAD_TOP\
349 UNUSED int t3 = SRC(3,-1);
351 #define PREDICT_4x4_LOAD_TOP_RIGHT\
355 UNUSED int t7 = SRC(7,-1);
357 #define F1(a,b) (((a)+(b)+1)>>1)
358 #define F2(a,b,c) (((a)+2*(b)+(c)+2)>>2)
360 static void predict_4x4_ddl( uint8_t *src )
363 PREDICT_4x4_LOAD_TOP_RIGHT
364 SRC(0,0)= F2(t0,t1,t2);
365 SRC(1,0)=SRC(0,1)= F2(t1,t2,t3);
366 SRC(2,0)=SRC(1,1)=SRC(0,2)= F2(t2,t3,t4);
367 SRC(3,0)=SRC(2,1)=SRC(1,2)=SRC(0,3)= F2(t3,t4,t5);
368 SRC(3,1)=SRC(2,2)=SRC(1,3)= F2(t4,t5,t6);
369 SRC(3,2)=SRC(2,3)= F2(t5,t6,t7);
370 SRC(3,3)= F2(t6,t7,t7);
372 static void predict_4x4_ddr( uint8_t *src )
375 PREDICT_4x4_LOAD_LEFT
377 SRC(3,0)= F2(t3,t2,t1);
378 SRC(2,0)=SRC(3,1)= F2(t2,t1,t0);
379 SRC(1,0)=SRC(2,1)=SRC(3,2)= F2(t1,t0,lt);
380 SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)= F2(t0,lt,l0);
381 SRC(0,1)=SRC(1,2)=SRC(2,3)= F2(lt,l0,l1);
382 SRC(0,2)=SRC(1,3)= F2(l0,l1,l2);
383 SRC(0,3)= F2(l1,l2,l3);
386 static void predict_4x4_vr( uint8_t *src )
389 PREDICT_4x4_LOAD_LEFT
391 SRC(0,3)= F2(l2,l1,l0);
392 SRC(0,2)= F2(l1,l0,lt);
393 SRC(0,1)=SRC(1,3)= F2(l0,lt,t0);
394 SRC(0,0)=SRC(1,2)= F1(lt,t0);
395 SRC(1,1)=SRC(2,3)= F2(lt,t0,t1);
396 SRC(1,0)=SRC(2,2)= F1(t0,t1);
397 SRC(2,1)=SRC(3,3)= F2(t0,t1,t2);
398 SRC(2,0)=SRC(3,2)= F1(t1,t2);
399 SRC(3,1)= F2(t1,t2,t3);
403 static void predict_4x4_hd( uint8_t *src )
406 PREDICT_4x4_LOAD_LEFT
409 SRC(1,3)= F2(l1,l2,l3);
410 SRC(0,2)=SRC(2,3)= F1(l1,l2);
411 SRC(1,2)=SRC(3,3)= F2(l0,l1,l2);
412 SRC(0,1)=SRC(2,2)= F1(l0,l1);
413 SRC(1,1)=SRC(3,2)= F2(lt,l0,l1);
414 SRC(0,0)=SRC(2,1)= F1(lt,l0);
415 SRC(1,0)=SRC(3,1)= F2(t0,lt,l0);
416 SRC(2,0)= F2(t1,t0,lt);
417 SRC(3,0)= F2(t2,t1,t0);
420 static void predict_4x4_vl( uint8_t *src )
423 PREDICT_4x4_LOAD_TOP_RIGHT
425 SRC(0,1)= F2(t0,t1,t2);
426 SRC(1,0)=SRC(0,2)= F1(t1,t2);
427 SRC(1,1)=SRC(0,3)= F2(t1,t2,t3);
428 SRC(2,0)=SRC(1,2)= F1(t2,t3);
429 SRC(2,1)=SRC(1,3)= F2(t2,t3,t4);
430 SRC(3,0)=SRC(2,2)= F1(t3,t4);
431 SRC(3,1)=SRC(2,3)= F2(t3,t4,t5);
433 SRC(3,3)= F2(t4,t5,t6);
436 static void predict_4x4_hu( uint8_t *src )
438 PREDICT_4x4_LOAD_LEFT
440 SRC(1,0)= F2(l0,l1,l2);
441 SRC(2,0)=SRC(0,1)= F1(l1,l2);
442 SRC(3,0)=SRC(1,1)= F2(l1,l2,l3);
443 SRC(2,1)=SRC(0,2)= F1(l2,l3);
444 SRC(3,1)=SRC(1,2)= F2(l2,l3,l3);
445 SRC(3,2)=SRC(1,3)=SRC(0,3)=
446 SRC(2,2)=SRC(2,3)=SRC(3,3)= l3;
449 /****************************************************************************
450 * 8x8 prediction for intra luma block
451 ****************************************************************************/
454 edge[14-y] = F2(SRC(-1,y-1), SRC(-1,y), SRC(-1,y+1));
456 edge[16+x] = F2(SRC(x-1,-1), SRC(x,-1), SRC(x+1,-1));
458 static void predict_8x8_filter( uint8_t *src, uint8_t edge[33], int i_neighbor, int i_filters )
460 /* edge[7..14] = l7..l0
462 * edge[16..31] = t0 .. t15
465 int have_lt = i_neighbor & MB_TOPLEFT;
466 if( i_filters & MB_LEFT )
468 edge[15] = (SRC(0,-1) + 2*SRC(-1,-1) + SRC(-1,0) + 2) >> 2;
469 edge[14] = ((have_lt ? SRC(-1,-1) : SRC(-1,0))
470 + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2;
471 PL(1) PL(2) PL(3) PL(4) PL(5) PL(6)
472 edge[7] = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2;
475 if( i_filters & MB_TOP )
477 int have_tr = i_neighbor & MB_TOPRIGHT;
478 edge[16] = ((have_lt ? SRC(-1,-1) : SRC(0,-1))
479 + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2;
480 PT(1) PT(2) PT(3) PT(4) PT(5) PT(6)
481 edge[23] = (SRC(6,-1) + 2*SRC(7,-1)
482 + (have_tr ? SRC(8,-1) : SRC(7,-1)) + 2) >> 2;
484 if( i_filters & MB_TOPRIGHT )
488 PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14)
490 edge[32] = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2;
494 M64( edge+24 ) = SRC(7,-1) * 0x0101010101010101ULL;
495 edge[32] = SRC(7,-1);
505 UNUSED int l##y = edge[14-y];
507 UNUSED int t##x = edge[16+x];
508 #define PREDICT_8x8_LOAD_TOPLEFT \
510 #define PREDICT_8x8_LOAD_LEFT \
511 PL(0) PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) PL(7)
512 #define PREDICT_8x8_LOAD_TOP \
513 PT(0) PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) PT(7)
514 #define PREDICT_8x8_LOAD_TOPRIGHT \
515 PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14) PT(15)
517 #define PREDICT_8x8_DC(v) \
518 for( int y = 0; y < 8; y++ ) { \
521 src += FDEC_STRIDE; \
524 static void predict_8x8_dc_128( uint8_t *src, uint8_t edge[33] )
526 PREDICT_8x8_DC(0x80808080);
528 static void predict_8x8_dc_left( uint8_t *src, uint8_t edge[33] )
530 PREDICT_8x8_LOAD_LEFT
531 uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3) * 0x01010101;
534 static void predict_8x8_dc_top( uint8_t *src, uint8_t edge[33] )
537 uint32_t dc = ((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3) * 0x01010101;
540 static void predict_8x8_dc( uint8_t *src, uint8_t edge[33] )
542 PREDICT_8x8_LOAD_LEFT
544 uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7
545 +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4) * 0x01010101;
548 static void predict_8x8_h( uint8_t *src, uint8_t edge[33] )
550 PREDICT_8x8_LOAD_LEFT
551 #define ROW(y) M32( src+y*FDEC_STRIDE+0 ) =\
552 M32( src+y*FDEC_STRIDE+4 ) = 0x01010101U * l##y;
553 ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
556 static void predict_8x8_v( uint8_t *src, uint8_t edge[33] )
558 uint64_t top = M64( edge+16 );
559 for( int y = 0; y < 8; y++ )
560 M64( src+y*FDEC_STRIDE ) = top;
562 static void predict_8x8_ddl( uint8_t *src, uint8_t edge[33] )
565 PREDICT_8x8_LOAD_TOPRIGHT
566 SRC(0,0)= F2(t0,t1,t2);
567 SRC(0,1)=SRC(1,0)= F2(t1,t2,t3);
568 SRC(0,2)=SRC(1,1)=SRC(2,0)= F2(t2,t3,t4);
569 SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= F2(t3,t4,t5);
570 SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= F2(t4,t5,t6);
571 SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= F2(t5,t6,t7);
572 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);
573 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);
574 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);
575 SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= F2(t9,t10,t11);
576 SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= F2(t10,t11,t12);
577 SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= F2(t11,t12,t13);
578 SRC(5,7)=SRC(6,6)=SRC(7,5)= F2(t12,t13,t14);
579 SRC(6,7)=SRC(7,6)= F2(t13,t14,t15);
580 SRC(7,7)= F2(t14,t15,t15);
582 static void predict_8x8_ddr( uint8_t *src, uint8_t edge[33] )
585 PREDICT_8x8_LOAD_LEFT
586 PREDICT_8x8_LOAD_TOPLEFT
587 SRC(0,7)= F2(l7,l6,l5);
588 SRC(0,6)=SRC(1,7)= F2(l6,l5,l4);
589 SRC(0,5)=SRC(1,6)=SRC(2,7)= F2(l5,l4,l3);
590 SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= F2(l4,l3,l2);
591 SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= F2(l3,l2,l1);
592 SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= F2(l2,l1,l0);
593 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);
594 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);
595 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);
596 SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= F2(t0,t1,t2);
597 SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= F2(t1,t2,t3);
598 SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= F2(t2,t3,t4);
599 SRC(5,0)=SRC(6,1)=SRC(7,2)= F2(t3,t4,t5);
600 SRC(6,0)=SRC(7,1)= F2(t4,t5,t6);
601 SRC(7,0)= F2(t5,t6,t7);
604 static void predict_8x8_vr( uint8_t *src, uint8_t edge[33] )
607 PREDICT_8x8_LOAD_LEFT
608 PREDICT_8x8_LOAD_TOPLEFT
609 SRC(0,6)= F2(l5,l4,l3);
610 SRC(0,7)= F2(l6,l5,l4);
611 SRC(0,4)=SRC(1,6)= F2(l3,l2,l1);
612 SRC(0,5)=SRC(1,7)= F2(l4,l3,l2);
613 SRC(0,2)=SRC(1,4)=SRC(2,6)= F2(l1,l0,lt);
614 SRC(0,3)=SRC(1,5)=SRC(2,7)= F2(l2,l1,l0);
615 SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= F2(l0,lt,t0);
616 SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= F1(lt,t0);
617 SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= F2(lt,t0,t1);
618 SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= F1(t0,t1);
619 SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= F2(t0,t1,t2);
620 SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= F1(t1,t2);
621 SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= F2(t1,t2,t3);
622 SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= F1(t2,t3);
623 SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= F2(t2,t3,t4);
624 SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= F1(t3,t4);
625 SRC(5,1)=SRC(6,3)=SRC(7,5)= F2(t3,t4,t5);
626 SRC(5,0)=SRC(6,2)=SRC(7,4)= F1(t4,t5);
627 SRC(6,1)=SRC(7,3)= F2(t4,t5,t6);
628 SRC(6,0)=SRC(7,2)= F1(t5,t6);
629 SRC(7,1)= F2(t5,t6,t7);
632 static void predict_8x8_hd( uint8_t *src, uint8_t edge[33] )
635 PREDICT_8x8_LOAD_LEFT
636 PREDICT_8x8_LOAD_TOPLEFT
637 int p1 = pack8to16(F1(l6,l7), F2(l5,l6,l7));
638 int p2 = pack8to16(F1(l5,l6), F2(l4,l5,l6));
639 int p3 = pack8to16(F1(l4,l5), F2(l3,l4,l5));
640 int p4 = pack8to16(F1(l3,l4), F2(l2,l3,l4));
641 int p5 = pack8to16(F1(l2,l3), F2(l1,l2,l3));
642 int p6 = pack8to16(F1(l1,l2), F2(l0,l1,l2));
643 int p7 = pack8to16(F1(l0,l1), F2(lt,l0,l1));
644 int p8 = pack8to16(F1(lt,l0), F2(l0,lt,t0));
645 int p9 = pack8to16(F2(t1,t0,lt), F2(t2,t1,t0));
646 int p10 = pack8to16(F2(t3,t2,t1), F2(t4,t3,t2));
647 int p11 = pack8to16(F2(t5,t4,t3), F2(t6,t5,t4));
648 SRC32(0,7)= pack16to32(p1,p2);
649 SRC32(0,6)= pack16to32(p2,p3);
650 SRC32(4,7)=SRC32(0,5)= pack16to32(p3,p4);
651 SRC32(4,6)=SRC32(0,4)= pack16to32(p4,p5);
652 SRC32(4,5)=SRC32(0,3)= pack16to32(p5,p6);
653 SRC32(4,4)=SRC32(0,2)= pack16to32(p6,p7);
654 SRC32(4,3)=SRC32(0,1)= pack16to32(p7,p8);
655 SRC32(4,2)=SRC32(0,0)= pack16to32(p8,p9);
656 SRC32(4,1)= pack16to32(p9,p10);
657 SRC32(4,0)= pack16to32(p10,p11);
659 static void predict_8x8_vl( uint8_t *src, uint8_t edge[33] )
662 PREDICT_8x8_LOAD_TOPRIGHT
664 SRC(0,1)= F2(t0,t1,t2);
665 SRC(0,2)=SRC(1,0)= F1(t1,t2);
666 SRC(0,3)=SRC(1,1)= F2(t1,t2,t3);
667 SRC(0,4)=SRC(1,2)=SRC(2,0)= F1(t2,t3);
668 SRC(0,5)=SRC(1,3)=SRC(2,1)= F2(t2,t3,t4);
669 SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= F1(t3,t4);
670 SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= F2(t3,t4,t5);
671 SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= F1(t4,t5);
672 SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= F2(t4,t5,t6);
673 SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= F1(t5,t6);
674 SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= F2(t5,t6,t7);
675 SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= F1(t6,t7);
676 SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= F2(t6,t7,t8);
677 SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= F1(t7,t8);
678 SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= F2(t7,t8,t9);
679 SRC(5,6)=SRC(6,4)=SRC(7,2)= F1(t8,t9);
680 SRC(5,7)=SRC(6,5)=SRC(7,3)= F2(t8,t9,t10);
681 SRC(6,6)=SRC(7,4)= F1(t9,t10);
682 SRC(6,7)=SRC(7,5)= F2(t9,t10,t11);
683 SRC(7,6)= F1(t10,t11);
684 SRC(7,7)= F2(t10,t11,t12);
686 static void predict_8x8_hu( uint8_t *src, uint8_t edge[33] )
688 PREDICT_8x8_LOAD_LEFT
689 int p1 = pack8to16(F1(l0,l1), F2(l0,l1,l2));
690 int p2 = pack8to16(F1(l1,l2), F2(l1,l2,l3));
691 int p3 = pack8to16(F1(l2,l3), F2(l2,l3,l4));
692 int p4 = pack8to16(F1(l3,l4), F2(l3,l4,l5));
693 int p5 = pack8to16(F1(l4,l5), F2(l4,l5,l6));
694 int p6 = pack8to16(F1(l5,l6), F2(l5,l6,l7));
695 int p7 = pack8to16(F1(l6,l7), F2(l6,l7,l7));
696 int p8 = pack8to16(l7,l7);
697 SRC32(0,0)= pack16to32(p1,p2);
698 SRC32(0,1)= pack16to32(p2,p3);
699 SRC32(4,0)=SRC32(0,2)= pack16to32(p3,p4);
700 SRC32(4,1)=SRC32(0,3)= pack16to32(p4,p5);
701 SRC32(4,2)=SRC32(0,4)= pack16to32(p5,p6);
702 SRC32(4,3)=SRC32(0,5)= pack16to32(p6,p7);
703 SRC32(4,4)=SRC32(0,6)= pack16to32(p7,p8);
704 SRC32(4,5)=SRC32(4,6)= SRC32(0,7) = SRC32(4,7) = pack16to32(p8,p8);
707 /****************************************************************************
708 * Exported functions:
709 ****************************************************************************/
710 void x264_predict_16x16_init( int cpu, x264_predict_t pf[7] )
712 pf[I_PRED_16x16_V ] = predict_16x16_v;
713 pf[I_PRED_16x16_H ] = predict_16x16_h;
714 pf[I_PRED_16x16_DC] = predict_16x16_dc;
715 pf[I_PRED_16x16_P ] = predict_16x16_p;
716 pf[I_PRED_16x16_DC_LEFT]= predict_16x16_dc_left;
717 pf[I_PRED_16x16_DC_TOP ]= predict_16x16_dc_top;
718 pf[I_PRED_16x16_DC_128 ]= predict_16x16_dc_128;
721 x264_predict_16x16_init_mmx( cpu, pf );
725 if( cpu&X264_CPU_ALTIVEC )
726 x264_predict_16x16_init_altivec( pf );
730 x264_predict_16x16_init_arm( cpu, pf );
734 void x264_predict_8x8c_init( int cpu, x264_predict_t pf[7] )
736 pf[I_PRED_CHROMA_V ] = predict_8x8c_v;
737 pf[I_PRED_CHROMA_H ] = predict_8x8c_h;
738 pf[I_PRED_CHROMA_DC] = predict_8x8c_dc;
739 pf[I_PRED_CHROMA_P ] = predict_8x8c_p;
740 pf[I_PRED_CHROMA_DC_LEFT]= predict_8x8c_dc_left;
741 pf[I_PRED_CHROMA_DC_TOP ]= predict_8x8c_dc_top;
742 pf[I_PRED_CHROMA_DC_128 ]= predict_8x8c_dc_128;
745 x264_predict_8x8c_init_mmx( cpu, pf );
749 if( cpu&X264_CPU_ALTIVEC )
750 x264_predict_8x8c_init_altivec( pf );
754 x264_predict_8x8c_init_arm( cpu, pf );
758 void x264_predict_8x8_init( int cpu, x264_predict8x8_t pf[12], x264_predict_8x8_filter_t *predict_filter )
760 pf[I_PRED_8x8_V] = predict_8x8_v;
761 pf[I_PRED_8x8_H] = predict_8x8_h;
762 pf[I_PRED_8x8_DC] = predict_8x8_dc;
763 pf[I_PRED_8x8_DDL] = predict_8x8_ddl;
764 pf[I_PRED_8x8_DDR] = predict_8x8_ddr;
765 pf[I_PRED_8x8_VR] = predict_8x8_vr;
766 pf[I_PRED_8x8_HD] = predict_8x8_hd;
767 pf[I_PRED_8x8_VL] = predict_8x8_vl;
768 pf[I_PRED_8x8_HU] = predict_8x8_hu;
769 pf[I_PRED_8x8_DC_LEFT]= predict_8x8_dc_left;
770 pf[I_PRED_8x8_DC_TOP] = predict_8x8_dc_top;
771 pf[I_PRED_8x8_DC_128] = predict_8x8_dc_128;
772 *predict_filter = predict_8x8_filter;
775 x264_predict_8x8_init_mmx( cpu, pf, predict_filter );
779 x264_predict_8x8_init_arm( cpu, pf, predict_filter );
783 void x264_predict_4x4_init( int cpu, x264_predict_t pf[12] )
785 pf[I_PRED_4x4_V] = predict_4x4_v;
786 pf[I_PRED_4x4_H] = predict_4x4_h;
787 pf[I_PRED_4x4_DC] = predict_4x4_dc;
788 pf[I_PRED_4x4_DDL] = predict_4x4_ddl;
789 pf[I_PRED_4x4_DDR] = predict_4x4_ddr;
790 pf[I_PRED_4x4_VR] = predict_4x4_vr;
791 pf[I_PRED_4x4_HD] = predict_4x4_hd;
792 pf[I_PRED_4x4_VL] = predict_4x4_vl;
793 pf[I_PRED_4x4_HU] = predict_4x4_hu;
794 pf[I_PRED_4x4_DC_LEFT]= predict_4x4_dc_left;
795 pf[I_PRED_4x4_DC_TOP] = predict_4x4_dc_top;
796 pf[I_PRED_4x4_DC_128] = predict_4x4_dc_128;
799 x264_predict_4x4_init_mmx( cpu, pf );
803 x264_predict_4x4_init_arm( cpu, pf );