1 /*****************************************************************************
2 * predict.c: intra prediction
3 *****************************************************************************
4 * Copyright (C) 2003-2014 x264 project
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>
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.
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.
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.
25 * This program is also available under a commercial proprietary license.
26 * For more information, contact us at licensing@x264.com.
27 *****************************************************************************/
29 /* predict4x4 are inspired from ffmpeg h264 decoder */
35 # include "x86/predict.h"
38 # include "ppc/predict.h"
41 # include "arm/predict.h"
44 # include "aarch64/predict.h"
47 /****************************************************************************
48 * 16x16 prediction for intra luma block
49 ****************************************************************************/
51 #define PREDICT_16x16_DC(v)\
52 for( int i = 0; i < 16; i++ )\
54 MPIXEL_X4( src+ 0 ) = v;\
55 MPIXEL_X4( src+ 4 ) = v;\
56 MPIXEL_X4( src+ 8 ) = v;\
57 MPIXEL_X4( src+12 ) = v;\
61 void x264_predict_16x16_dc_c( pixel *src )
65 for( int i = 0; i < 16; i++ )
67 dc += src[-1 + i * FDEC_STRIDE];
68 dc += src[i - FDEC_STRIDE];
70 pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 16 ) >> 5 );
72 PREDICT_16x16_DC( dcsplat );
74 static void x264_predict_16x16_dc_left_c( pixel *src )
78 for( int i = 0; i < 16; i++ )
79 dc += src[-1 + i * FDEC_STRIDE];
80 pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 );
82 PREDICT_16x16_DC( dcsplat );
84 static void x264_predict_16x16_dc_top_c( pixel *src )
88 for( int i = 0; i < 16; i++ )
89 dc += src[i - FDEC_STRIDE];
90 pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 );
92 PREDICT_16x16_DC( dcsplat );
94 static void x264_predict_16x16_dc_128_c( pixel *src )
96 PREDICT_16x16_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );
98 void x264_predict_16x16_h_c( pixel *src )
100 for( int i = 0; i < 16; i++ )
102 const pixel4 v = PIXEL_SPLAT_X4( src[-1] );
103 MPIXEL_X4( src+ 0 ) = v;
104 MPIXEL_X4( src+ 4 ) = v;
105 MPIXEL_X4( src+ 8 ) = v;
106 MPIXEL_X4( src+12 ) = v;
110 void x264_predict_16x16_v_c( pixel *src )
112 pixel4 v0 = MPIXEL_X4( &src[ 0-FDEC_STRIDE] );
113 pixel4 v1 = MPIXEL_X4( &src[ 4-FDEC_STRIDE] );
114 pixel4 v2 = MPIXEL_X4( &src[ 8-FDEC_STRIDE] );
115 pixel4 v3 = MPIXEL_X4( &src[12-FDEC_STRIDE] );
117 for( int i = 0; i < 16; i++ )
119 MPIXEL_X4( src+ 0 ) = v0;
120 MPIXEL_X4( src+ 4 ) = v1;
121 MPIXEL_X4( src+ 8 ) = v2;
122 MPIXEL_X4( src+12 ) = v3;
126 void x264_predict_16x16_p_c( pixel *src )
130 /* calculate H and V */
131 for( int i = 0; i <= 7; i++ )
133 H += ( i + 1 ) * ( src[ 8 + i - FDEC_STRIDE ] - src[6 -i -FDEC_STRIDE] );
134 V += ( i + 1 ) * ( src[-1 + (8+i)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );
137 int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[15 - FDEC_STRIDE] );
138 int b = ( 5 * H + 32 ) >> 6;
139 int c = ( 5 * V + 32 ) >> 6;
141 int i00 = a - b * 7 - c * 7 + 16;
143 for( int y = 0; y < 16; y++ )
146 for( int x = 0; x < 16; x++ )
148 src[x] = x264_clip_pixel( pix>>5 );
157 /****************************************************************************
158 * 8x8 prediction for intra chroma block (4:2:0)
159 ****************************************************************************/
161 static void x264_predict_8x8c_dc_128_c( pixel *src )
163 for( int y = 0; y < 8; y++ )
165 MPIXEL_X4( src+0 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
166 MPIXEL_X4( src+4 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
170 static void x264_predict_8x8c_dc_left_c( pixel *src )
172 int dc0 = 0, dc1 = 0;
174 for( int y = 0; y < 4; y++ )
176 dc0 += src[y * FDEC_STRIDE - 1];
177 dc1 += src[(y+4) * FDEC_STRIDE - 1];
179 pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );
180 pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );
182 for( int y = 0; y < 4; y++ )
184 MPIXEL_X4( src+0 ) = dc0splat;
185 MPIXEL_X4( src+4 ) = dc0splat;
188 for( int y = 0; y < 4; y++ )
190 MPIXEL_X4( src+0 ) = dc1splat;
191 MPIXEL_X4( src+4 ) = dc1splat;
196 static void x264_predict_8x8c_dc_top_c( pixel *src )
198 int dc0 = 0, dc1 = 0;
200 for( int x = 0; x < 4; x++ )
202 dc0 += src[x - FDEC_STRIDE];
203 dc1 += src[x + 4 - FDEC_STRIDE];
205 pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );
206 pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );
208 for( int y = 0; y < 8; y++ )
210 MPIXEL_X4( src+0 ) = dc0splat;
211 MPIXEL_X4( src+4 ) = dc1splat;
215 void x264_predict_8x8c_dc_c( pixel *src )
217 int s0 = 0, s1 = 0, s2 = 0, s3 = 0;
224 for( int i = 0; i < 4; i++ )
226 s0 += src[i - FDEC_STRIDE];
227 s1 += src[i + 4 - FDEC_STRIDE];
228 s2 += src[-1 + i * FDEC_STRIDE];
229 s3 += src[-1 + (i+4)*FDEC_STRIDE];
235 pixel4 dc0 = PIXEL_SPLAT_X4( ( s0 + s2 + 4 ) >> 3 );
236 pixel4 dc1 = PIXEL_SPLAT_X4( ( s1 + 2 ) >> 2 );
237 pixel4 dc2 = PIXEL_SPLAT_X4( ( s3 + 2 ) >> 2 );
238 pixel4 dc3 = PIXEL_SPLAT_X4( ( s1 + s3 + 4 ) >> 3 );
240 for( int y = 0; y < 4; y++ )
242 MPIXEL_X4( src+0 ) = dc0;
243 MPIXEL_X4( src+4 ) = dc1;
247 for( int y = 0; y < 4; y++ )
249 MPIXEL_X4( src+0 ) = dc2;
250 MPIXEL_X4( src+4 ) = dc3;
254 void x264_predict_8x8c_h_c( pixel *src )
256 for( int i = 0; i < 8; i++ )
258 pixel4 v = PIXEL_SPLAT_X4( src[-1] );
259 MPIXEL_X4( src+0 ) = v;
260 MPIXEL_X4( src+4 ) = v;
264 void x264_predict_8x8c_v_c( pixel *src )
266 pixel4 v0 = MPIXEL_X4( src+0-FDEC_STRIDE );
267 pixel4 v1 = MPIXEL_X4( src+4-FDEC_STRIDE );
269 for( int i = 0; i < 8; i++ )
271 MPIXEL_X4( src+0 ) = v0;
272 MPIXEL_X4( src+4 ) = v1;
276 void x264_predict_8x8c_p_c( pixel *src )
280 for( int i = 0; i < 4; i++ )
282 H += ( i + 1 ) * ( src[4+i - FDEC_STRIDE] - src[2 - i -FDEC_STRIDE] );
283 V += ( i + 1 ) * ( src[-1 +(i+4)*FDEC_STRIDE] - src[-1+(2-i)*FDEC_STRIDE] );
286 int a = 16 * ( src[-1+7*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );
287 int b = ( 17 * H + 16 ) >> 5;
288 int c = ( 17 * V + 16 ) >> 5;
289 int i00 = a -3*b -3*c + 16;
291 for( int y = 0; y < 8; y++ )
294 for( int x = 0; x < 8; x++ )
296 src[x] = x264_clip_pixel( pix>>5 );
304 /****************************************************************************
305 * 8x16 prediction for intra chroma block (4:2:2)
306 ****************************************************************************/
308 static void x264_predict_8x16c_dc_128_c( pixel *src )
310 for( int y = 0; y < 16; y++ )
312 MPIXEL_X4( src+0 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
313 MPIXEL_X4( src+4 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );
317 static void x264_predict_8x16c_dc_left_c( pixel *src )
319 for( int i = 0; i < 4; i++ )
323 for( int y = 0; y < 4; y++ )
324 dc += src[y*FDEC_STRIDE - 1];
326 pixel4 dcsplat = PIXEL_SPLAT_X4( (dc + 2) >> 2 );
328 for( int y = 0; y < 4; y++ )
330 MPIXEL_X4( src+0 ) = dcsplat;
331 MPIXEL_X4( src+4 ) = dcsplat;
336 static void x264_predict_8x16c_dc_top_c( pixel *src )
338 int dc0 = 0, dc1 = 0;
340 for(int x = 0; x < 4; x++ )
342 dc0 += src[x - FDEC_STRIDE];
343 dc1 += src[x + 4 - FDEC_STRIDE];
345 pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );
346 pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );
348 for( int y = 0; y < 16; y++ )
350 MPIXEL_X4( src+0 ) = dc0splat;
351 MPIXEL_X4( src+4 ) = dc1splat;
355 void x264_predict_8x16c_dc_c( pixel *src )
357 int s0 = 0, s1 = 0, s2 = 0, s3 = 0, s4 = 0, s5 = 0;
366 for( int i = 0; i < 4; i++ )
368 s0 += src[i+0 - FDEC_STRIDE];
369 s1 += src[i+4 - FDEC_STRIDE];
370 s2 += src[-1 + (i+0) * FDEC_STRIDE];
371 s3 += src[-1 + (i+4) * FDEC_STRIDE];
372 s4 += src[-1 + (i+8) * FDEC_STRIDE];
373 s5 += src[-1 + (i+12) * FDEC_STRIDE];
381 pixel4 dc0 = PIXEL_SPLAT_X4( ( s0 + s2 + 4 ) >> 3 );
382 pixel4 dc1 = PIXEL_SPLAT_X4( ( s1 + 2 ) >> 2 );
383 pixel4 dc2 = PIXEL_SPLAT_X4( ( s3 + 2 ) >> 2 );
384 pixel4 dc3 = PIXEL_SPLAT_X4( ( s1 + s3 + 4 ) >> 3 );
385 pixel4 dc4 = PIXEL_SPLAT_X4( ( s4 + 2 ) >> 2 );
386 pixel4 dc5 = PIXEL_SPLAT_X4( ( s1 + s4 + 4 ) >> 3 );
387 pixel4 dc6 = PIXEL_SPLAT_X4( ( s5 + 2 ) >> 2 );
388 pixel4 dc7 = PIXEL_SPLAT_X4( ( s1 + s5 + 4 ) >> 3 );
390 for( int y = 0; y < 4; y++ )
392 MPIXEL_X4( src+0 ) = dc0;
393 MPIXEL_X4( src+4 ) = dc1;
396 for( int y = 0; y < 4; y++ )
398 MPIXEL_X4( src+0 ) = dc2;
399 MPIXEL_X4( src+4 ) = dc3;
402 for( int y = 0; y < 4; y++ )
404 MPIXEL_X4( src+0 ) = dc4;
405 MPIXEL_X4( src+4 ) = dc5;
408 for( int y = 0; y < 4; y++ )
410 MPIXEL_X4( src+0 ) = dc6;
411 MPIXEL_X4( src+4 ) = dc7;
415 void x264_predict_8x16c_h_c( pixel *src )
417 for( int i = 0; i < 16; i++ )
419 pixel4 v = PIXEL_SPLAT_X4( src[-1] );
420 MPIXEL_X4( src+0 ) = v;
421 MPIXEL_X4( src+4 ) = v;
425 void x264_predict_8x16c_v_c( pixel *src )
427 pixel4 v0 = MPIXEL_X4( src+0-FDEC_STRIDE );
428 pixel4 v1 = MPIXEL_X4( src+4-FDEC_STRIDE );
430 for( int i = 0; i < 16; i++ )
432 MPIXEL_X4( src+0 ) = v0;
433 MPIXEL_X4( src+4 ) = v1;
437 void x264_predict_8x16c_p_c( pixel *src )
442 for( int i = 0; i < 4; i++ )
443 H += ( i + 1 ) * ( src[4 + i - FDEC_STRIDE] - src[2 - i - FDEC_STRIDE] );
444 for( int i = 0; i < 8; i++ )
445 V += ( i + 1 ) * ( src[-1 + (i+8)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );
447 int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );
448 int b = ( 17 * H + 16 ) >> 5;
449 int c = ( 5 * V + 32 ) >> 6;
450 int i00 = a -3*b -7*c + 16;
452 for( int y = 0; y < 16; y++ )
455 for( int x = 0; x < 8; x++ )
457 src[x] = x264_clip_pixel( pix>>5 );
465 /****************************************************************************
466 * 4x4 prediction for intra luma block
467 ****************************************************************************/
469 #define SRC(x,y) src[(x)+(y)*FDEC_STRIDE]
470 #define SRC_X4(x,y) MPIXEL_X4( &SRC(x,y) )
472 #define PREDICT_4x4_DC(v)\
473 SRC_X4(0,0) = SRC_X4(0,1) = SRC_X4(0,2) = SRC_X4(0,3) = v;
475 static void x264_predict_4x4_dc_128_c( pixel *src )
477 PREDICT_4x4_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );
479 static void x264_predict_4x4_dc_left_c( pixel *src )
481 pixel4 dc = PIXEL_SPLAT_X4( (SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) + 2) >> 2 );
482 PREDICT_4x4_DC( dc );
484 static void x264_predict_4x4_dc_top_c( pixel *src )
486 pixel4 dc = PIXEL_SPLAT_X4( (SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 2) >> 2 );
487 PREDICT_4x4_DC( dc );
489 void x264_predict_4x4_dc_c( pixel *src )
491 pixel4 dc = PIXEL_SPLAT_X4( (SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) +
492 SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 4) >> 3 );
493 PREDICT_4x4_DC( dc );
495 void x264_predict_4x4_h_c( pixel *src )
497 SRC_X4(0,0) = PIXEL_SPLAT_X4( SRC(-1,0) );
498 SRC_X4(0,1) = PIXEL_SPLAT_X4( SRC(-1,1) );
499 SRC_X4(0,2) = PIXEL_SPLAT_X4( SRC(-1,2) );
500 SRC_X4(0,3) = PIXEL_SPLAT_X4( SRC(-1,3) );
502 void x264_predict_4x4_v_c( pixel *src )
504 PREDICT_4x4_DC(SRC_X4(0,-1));
507 #define PREDICT_4x4_LOAD_LEFT\
511 UNUSED int l3 = SRC(-1,3);
513 #define PREDICT_4x4_LOAD_TOP\
517 UNUSED int t3 = SRC(3,-1);
519 #define PREDICT_4x4_LOAD_TOP_RIGHT\
523 UNUSED int t7 = SRC(7,-1);
525 #define F1(a,b) (((a)+(b)+1)>>1)
526 #define F2(a,b,c) (((a)+2*(b)+(c)+2)>>2)
528 static void x264_predict_4x4_ddl_c( pixel *src )
531 PREDICT_4x4_LOAD_TOP_RIGHT
532 SRC(0,0)= F2(t0,t1,t2);
533 SRC(1,0)=SRC(0,1)= F2(t1,t2,t3);
534 SRC(2,0)=SRC(1,1)=SRC(0,2)= F2(t2,t3,t4);
535 SRC(3,0)=SRC(2,1)=SRC(1,2)=SRC(0,3)= F2(t3,t4,t5);
536 SRC(3,1)=SRC(2,2)=SRC(1,3)= F2(t4,t5,t6);
537 SRC(3,2)=SRC(2,3)= F2(t5,t6,t7);
538 SRC(3,3)= F2(t6,t7,t7);
540 static void x264_predict_4x4_ddr_c( pixel *src )
543 PREDICT_4x4_LOAD_LEFT
545 SRC(3,0)= F2(t3,t2,t1);
546 SRC(2,0)=SRC(3,1)= F2(t2,t1,t0);
547 SRC(1,0)=SRC(2,1)=SRC(3,2)= F2(t1,t0,lt);
548 SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)= F2(t0,lt,l0);
549 SRC(0,1)=SRC(1,2)=SRC(2,3)= F2(lt,l0,l1);
550 SRC(0,2)=SRC(1,3)= F2(l0,l1,l2);
551 SRC(0,3)= F2(l1,l2,l3);
554 static void x264_predict_4x4_vr_c( pixel *src )
557 PREDICT_4x4_LOAD_LEFT
559 SRC(0,3)= F2(l2,l1,l0);
560 SRC(0,2)= F2(l1,l0,lt);
561 SRC(0,1)=SRC(1,3)= F2(l0,lt,t0);
562 SRC(0,0)=SRC(1,2)= F1(lt,t0);
563 SRC(1,1)=SRC(2,3)= F2(lt,t0,t1);
564 SRC(1,0)=SRC(2,2)= F1(t0,t1);
565 SRC(2,1)=SRC(3,3)= F2(t0,t1,t2);
566 SRC(2,0)=SRC(3,2)= F1(t1,t2);
567 SRC(3,1)= F2(t1,t2,t3);
571 static void x264_predict_4x4_hd_c( pixel *src )
574 PREDICT_4x4_LOAD_LEFT
577 SRC(1,3)= F2(l1,l2,l3);
578 SRC(0,2)=SRC(2,3)= F1(l1,l2);
579 SRC(1,2)=SRC(3,3)= F2(l0,l1,l2);
580 SRC(0,1)=SRC(2,2)= F1(l0,l1);
581 SRC(1,1)=SRC(3,2)= F2(lt,l0,l1);
582 SRC(0,0)=SRC(2,1)= F1(lt,l0);
583 SRC(1,0)=SRC(3,1)= F2(t0,lt,l0);
584 SRC(2,0)= F2(t1,t0,lt);
585 SRC(3,0)= F2(t2,t1,t0);
588 static void x264_predict_4x4_vl_c( pixel *src )
591 PREDICT_4x4_LOAD_TOP_RIGHT
593 SRC(0,1)= F2(t0,t1,t2);
594 SRC(1,0)=SRC(0,2)= F1(t1,t2);
595 SRC(1,1)=SRC(0,3)= F2(t1,t2,t3);
596 SRC(2,0)=SRC(1,2)= F1(t2,t3);
597 SRC(2,1)=SRC(1,3)= F2(t2,t3,t4);
598 SRC(3,0)=SRC(2,2)= F1(t3,t4);
599 SRC(3,1)=SRC(2,3)= F2(t3,t4,t5);
601 SRC(3,3)= F2(t4,t5,t6);
604 static void x264_predict_4x4_hu_c( pixel *src )
606 PREDICT_4x4_LOAD_LEFT
608 SRC(1,0)= F2(l0,l1,l2);
609 SRC(2,0)=SRC(0,1)= F1(l1,l2);
610 SRC(3,0)=SRC(1,1)= F2(l1,l2,l3);
611 SRC(2,1)=SRC(0,2)= F1(l2,l3);
612 SRC(3,1)=SRC(1,2)= F2(l2,l3,l3);
613 SRC(3,2)=SRC(1,3)=SRC(0,3)=
614 SRC(2,2)=SRC(2,3)=SRC(3,3)= l3;
617 /****************************************************************************
618 * 8x8 prediction for intra luma block
619 ****************************************************************************/
622 edge[14-y] = F2(SRC(-1,y-1), SRC(-1,y), SRC(-1,y+1));
624 edge[16+x] = F2(SRC(x-1,-1), SRC(x,-1), SRC(x+1,-1));
626 static void x264_predict_8x8_filter_c( pixel *src, pixel edge[36], int i_neighbor, int i_filters )
628 /* edge[7..14] = l7..l0
630 * edge[16..31] = t0 .. t15
633 int have_lt = i_neighbor & MB_TOPLEFT;
634 if( i_filters & MB_LEFT )
636 edge[15] = (SRC(0,-1) + 2*SRC(-1,-1) + SRC(-1,0) + 2) >> 2;
637 edge[14] = ((have_lt ? SRC(-1,-1) : SRC(-1,0))
638 + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2;
639 PL(1) PL(2) PL(3) PL(4) PL(5) PL(6)
641 edge[7] = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2;
644 if( i_filters & MB_TOP )
646 int have_tr = i_neighbor & MB_TOPRIGHT;
647 edge[16] = ((have_lt ? SRC(-1,-1) : SRC(0,-1))
648 + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2;
649 PT(1) PT(2) PT(3) PT(4) PT(5) PT(6)
650 edge[23] = (SRC(6,-1) + 2*SRC(7,-1)
651 + (have_tr ? SRC(8,-1) : SRC(7,-1)) + 2) >> 2;
653 if( i_filters & MB_TOPRIGHT )
657 PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14)
659 edge[32] = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2;
663 MPIXEL_X4( edge+24 ) = PIXEL_SPLAT_X4( SRC(7,-1) );
664 MPIXEL_X4( edge+28 ) = PIXEL_SPLAT_X4( SRC(7,-1) );
665 edge[32] = SRC(7,-1);
675 UNUSED int l##y = edge[14-y];
677 UNUSED int t##x = edge[16+x];
678 #define PREDICT_8x8_LOAD_TOPLEFT \
680 #define PREDICT_8x8_LOAD_LEFT \
681 PL(0) PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) PL(7)
682 #define PREDICT_8x8_LOAD_TOP \
683 PT(0) PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) PT(7)
684 #define PREDICT_8x8_LOAD_TOPRIGHT \
685 PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14) PT(15)
687 #define PREDICT_8x8_DC(v) \
688 for( int y = 0; y < 8; y++ ) { \
689 MPIXEL_X4( src+0 ) = v; \
690 MPIXEL_X4( src+4 ) = v; \
691 src += FDEC_STRIDE; \
694 static void x264_predict_8x8_dc_128_c( pixel *src, pixel edge[36] )
696 PREDICT_8x8_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );
698 static void x264_predict_8x8_dc_left_c( pixel *src, pixel edge[36] )
700 PREDICT_8x8_LOAD_LEFT
701 pixel4 dc = PIXEL_SPLAT_X4( (l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3 );
702 PREDICT_8x8_DC( dc );
704 static void x264_predict_8x8_dc_top_c( pixel *src, pixel edge[36] )
707 pixel4 dc = PIXEL_SPLAT_X4( (t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3 );
708 PREDICT_8x8_DC( dc );
710 void x264_predict_8x8_dc_c( pixel *src, pixel edge[36] )
712 PREDICT_8x8_LOAD_LEFT
714 pixel4 dc = PIXEL_SPLAT_X4( (l0+l1+l2+l3+l4+l5+l6+l7+t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4 );
715 PREDICT_8x8_DC( dc );
717 void x264_predict_8x8_h_c( pixel *src, pixel edge[36] )
719 PREDICT_8x8_LOAD_LEFT
720 #define ROW(y) MPIXEL_X4( src+y*FDEC_STRIDE+0 ) =\
721 MPIXEL_X4( src+y*FDEC_STRIDE+4 ) = PIXEL_SPLAT_X4( l##y );
722 ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
725 void x264_predict_8x8_v_c( pixel *src, pixel edge[36] )
727 pixel4 top[2] = { MPIXEL_X4( edge+16 ),
728 MPIXEL_X4( edge+20 ) };
729 for( int y = 0; y < 8; y++ )
731 MPIXEL_X4( src+y*FDEC_STRIDE+0 ) = top[0];
732 MPIXEL_X4( src+y*FDEC_STRIDE+4 ) = top[1];
735 static void x264_predict_8x8_ddl_c( pixel *src, pixel edge[36] )
738 PREDICT_8x8_LOAD_TOPRIGHT
739 SRC(0,0)= F2(t0,t1,t2);
740 SRC(0,1)=SRC(1,0)= F2(t1,t2,t3);
741 SRC(0,2)=SRC(1,1)=SRC(2,0)= F2(t2,t3,t4);
742 SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= F2(t3,t4,t5);
743 SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= F2(t4,t5,t6);
744 SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= F2(t5,t6,t7);
745 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);
746 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);
747 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);
748 SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= F2(t9,t10,t11);
749 SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= F2(t10,t11,t12);
750 SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= F2(t11,t12,t13);
751 SRC(5,7)=SRC(6,6)=SRC(7,5)= F2(t12,t13,t14);
752 SRC(6,7)=SRC(7,6)= F2(t13,t14,t15);
753 SRC(7,7)= F2(t14,t15,t15);
755 static void x264_predict_8x8_ddr_c( pixel *src, pixel edge[36] )
758 PREDICT_8x8_LOAD_LEFT
759 PREDICT_8x8_LOAD_TOPLEFT
760 SRC(0,7)= F2(l7,l6,l5);
761 SRC(0,6)=SRC(1,7)= F2(l6,l5,l4);
762 SRC(0,5)=SRC(1,6)=SRC(2,7)= F2(l5,l4,l3);
763 SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= F2(l4,l3,l2);
764 SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= F2(l3,l2,l1);
765 SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= F2(l2,l1,l0);
766 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);
767 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);
768 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);
769 SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= F2(t0,t1,t2);
770 SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= F2(t1,t2,t3);
771 SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= F2(t2,t3,t4);
772 SRC(5,0)=SRC(6,1)=SRC(7,2)= F2(t3,t4,t5);
773 SRC(6,0)=SRC(7,1)= F2(t4,t5,t6);
774 SRC(7,0)= F2(t5,t6,t7);
777 static void x264_predict_8x8_vr_c( pixel *src, pixel edge[36] )
780 PREDICT_8x8_LOAD_LEFT
781 PREDICT_8x8_LOAD_TOPLEFT
782 SRC(0,6)= F2(l5,l4,l3);
783 SRC(0,7)= F2(l6,l5,l4);
784 SRC(0,4)=SRC(1,6)= F2(l3,l2,l1);
785 SRC(0,5)=SRC(1,7)= F2(l4,l3,l2);
786 SRC(0,2)=SRC(1,4)=SRC(2,6)= F2(l1,l0,lt);
787 SRC(0,3)=SRC(1,5)=SRC(2,7)= F2(l2,l1,l0);
788 SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= F2(l0,lt,t0);
789 SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= F1(lt,t0);
790 SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= F2(lt,t0,t1);
791 SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= F1(t0,t1);
792 SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= F2(t0,t1,t2);
793 SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= F1(t1,t2);
794 SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= F2(t1,t2,t3);
795 SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= F1(t2,t3);
796 SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= F2(t2,t3,t4);
797 SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= F1(t3,t4);
798 SRC(5,1)=SRC(6,3)=SRC(7,5)= F2(t3,t4,t5);
799 SRC(5,0)=SRC(6,2)=SRC(7,4)= F1(t4,t5);
800 SRC(6,1)=SRC(7,3)= F2(t4,t5,t6);
801 SRC(6,0)=SRC(7,2)= F1(t5,t6);
802 SRC(7,1)= F2(t5,t6,t7);
805 static void x264_predict_8x8_hd_c( pixel *src, pixel edge[36] )
808 PREDICT_8x8_LOAD_LEFT
809 PREDICT_8x8_LOAD_TOPLEFT
810 int p1 = pack_pixel_1to2(F1(l6,l7), F2(l5,l6,l7));
811 int p2 = pack_pixel_1to2(F1(l5,l6), F2(l4,l5,l6));
812 int p3 = pack_pixel_1to2(F1(l4,l5), F2(l3,l4,l5));
813 int p4 = pack_pixel_1to2(F1(l3,l4), F2(l2,l3,l4));
814 int p5 = pack_pixel_1to2(F1(l2,l3), F2(l1,l2,l3));
815 int p6 = pack_pixel_1to2(F1(l1,l2), F2(l0,l1,l2));
816 int p7 = pack_pixel_1to2(F1(l0,l1), F2(lt,l0,l1));
817 int p8 = pack_pixel_1to2(F1(lt,l0), F2(l0,lt,t0));
818 int p9 = pack_pixel_1to2(F2(t1,t0,lt), F2(t2,t1,t0));
819 int p10 = pack_pixel_1to2(F2(t3,t2,t1), F2(t4,t3,t2));
820 int p11 = pack_pixel_1to2(F2(t5,t4,t3), F2(t6,t5,t4));
821 SRC_X4(0,7)= pack_pixel_2to4(p1,p2);
822 SRC_X4(0,6)= pack_pixel_2to4(p2,p3);
823 SRC_X4(4,7)=SRC_X4(0,5)= pack_pixel_2to4(p3,p4);
824 SRC_X4(4,6)=SRC_X4(0,4)= pack_pixel_2to4(p4,p5);
825 SRC_X4(4,5)=SRC_X4(0,3)= pack_pixel_2to4(p5,p6);
826 SRC_X4(4,4)=SRC_X4(0,2)= pack_pixel_2to4(p6,p7);
827 SRC_X4(4,3)=SRC_X4(0,1)= pack_pixel_2to4(p7,p8);
828 SRC_X4(4,2)=SRC_X4(0,0)= pack_pixel_2to4(p8,p9);
829 SRC_X4(4,1)= pack_pixel_2to4(p9,p10);
830 SRC_X4(4,0)= pack_pixel_2to4(p10,p11);
832 static void x264_predict_8x8_vl_c( pixel *src, pixel edge[36] )
835 PREDICT_8x8_LOAD_TOPRIGHT
837 SRC(0,1)= F2(t0,t1,t2);
838 SRC(0,2)=SRC(1,0)= F1(t1,t2);
839 SRC(0,3)=SRC(1,1)= F2(t1,t2,t3);
840 SRC(0,4)=SRC(1,2)=SRC(2,0)= F1(t2,t3);
841 SRC(0,5)=SRC(1,3)=SRC(2,1)= F2(t2,t3,t4);
842 SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= F1(t3,t4);
843 SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= F2(t3,t4,t5);
844 SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= F1(t4,t5);
845 SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= F2(t4,t5,t6);
846 SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= F1(t5,t6);
847 SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= F2(t5,t6,t7);
848 SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= F1(t6,t7);
849 SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= F2(t6,t7,t8);
850 SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= F1(t7,t8);
851 SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= F2(t7,t8,t9);
852 SRC(5,6)=SRC(6,4)=SRC(7,2)= F1(t8,t9);
853 SRC(5,7)=SRC(6,5)=SRC(7,3)= F2(t8,t9,t10);
854 SRC(6,6)=SRC(7,4)= F1(t9,t10);
855 SRC(6,7)=SRC(7,5)= F2(t9,t10,t11);
856 SRC(7,6)= F1(t10,t11);
857 SRC(7,7)= F2(t10,t11,t12);
859 static void x264_predict_8x8_hu_c( pixel *src, pixel edge[36] )
861 PREDICT_8x8_LOAD_LEFT
862 int p1 = pack_pixel_1to2(F1(l0,l1), F2(l0,l1,l2));
863 int p2 = pack_pixel_1to2(F1(l1,l2), F2(l1,l2,l3));
864 int p3 = pack_pixel_1to2(F1(l2,l3), F2(l2,l3,l4));
865 int p4 = pack_pixel_1to2(F1(l3,l4), F2(l3,l4,l5));
866 int p5 = pack_pixel_1to2(F1(l4,l5), F2(l4,l5,l6));
867 int p6 = pack_pixel_1to2(F1(l5,l6), F2(l5,l6,l7));
868 int p7 = pack_pixel_1to2(F1(l6,l7), F2(l6,l7,l7));
869 int p8 = pack_pixel_1to2(l7,l7);
870 SRC_X4(0,0)= pack_pixel_2to4(p1,p2);
871 SRC_X4(0,1)= pack_pixel_2to4(p2,p3);
872 SRC_X4(4,0)=SRC_X4(0,2)= pack_pixel_2to4(p3,p4);
873 SRC_X4(4,1)=SRC_X4(0,3)= pack_pixel_2to4(p4,p5);
874 SRC_X4(4,2)=SRC_X4(0,4)= pack_pixel_2to4(p5,p6);
875 SRC_X4(4,3)=SRC_X4(0,5)= pack_pixel_2to4(p6,p7);
876 SRC_X4(4,4)=SRC_X4(0,6)= pack_pixel_2to4(p7,p8);
877 SRC_X4(4,5)=SRC_X4(4,6)= SRC_X4(0,7) = SRC_X4(4,7) = pack_pixel_2to4(p8,p8);
880 /****************************************************************************
881 * Exported functions:
882 ****************************************************************************/
883 void x264_predict_16x16_init( int cpu, x264_predict_t pf[7] )
885 pf[I_PRED_16x16_V ] = x264_predict_16x16_v_c;
886 pf[I_PRED_16x16_H ] = x264_predict_16x16_h_c;
887 pf[I_PRED_16x16_DC] = x264_predict_16x16_dc_c;
888 pf[I_PRED_16x16_P ] = x264_predict_16x16_p_c;
889 pf[I_PRED_16x16_DC_LEFT]= x264_predict_16x16_dc_left_c;
890 pf[I_PRED_16x16_DC_TOP ]= x264_predict_16x16_dc_top_c;
891 pf[I_PRED_16x16_DC_128 ]= x264_predict_16x16_dc_128_c;
894 x264_predict_16x16_init_mmx( cpu, pf );
898 if( cpu&X264_CPU_ALTIVEC )
899 x264_predict_16x16_init_altivec( pf );
903 x264_predict_16x16_init_arm( cpu, pf );
907 x264_predict_16x16_init_aarch64( cpu, pf );
911 void x264_predict_8x8c_init( int cpu, x264_predict_t pf[7] )
913 pf[I_PRED_CHROMA_V ] = x264_predict_8x8c_v_c;
914 pf[I_PRED_CHROMA_H ] = x264_predict_8x8c_h_c;
915 pf[I_PRED_CHROMA_DC] = x264_predict_8x8c_dc_c;
916 pf[I_PRED_CHROMA_P ] = x264_predict_8x8c_p_c;
917 pf[I_PRED_CHROMA_DC_LEFT]= x264_predict_8x8c_dc_left_c;
918 pf[I_PRED_CHROMA_DC_TOP ]= x264_predict_8x8c_dc_top_c;
919 pf[I_PRED_CHROMA_DC_128 ]= x264_predict_8x8c_dc_128_c;
922 x264_predict_8x8c_init_mmx( cpu, pf );
926 if( cpu&X264_CPU_ALTIVEC )
927 x264_predict_8x8c_init_altivec( pf );
931 x264_predict_8x8c_init_arm( cpu, pf );
935 x264_predict_8x8c_init_aarch64( cpu, pf );
939 void x264_predict_8x16c_init( int cpu, x264_predict_t pf[7] )
941 pf[I_PRED_CHROMA_V ] = x264_predict_8x16c_v_c;
942 pf[I_PRED_CHROMA_H ] = x264_predict_8x16c_h_c;
943 pf[I_PRED_CHROMA_DC] = x264_predict_8x16c_dc_c;
944 pf[I_PRED_CHROMA_P ] = x264_predict_8x16c_p_c;
945 pf[I_PRED_CHROMA_DC_LEFT]= x264_predict_8x16c_dc_left_c;
946 pf[I_PRED_CHROMA_DC_TOP ]= x264_predict_8x16c_dc_top_c;
947 pf[I_PRED_CHROMA_DC_128 ]= x264_predict_8x16c_dc_128_c;
950 x264_predict_8x16c_init_mmx( cpu, pf );
954 x264_predict_8x16c_init_aarch64( cpu, pf );
958 void x264_predict_8x8_init( int cpu, x264_predict8x8_t pf[12], x264_predict_8x8_filter_t *predict_filter )
960 pf[I_PRED_8x8_V] = x264_predict_8x8_v_c;
961 pf[I_PRED_8x8_H] = x264_predict_8x8_h_c;
962 pf[I_PRED_8x8_DC] = x264_predict_8x8_dc_c;
963 pf[I_PRED_8x8_DDL] = x264_predict_8x8_ddl_c;
964 pf[I_PRED_8x8_DDR] = x264_predict_8x8_ddr_c;
965 pf[I_PRED_8x8_VR] = x264_predict_8x8_vr_c;
966 pf[I_PRED_8x8_HD] = x264_predict_8x8_hd_c;
967 pf[I_PRED_8x8_VL] = x264_predict_8x8_vl_c;
968 pf[I_PRED_8x8_HU] = x264_predict_8x8_hu_c;
969 pf[I_PRED_8x8_DC_LEFT]= x264_predict_8x8_dc_left_c;
970 pf[I_PRED_8x8_DC_TOP] = x264_predict_8x8_dc_top_c;
971 pf[I_PRED_8x8_DC_128] = x264_predict_8x8_dc_128_c;
972 *predict_filter = x264_predict_8x8_filter_c;
975 x264_predict_8x8_init_mmx( cpu, pf, predict_filter );
979 x264_predict_8x8_init_arm( cpu, pf, predict_filter );
983 x264_predict_8x8_init_aarch64( cpu, pf, predict_filter );
987 void x264_predict_4x4_init( int cpu, x264_predict_t pf[12] )
989 pf[I_PRED_4x4_V] = x264_predict_4x4_v_c;
990 pf[I_PRED_4x4_H] = x264_predict_4x4_h_c;
991 pf[I_PRED_4x4_DC] = x264_predict_4x4_dc_c;
992 pf[I_PRED_4x4_DDL] = x264_predict_4x4_ddl_c;
993 pf[I_PRED_4x4_DDR] = x264_predict_4x4_ddr_c;
994 pf[I_PRED_4x4_VR] = x264_predict_4x4_vr_c;
995 pf[I_PRED_4x4_HD] = x264_predict_4x4_hd_c;
996 pf[I_PRED_4x4_VL] = x264_predict_4x4_vl_c;
997 pf[I_PRED_4x4_HU] = x264_predict_4x4_hu_c;
998 pf[I_PRED_4x4_DC_LEFT]= x264_predict_4x4_dc_left_c;
999 pf[I_PRED_4x4_DC_TOP] = x264_predict_4x4_dc_top_c;
1000 pf[I_PRED_4x4_DC_128] = x264_predict_4x4_dc_128_c;
1003 x264_predict_4x4_init_mmx( cpu, pf );
1007 x264_predict_4x4_init_arm( cpu, pf );
1011 x264_predict_4x4_init_aarch64( cpu, pf );