falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
28457 Commits
32 Branches
759MB
Tree: 44ca80df34
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '44ca80df34'
${ noResults }
FFmpeg/libavcodec/h264pred_template.c

990 lines
33KB
Raw Blame History 

  1. /*

  2.  * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder

  3.  * Copyright (c) 2003-2011 Michael Niedermayer <michaelni@gmx.at>

  4.  *

  5.  * This file is part of Libav.

  6.  *

  7.  * Libav 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.  * Libav 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 Libav; 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 "mathops.h"

  29. #include "dsputil.h"

  30. 

  31. #define BIT_DEPTH 8

  32. 

  33. #define pixel uint8_t

  34. #define pixel4 uint32_t

  35. #define dctcoef DCTELEM

  36. 

  37. #define INIT_CLIP uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;

  38. #define CLIP(a) cm[a]

  39. #define FUNC(a) a

  40. #define FUNCC(a) a ## _c

  41. #define PIXEL_SPLAT_X4(x) ((x)*0x01010101U)

  42. #define AV_WN4P  AV_WN32

  43. #define AV_WN4PA AV_WN32A

  44. 

  45. static void FUNCC(pred4x4_vertical)(uint8_t *_src, const uint8_t *topright, int _stride){

  46.     pixel *src = (pixel*)_src;

  47.     int stride = _stride/sizeof(pixel);

  48.     const pixel4 a= ((pixel4*)(src-stride))[0];

  49.     ((pixel4*)(src+0*stride))[0]= a;

  50.     ((pixel4*)(src+1*stride))[0]= a;

  51.     ((pixel4*)(src+2*stride))[0]= a;

  52.     ((pixel4*)(src+3*stride))[0]= a;

  53. }

  54. 

  55. static void FUNCC(pred4x4_horizontal)(uint8_t *_src, const uint8_t *topright, int _stride){

  56.     pixel *src = (pixel*)_src;

  57.     int stride = _stride/sizeof(pixel);

  58.     ((pixel4*)(src+0*stride))[0]= PIXEL_SPLAT_X4(src[-1+0*stride]);

  59.     ((pixel4*)(src+1*stride))[0]= PIXEL_SPLAT_X4(src[-1+1*stride]);

  60.     ((pixel4*)(src+2*stride))[0]= PIXEL_SPLAT_X4(src[-1+2*stride]);

  61.     ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(src[-1+3*stride]);

  62. }

  63. 

  64. static void FUNCC(pred4x4_dc)(uint8_t *_src, const uint8_t *topright, int _stride){

  65.     pixel *src = (pixel*)_src;

  66.     int stride = _stride/sizeof(pixel);

  67.     const int dc= (  src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]

  68.                    + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;

  69. 

  70.     ((pixel4*)(src+0*stride))[0]=

  71.     ((pixel4*)(src+1*stride))[0]=

  72.     ((pixel4*)(src+2*stride))[0]=

  73.     ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(dc);

  74. }

  75. 

  76. static void FUNCC(pred4x4_left_dc)(uint8_t *_src, const uint8_t *topright, int _stride){

  77.     pixel *src = (pixel*)_src;

  78.     int stride = _stride/sizeof(pixel);

  79.     const int dc= (  src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;

  80. 

  81.     ((pixel4*)(src+0*stride))[0]=

  82.     ((pixel4*)(src+1*stride))[0]=

  83.     ((pixel4*)(src+2*stride))[0]=

  84.     ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(dc);

  85. }

  86. 

  87. static void FUNCC(pred4x4_top_dc)(uint8_t *_src, const uint8_t *topright, int _stride){

  88.     pixel *src = (pixel*)_src;

  89.     int stride = _stride/sizeof(pixel);

  90.     const int dc= (  src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;

  91. 

  92.     ((pixel4*)(src+0*stride))[0]=

  93.     ((pixel4*)(src+1*stride))[0]=

  94.     ((pixel4*)(src+2*stride))[0]=

  95.     ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(dc);

  96. }

  97. 

  98. static void FUNCC(pred4x4_128_dc)(uint8_t *_src, const uint8_t *topright, int _stride){

  99.     pixel *src = (pixel*)_src;

  100.     int stride = _stride/sizeof(pixel);

  101.     ((pixel4*)(src+0*stride))[0]=

  102.     ((pixel4*)(src+1*stride))[0]=

  103.     ((pixel4*)(src+2*stride))[0]=

  104.     ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1));

  105. }

  106. 

  107. static void FUNCC(pred4x4_127_dc)(uint8_t *_src, const uint8_t *topright, int _stride){

  108.     pixel *src = (pixel*)_src;

  109.     int stride = _stride/sizeof(pixel);

  110.     ((pixel4*)(src+0*stride))[0]=

  111.     ((pixel4*)(src+1*stride))[0]=

  112.     ((pixel4*)(src+2*stride))[0]=

  113.     ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))-1);

  114. }

  115. 

  116. static void FUNCC(pred4x4_129_dc)(uint8_t *_src, const uint8_t *topright, int _stride){

  117.     pixel *src = (pixel*)_src;

  118.     int stride = _stride/sizeof(pixel);

  119.     ((pixel4*)(src+0*stride))[0]=

  120.     ((pixel4*)(src+1*stride))[0]=

  121.     ((pixel4*)(src+2*stride))[0]=

  122.     ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))+1);

  123. }

  124. 

  125. 

  126. #define LOAD_TOP_RIGHT_EDGE\

  127.     const int av_unused t4= topright[0];\

  128.     const int av_unused t5= topright[1];\

  129.     const int av_unused t6= topright[2];\

  130.     const int av_unused t7= topright[3];\

  131. 

  132. #define LOAD_DOWN_LEFT_EDGE\

  133.     const int av_unused l4= src[-1+4*stride];\

  134.     const int av_unused l5= src[-1+5*stride];\

  135.     const int av_unused l6= src[-1+6*stride];\

  136.     const int av_unused l7= src[-1+7*stride];\

  137. 

  138. #define LOAD_LEFT_EDGE\

  139.     const int av_unused l0= src[-1+0*stride];\

  140.     const int av_unused l1= src[-1+1*stride];\

  141.     const int av_unused l2= src[-1+2*stride];\

  142.     const int av_unused l3= src[-1+3*stride];\

  143. 

  144. #define LOAD_TOP_EDGE\

  145.     const int av_unused t0= src[ 0-1*stride];\

  146.     const int av_unused t1= src[ 1-1*stride];\

  147.     const int av_unused t2= src[ 2-1*stride];\

  148.     const int av_unused t3= src[ 3-1*stride];\

  149. 

  150. static void FUNCC(pred4x4_down_right)(uint8_t *_src, const uint8_t *topright, int _stride){

  151.     pixel *src = (pixel*)_src;

  152.     int stride = _stride/sizeof(pixel);

  153.     const int lt= src[-1-1*stride];

  154.     LOAD_TOP_EDGE

  155.     LOAD_LEFT_EDGE

  156. 

  157.     src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2;

  158.     src[0+2*stride]=

  159.     src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2;

  160.     src[0+1*stride]=

  161.     src[1+2*stride]=

  162.     src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2;

  163.     src[0+0*stride]=

  164.     src[1+1*stride]=

  165.     src[2+2*stride]=

  166.     src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2;

  167.     src[1+0*stride]=

  168.     src[2+1*stride]=

  169.     src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2;

  170.     src[2+0*stride]=

  171.     src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;

  172.     src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;

  173. }

  174. 

  175. static void FUNCC(pred4x4_down_left)(uint8_t *_src, const uint8_t *_topright, int _stride){

  176.     pixel *src = (pixel*)_src;

  177.     const pixel *topright = (const pixel*)_topright;

  178.     int stride = _stride/sizeof(pixel);

  179.     LOAD_TOP_EDGE

  180.     LOAD_TOP_RIGHT_EDGE

  181. //    LOAD_LEFT_EDGE

  182. 

  183.     src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2;

  184.     src[1+0*stride]=

  185.     src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2;

  186.     src[2+0*stride]=

  187.     src[1+1*stride]=

  188.     src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2;

  189.     src[3+0*stride]=

  190.     src[2+1*stride]=

  191.     src[1+2*stride]=

  192.     src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2;

  193.     src[3+1*stride]=

  194.     src[2+2*stride]=

  195.     src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2;

  196.     src[3+2*stride]=

  197.     src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2;

  198.     src[3+3*stride]=(t6 + 3*t7 + 2)>>2;

  199. }

  200. 

  201. static void FUNCC(pred4x4_vertical_right)(uint8_t *_src, const uint8_t *topright, int _stride){

  202.     pixel *src = (pixel*)_src;

  203.     int stride = _stride/sizeof(pixel);

  204.     const int lt= src[-1-1*stride];

  205.     LOAD_TOP_EDGE

  206.     LOAD_LEFT_EDGE

  207. 

  208.     src[0+0*stride]=

  209.     src[1+2*stride]=(lt + t0 + 1)>>1;

  210.     src[1+0*stride]=

  211.     src[2+2*stride]=(t0 + t1 + 1)>>1;

  212.     src[2+0*stride]=

  213.     src[3+2*stride]=(t1 + t2 + 1)>>1;

  214.     src[3+0*stride]=(t2 + t3 + 1)>>1;

  215.     src[0+1*stride]=

  216.     src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;

  217.     src[1+1*stride]=

  218.     src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;

  219.     src[2+1*stride]=

  220.     src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;

  221.     src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;

  222.     src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;

  223.     src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;

  224. }

  225. 

  226. static void FUNCC(pred4x4_vertical_left)(uint8_t *_src, const uint8_t *_topright, int _stride){

  227.     pixel *src = (pixel*)_src;

  228.     const pixel *topright = (const pixel*)_topright;

  229.     int stride = _stride/sizeof(pixel);

  230.     LOAD_TOP_EDGE

  231.     LOAD_TOP_RIGHT_EDGE

  232. 

  233.     src[0+0*stride]=(t0 + t1 + 1)>>1;

  234.     src[1+0*stride]=

  235.     src[0+2*stride]=(t1 + t2 + 1)>>1;

  236.     src[2+0*stride]=

  237.     src[1+2*stride]=(t2 + t3 + 1)>>1;

  238.     src[3+0*stride]=

  239.     src[2+2*stride]=(t3 + t4+ 1)>>1;

  240.     src[3+2*stride]=(t4 + t5+ 1)>>1;

  241.     src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;

  242.     src[1+1*stride]=

  243.     src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;

  244.     src[2+1*stride]=

  245.     src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;

  246.     src[3+1*stride]=

  247.     src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;

  248.     src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;

  249. }

  250. 

  251. static void FUNCC(pred4x4_horizontal_up)(uint8_t *_src, const uint8_t *topright, int _stride){

  252.     pixel *src = (pixel*)_src;

  253.     int stride = _stride/sizeof(pixel);

  254.     LOAD_LEFT_EDGE

  255. 

  256.     src[0+0*stride]=(l0 + l1 + 1)>>1;

  257.     src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;

  258.     src[2+0*stride]=

  259.     src[0+1*stride]=(l1 + l2 + 1)>>1;

  260.     src[3+0*stride]=

  261.     src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;

  262.     src[2+1*stride]=

  263.     src[0+2*stride]=(l2 + l3 + 1)>>1;

  264.     src[3+1*stride]=

  265.     src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;

  266.     src[3+2*stride]=

  267.     src[1+3*stride]=

  268.     src[0+3*stride]=

  269.     src[2+2*stride]=

  270.     src[2+3*stride]=

  271.     src[3+3*stride]=l3;

  272. }

  273. 

  274. static void FUNCC(pred4x4_horizontal_down)(uint8_t *_src, const uint8_t *topright, int _stride){

  275.     pixel *src = (pixel*)_src;

  276.     int stride = _stride/sizeof(pixel);

  277.     const int lt= src[-1-1*stride];

  278.     LOAD_TOP_EDGE

  279.     LOAD_LEFT_EDGE

  280. 

  281.     src[0+0*stride]=

  282.     src[2+1*stride]=(lt + l0 + 1)>>1;

  283.     src[1+0*stride]=

  284.     src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2;

  285.     src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2;

  286.     src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2;

  287.     src[0+1*stride]=

  288.     src[2+2*stride]=(l0 + l1 + 1)>>1;

  289.     src[1+1*stride]=

  290.     src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2;

  291.     src[0+2*stride]=

  292.     src[2+3*stride]=(l1 + l2+ 1)>>1;

  293.     src[1+2*stride]=

  294.     src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;

  295.     src[0+3*stride]=(l2 + l3 + 1)>>1;

  296.     src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;

  297. }

  298. 

  299. static void FUNCC(pred16x16_vertical)(uint8_t *_src, int _stride){

  300.     int i;

  301.     pixel *src = (pixel*)_src;

  302.     int stride = _stride/sizeof(pixel);

  303.     const pixel4 a = ((pixel4*)(src-stride))[0];

  304.     const pixel4 b = ((pixel4*)(src-stride))[1];

  305.     const pixel4 c = ((pixel4*)(src-stride))[2];

  306.     const pixel4 d = ((pixel4*)(src-stride))[3];

  307. 

  308.     for(i=0; i<16; i++){

  309.         ((pixel4*)(src+i*stride))[0] = a;

  310.         ((pixel4*)(src+i*stride))[1] = b;

  311.         ((pixel4*)(src+i*stride))[2] = c;

  312.         ((pixel4*)(src+i*stride))[3] = d;

  313.     }

  314. }

  315. 

  316. static void FUNCC(pred16x16_horizontal)(uint8_t *_src, int stride){

  317.     int i;

  318.     pixel *src = (pixel*)_src;

  319.     stride /= sizeof(pixel);

  320. 

  321.     for(i=0; i<16; i++){

  322.         ((pixel4*)(src+i*stride))[0] =

  323.         ((pixel4*)(src+i*stride))[1] =

  324.         ((pixel4*)(src+i*stride))[2] =

  325.         ((pixel4*)(src+i*stride))[3] = PIXEL_SPLAT_X4(src[-1+i*stride]);

  326.     }

  327. }

  328. 

  329. #define PREDICT_16x16_DC(v)\

  330.     for(i=0; i<16; i++){\

  331.         AV_WN4P(src+ 0, v);\

  332.         AV_WN4P(src+ 4, v);\

  333.         AV_WN4P(src+ 8, v);\

  334.         AV_WN4P(src+12, v);\

  335.         src += stride;\

  336.     }

  337. 

  338. static void FUNCC(pred16x16_dc)(uint8_t *_src, int stride){

  339.     int i, dc=0;

  340.     pixel *src = (pixel*)_src;

  341.     pixel4 dcsplat;

  342.     stride /= sizeof(pixel);

  343. 

  344.     for(i=0;i<16; i++){

  345.         dc+= src[-1+i*stride];

  346.     }

  347. 

  348.     for(i=0;i<16; i++){

  349.         dc+= src[i-stride];

  350.     }

  351. 

  352.     dcsplat = PIXEL_SPLAT_X4((dc+16)>>5);

  353.     PREDICT_16x16_DC(dcsplat);

  354. }

  355. 

  356. static void FUNCC(pred16x16_left_dc)(uint8_t *_src, int stride){

  357.     int i, dc=0;

  358.     pixel *src = (pixel*)_src;

  359.     pixel4 dcsplat;

  360.     stride /= sizeof(pixel);

  361. 

  362.     for(i=0;i<16; i++){

  363.         dc+= src[-1+i*stride];

  364.     }

  365. 

  366.     dcsplat = PIXEL_SPLAT_X4((dc+8)>>4);

  367.     PREDICT_16x16_DC(dcsplat);

  368. }

  369. 

  370. static void FUNCC(pred16x16_top_dc)(uint8_t *_src, int stride){

  371.     int i, dc=0;

  372.     pixel *src = (pixel*)_src;

  373.     pixel4 dcsplat;

  374.     stride /= sizeof(pixel);

  375. 

  376.     for(i=0;i<16; i++){

  377.         dc+= src[i-stride];

  378.     }

  379. 

  380.     dcsplat = PIXEL_SPLAT_X4((dc+8)>>4);

  381.     PREDICT_16x16_DC(dcsplat);

  382. }

  383. 

  384. #define PRED16x16_X(n, v) \

  385. static void FUNCC(pred16x16_##n##_dc)(uint8_t *_src, int stride){\

  386.     int i;\

  387.     pixel *src = (pixel*)_src;\

  388.     stride /= sizeof(pixel);\

  389.     PREDICT_16x16_DC(PIXEL_SPLAT_X4(v));\

  390. }

  391. 

  392. PRED16x16_X(127, (1<<(BIT_DEPTH-1))-1);

  393. PRED16x16_X(128, (1<<(BIT_DEPTH-1))+0);

  394. PRED16x16_X(129, (1<<(BIT_DEPTH-1))+1);

  395. 

  396. static inline void FUNCC(pred16x16_plane_compat)(uint8_t *_src, int _stride, const int svq3, const int rv40){

  397.   int i, j, k;

  398.   int a;

  399.   INIT_CLIP

  400.   pixel *src = (pixel*)_src;

  401.   int stride = _stride/sizeof(pixel);

  402.   const pixel * const src0 = src +7-stride;

  403.   const pixel *       src1 = src +8*stride-1;

  404.   const pixel *       src2 = src1-2*stride;    // == src+6*stride-1;

  405.   int H = src0[1] - src0[-1];

  406.   int V = src1[0] - src2[ 0];

  407.   for(k=2; k<=8; ++k) {

  408.     src1 += stride; src2 -= stride;

  409.     H += k*(src0[k] - src0[-k]);

  410.     V += k*(src1[0] - src2[ 0]);

  411.   }

  412.   if(svq3){

  413.     H = ( 5*(H/4) ) / 16;

  414.     V = ( 5*(V/4) ) / 16;

  415. 

  416.     /* required for 100% accuracy */

  417.     i = H; H = V; V = i;

  418.   }else if(rv40){

  419.     H = ( H + (H>>2) ) >> 4;

  420.     V = ( V + (V>>2) ) >> 4;

  421.   }else{

  422.     H = ( 5*H+32 ) >> 6;

  423.     V = ( 5*V+32 ) >> 6;

  424.   }

  425. 

  426.   a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);

  427.   for(j=16; j>0; --j) {

  428.     int b = a;

  429.     a += V;

  430.     for(i=-16; i<0; i+=4) {

  431.       src[16+i] = CLIP((b    ) >> 5);

  432.       src[17+i] = CLIP((b+  H) >> 5);

  433.       src[18+i] = CLIP((b+2*H) >> 5);

  434.       src[19+i] = CLIP((b+3*H) >> 5);

  435.       b += 4*H;

  436.     }

  437.     src += stride;

  438.   }

  439. }

  440. 

  441. static void FUNCC(pred16x16_plane)(uint8_t *src, int stride){

  442.     FUNCC(pred16x16_plane_compat)(src, stride, 0, 0);

  443. }

  444. 

  445. static void FUNCC(pred8x8_vertical)(uint8_t *_src, int _stride){

  446.     int i;

  447.     pixel *src = (pixel*)_src;

  448.     int stride = _stride/sizeof(pixel);

  449.     const pixel4 a= ((pixel4*)(src-stride))[0];

  450.     const pixel4 b= ((pixel4*)(src-stride))[1];

  451. 

  452.     for(i=0; i<8; i++){

  453.         ((pixel4*)(src+i*stride))[0]= a;

  454.         ((pixel4*)(src+i*stride))[1]= b;

  455.     }

  456. }

  457. 

  458. static void FUNCC(pred8x8_horizontal)(uint8_t *_src, int stride){

  459.     int i;

  460.     pixel *src = (pixel*)_src;

  461.     stride /= sizeof(pixel);

  462. 

  463.     for(i=0; i<8; i++){

  464.         ((pixel4*)(src+i*stride))[0]=

  465.         ((pixel4*)(src+i*stride))[1]= PIXEL_SPLAT_X4(src[-1+i*stride]);

  466.     }

  467. }

  468. 

  469. #define PRED8x8_X(n, v)\

  470. static void FUNCC(pred8x8_##n##_dc)(uint8_t *_src, int stride){\

  471.     int i;\

  472.     pixel *src = (pixel*)_src;\

  473.     stride /= sizeof(pixel);\

  474.     for(i=0; i<8; i++){\

  475.         ((pixel4*)(src+i*stride))[0]=\

  476.         ((pixel4*)(src+i*stride))[1]= PIXEL_SPLAT_X4(v);\

  477.     }\

  478. }

  479. 

  480. PRED8x8_X(127, (1<<(BIT_DEPTH-1))-1);

  481. PRED8x8_X(128, (1<<(BIT_DEPTH-1))+0);

  482. PRED8x8_X(129, (1<<(BIT_DEPTH-1))+1);

  483. 

  484. static void FUNCC(pred8x8_left_dc)(uint8_t *_src, int stride){

  485.     int i;

  486.     int dc0, dc2;

  487.     pixel4 dc0splat, dc2splat;

  488.     pixel *src = (pixel*)_src;

  489.     stride /= sizeof(pixel);

  490. 

  491.     dc0=dc2=0;

  492.     for(i=0;i<4; i++){

  493.         dc0+= src[-1+i*stride];

  494.         dc2+= src[-1+(i+4)*stride];

  495.     }

  496.     dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);

  497.     dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);

  498. 

  499.     for(i=0; i<4; i++){

  500.         ((pixel4*)(src+i*stride))[0]=

  501.         ((pixel4*)(src+i*stride))[1]= dc0splat;

  502.     }

  503.     for(i=4; i<8; i++){

  504.         ((pixel4*)(src+i*stride))[0]=

  505.         ((pixel4*)(src+i*stride))[1]= dc2splat;

  506.     }

  507. }

  508. 

  509. static void FUNCC(pred8x8_top_dc)(uint8_t *_src, int stride){

  510.     int i;

  511.     int dc0, dc1;

  512.     pixel4 dc0splat, dc1splat;

  513.     pixel *src = (pixel*)_src;

  514.     stride /= sizeof(pixel);

  515. 

  516.     dc0=dc1=0;

  517.     for(i=0;i<4; i++){

  518.         dc0+= src[i-stride];

  519.         dc1+= src[4+i-stride];

  520.     }

  521.     dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);

  522.     dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);

  523. 

  524.     for(i=0; i<4; i++){

  525.         ((pixel4*)(src+i*stride))[0]= dc0splat;

  526.         ((pixel4*)(src+i*stride))[1]= dc1splat;

  527.     }

  528.     for(i=4; i<8; i++){

  529.         ((pixel4*)(src+i*stride))[0]= dc0splat;

  530.         ((pixel4*)(src+i*stride))[1]= dc1splat;

  531.     }

  532. }

  533. 

  534. static void FUNCC(pred8x8_dc)(uint8_t *_src, int stride){

  535.     int i;

  536.     int dc0, dc1, dc2;

  537.     pixel4 dc0splat, dc1splat, dc2splat, dc3splat;

  538.     pixel *src = (pixel*)_src;

  539.     stride /= sizeof(pixel);

  540. 

  541.     dc0=dc1=dc2=0;

  542.     for(i=0;i<4; i++){

  543.         dc0+= src[-1+i*stride] + src[i-stride];

  544.         dc1+= src[4+i-stride];

  545.         dc2+= src[-1+(i+4)*stride];

  546.     }

  547.     dc0splat = PIXEL_SPLAT_X4((dc0 + 4)>>3);

  548.     dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);

  549.     dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);

  550.     dc3splat = PIXEL_SPLAT_X4((dc1 + dc2 + 4)>>3);

  551. 

  552.     for(i=0; i<4; i++){

  553.         ((pixel4*)(src+i*stride))[0]= dc0splat;

  554.         ((pixel4*)(src+i*stride))[1]= dc1splat;

  555.     }

  556.     for(i=4; i<8; i++){

  557.         ((pixel4*)(src+i*stride))[0]= dc2splat;

  558.         ((pixel4*)(src+i*stride))[1]= dc3splat;

  559.     }

  560. }

  561. 

  562. //the following 4 function should not be optimized!

  563. static void FUNC(pred8x8_mad_cow_dc_l0t)(uint8_t *src, int stride){

  564.     FUNCC(pred8x8_top_dc)(src, stride);

  565.     FUNCC(pred4x4_dc)(src, NULL, stride);

  566. }

  567. 

  568. static void FUNC(pred8x8_mad_cow_dc_0lt)(uint8_t *src, int stride){

  569.     FUNCC(pred8x8_dc)(src, stride);

  570.     FUNCC(pred4x4_top_dc)(src, NULL, stride);

  571. }

  572. 

  573. static void FUNC(pred8x8_mad_cow_dc_l00)(uint8_t *src, int stride){

  574.     FUNCC(pred8x8_left_dc)(src, stride);

  575.     FUNCC(pred4x4_128_dc)(src + 4*stride                  , NULL, stride);

  576.     FUNCC(pred4x4_128_dc)(src + 4*stride + 4*sizeof(pixel), NULL, stride);

  577. }

  578. 

  579. static void FUNC(pred8x8_mad_cow_dc_0l0)(uint8_t *src, int stride){

  580.     FUNCC(pred8x8_left_dc)(src, stride);

  581.     FUNCC(pred4x4_128_dc)(src                  , NULL, stride);

  582.     FUNCC(pred4x4_128_dc)(src + 4*sizeof(pixel), NULL, stride);

  583. }

  584. 

  585. static void FUNCC(pred8x8_plane)(uint8_t *_src, int _stride){

  586.   int j, k;

  587.   int a;

  588.   INIT_CLIP

  589.   pixel *src = (pixel*)_src;

  590.   int stride = _stride/sizeof(pixel);

  591.   const pixel * const src0 = src +3-stride;

  592.   const pixel *       src1 = src +4*stride-1;

  593.   const pixel *       src2 = src1-2*stride;    // == src+2*stride-1;

  594.   int H = src0[1] - src0[-1];

  595.   int V = src1[0] - src2[ 0];

  596.   for(k=2; k<=4; ++k) {

  597.     src1 += stride; src2 -= stride;

  598.     H += k*(src0[k] - src0[-k]);

  599.     V += k*(src1[0] - src2[ 0]);

  600.   }

  601.   H = ( 17*H+16 ) >> 5;

  602.   V = ( 17*V+16 ) >> 5;

  603. 

  604.   a = 16*(src1[0] + src2[8]+1) - 3*(V+H);

  605.   for(j=8; j>0; --j) {

  606.     int b = a;

  607.     a += V;

  608.     src[0] = CLIP((b    ) >> 5);

  609.     src[1] = CLIP((b+  H) >> 5);

  610.     src[2] = CLIP((b+2*H) >> 5);

  611.     src[3] = CLIP((b+3*H) >> 5);

  612.     src[4] = CLIP((b+4*H) >> 5);

  613.     src[5] = CLIP((b+5*H) >> 5);

  614.     src[6] = CLIP((b+6*H) >> 5);

  615.     src[7] = CLIP((b+7*H) >> 5);

  616.     src += stride;

  617.   }

  618. }

  619. 

  620. #define SRC(x,y) src[(x)+(y)*stride]

  621. #define PL(y) \

  622.     const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;

  623. #define PREDICT_8x8_LOAD_LEFT \

  624.     const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \

  625.                      + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \

  626.     PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \

  627.     const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2

  628. 

  629. #define PT(x) \

  630.     const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;

  631. #define PREDICT_8x8_LOAD_TOP \

  632.     const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \

  633.                      + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \

  634.     PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \

  635.     const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \

  636.                      + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2

  637. 

  638. #define PTR(x) \

  639.     t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;

  640. #define PREDICT_8x8_LOAD_TOPRIGHT \

  641.     int t8, t9, t10, t11, t12, t13, t14, t15; \

  642.     if(has_topright) { \

  643.         PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \

  644.         t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \

  645.     } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);

  646. 

  647. #define PREDICT_8x8_LOAD_TOPLEFT \

  648.     const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2

  649. 

  650. #define PREDICT_8x8_DC(v) \

  651.     int y; \

  652.     for( y = 0; y < 8; y++ ) { \

  653.         ((pixel4*)src)[0] = \

  654.         ((pixel4*)src)[1] = v; \

  655.         src += stride; \

  656.     }

  657. 

  658. static void FUNCC(pred8x8l_128_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  659. {

  660.     pixel *src = (pixel*)_src;

  661.     int stride = _stride/sizeof(pixel);

  662. 

  663.     PREDICT_8x8_DC(PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1)));

  664. }

  665. static void FUNCC(pred8x8l_left_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  666. {

  667.     pixel *src = (pixel*)_src;

  668.     int stride = _stride/sizeof(pixel);

  669. 

  670.     PREDICT_8x8_LOAD_LEFT;

  671.     const pixel4 dc = PIXEL_SPLAT_X4((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3);

  672.     PREDICT_8x8_DC(dc);

  673. }

  674. static void FUNCC(pred8x8l_top_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  675. {

  676.     pixel *src = (pixel*)_src;

  677.     int stride = _stride/sizeof(pixel);

  678. 

  679.     PREDICT_8x8_LOAD_TOP;

  680.     const pixel4 dc = PIXEL_SPLAT_X4((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3);

  681.     PREDICT_8x8_DC(dc);

  682. }

  683. static void FUNCC(pred8x8l_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  684. {

  685.     pixel *src = (pixel*)_src;

  686.     int stride = _stride/sizeof(pixel);

  687. 

  688.     PREDICT_8x8_LOAD_LEFT;

  689.     PREDICT_8x8_LOAD_TOP;

  690.     const pixel4 dc = PIXEL_SPLAT_X4((l0+l1+l2+l3+l4+l5+l6+l7

  691.                                      +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4);

  692.     PREDICT_8x8_DC(dc);

  693. }

  694. static void FUNCC(pred8x8l_horizontal)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  695. {

  696.     pixel *src = (pixel*)_src;

  697.     int stride = _stride/sizeof(pixel);

  698. 

  699.     PREDICT_8x8_LOAD_LEFT;

  700. #define ROW(y) ((pixel4*)(src+y*stride))[0] =\

  701.                ((pixel4*)(src+y*stride))[1] = PIXEL_SPLAT_X4(l##y)

  702.     ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);

  703. #undef ROW

  704. }

  705. static void FUNCC(pred8x8l_vertical)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  706. {

  707.     int y;

  708.     pixel *src = (pixel*)_src;

  709.     int stride = _stride/sizeof(pixel);

  710. 

  711.     PREDICT_8x8_LOAD_TOP;

  712.     src[0] = t0;

  713.     src[1] = t1;

  714.     src[2] = t2;

  715.     src[3] = t3;

  716.     src[4] = t4;

  717.     src[5] = t5;

  718.     src[6] = t6;

  719.     src[7] = t7;

  720.     for( y = 1; y < 8; y++ ) {

  721.         ((pixel4*)(src+y*stride))[0] = ((pixel4*)src)[0];

  722.         ((pixel4*)(src+y*stride))[1] = ((pixel4*)src)[1];

  723.     }

  724. }

  725. static void FUNCC(pred8x8l_down_left)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  726. {

  727.     pixel *src = (pixel*)_src;

  728.     int stride = _stride/sizeof(pixel);

  729.     PREDICT_8x8_LOAD_TOP;

  730.     PREDICT_8x8_LOAD_TOPRIGHT;

  731.     SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2;

  732.     SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2;

  733.     SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2;

  734.     SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2;

  735.     SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2;

  736.     SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2;

  737.     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;

  738.     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;

  739.     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;

  740.     SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2;

  741.     SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2;

  742.     SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2;

  743.     SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2;

  744.     SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;

  745.     SRC(7,7)= (t14 + 3*t15 + 2) >> 2;

  746. }

  747. static void FUNCC(pred8x8l_down_right)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  748. {

  749.     pixel *src = (pixel*)_src;

  750.     int stride = _stride/sizeof(pixel);

  751.     PREDICT_8x8_LOAD_TOP;

  752.     PREDICT_8x8_LOAD_LEFT;

  753.     PREDICT_8x8_LOAD_TOPLEFT;

  754.     SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2;

  755.     SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2;

  756.     SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2;

  757.     SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2;

  758.     SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2;

  759.     SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2;

  760.     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;

  761.     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;

  762.     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;

  763.     SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2;

  764.     SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2;

  765.     SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2;

  766.     SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2;

  767.     SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;

  768.     SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;

  769. }

  770. static void FUNCC(pred8x8l_vertical_right)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  771. {

  772.     pixel *src = (pixel*)_src;

  773.     int stride = _stride/sizeof(pixel);

  774.     PREDICT_8x8_LOAD_TOP;

  775.     PREDICT_8x8_LOAD_LEFT;

  776.     PREDICT_8x8_LOAD_TOPLEFT;

  777.     SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2;

  778.     SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2;

  779.     SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2;

  780.     SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2;

  781.     SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2;

  782.     SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2;

  783.     SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2;

  784.     SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1;

  785.     SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2;

  786.     SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1;

  787.     SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2;

  788.     SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1;

  789.     SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2;

  790.     SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1;

  791.     SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2;

  792.     SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1;

  793.     SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2;

  794.     SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1;

  795.     SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2;

  796.     SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1;

  797.     SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;

  798.     SRC(7,0)= (t6 + t7 + 1) >> 1;

  799. }

  800. static void FUNCC(pred8x8l_horizontal_down)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  801. {

  802.     pixel *src = (pixel*)_src;

  803.     int stride = _stride/sizeof(pixel);

  804.     PREDICT_8x8_LOAD_TOP;

  805.     PREDICT_8x8_LOAD_LEFT;

  806.     PREDICT_8x8_LOAD_TOPLEFT;

  807.     SRC(0,7)= (l6 + l7 + 1) >> 1;

  808.     SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2;

  809.     SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1;

  810.     SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2;

  811.     SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1;

  812.     SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2;

  813.     SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1;

  814.     SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2;

  815.     SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1;

  816.     SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2;

  817.     SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1;

  818.     SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2;

  819.     SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1;

  820.     SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2;

  821.     SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1;

  822.     SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2;

  823.     SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2;

  824.     SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2;

  825.     SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2;

  826.     SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2;

  827.     SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;

  828.     SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;

  829. }

  830. static void FUNCC(pred8x8l_vertical_left)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  831. {

  832.     pixel *src = (pixel*)_src;

  833.     int stride = _stride/sizeof(pixel);

  834.     PREDICT_8x8_LOAD_TOP;

  835.     PREDICT_8x8_LOAD_TOPRIGHT;

  836.     SRC(0,0)= (t0 + t1 + 1) >> 1;

  837.     SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2;

  838.     SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1;

  839.     SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2;

  840.     SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1;

  841.     SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2;

  842.     SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1;

  843.     SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2;

  844.     SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1;

  845.     SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2;

  846.     SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1;

  847.     SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2;

  848.     SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1;

  849.     SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2;

  850.     SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1;

  851.     SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2;

  852.     SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1;

  853.     SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2;

  854.     SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1;

  855.     SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2;

  856.     SRC(7,6)= (t10 + t11 + 1) >> 1;

  857.     SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;

  858. }

  859. static void FUNCC(pred8x8l_horizontal_up)(uint8_t *_src, int has_topleft, int has_topright, int _stride)

  860. {

  861.     pixel *src = (pixel*)_src;

  862.     int stride = _stride/sizeof(pixel);

  863.     PREDICT_8x8_LOAD_LEFT;

  864.     SRC(0,0)= (l0 + l1 + 1) >> 1;

  865.     SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2;

  866.     SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1;

  867.     SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2;

  868.     SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1;

  869.     SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2;

  870.     SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1;

  871.     SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2;

  872.     SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1;

  873.     SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2;

  874.     SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1;

  875.     SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2;

  876.     SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1;

  877.     SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2;

  878.     SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)=

  879.     SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)=

  880.     SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=

  881.     SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;

  882. }

  883. #undef PREDICT_8x8_LOAD_LEFT

  884. #undef PREDICT_8x8_LOAD_TOP

  885. #undef PREDICT_8x8_LOAD_TOPLEFT

  886. #undef PREDICT_8x8_LOAD_TOPRIGHT

  887. #undef PREDICT_8x8_DC

  888. #undef PTR

  889. #undef PT

  890. #undef PL

  891. #undef SRC

  892. 

  893. static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, const DCTELEM *_block, int stride){

  894.     int i;

  895.     pixel *pix = (pixel*)_pix;

  896.     const dctcoef *block = (const dctcoef*)_block;

  897.     stride /= sizeof(pixel);

  898.     pix -= stride;

  899.     for(i=0; i<4; i++){

  900.         pixel v = pix[0];

  901.         pix[1*stride]= v += block[0];

  902.         pix[2*stride]= v += block[4];

  903.         pix[3*stride]= v += block[8];

  904.         pix[4*stride]= v +  block[12];

  905.         pix++;

  906.         block++;

  907.     }

  908. }

  909. 

  910. static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, const DCTELEM *_block, int stride){

  911.     int i;

  912.     pixel *pix = (pixel*)_pix;

  913.     const dctcoef *block = (const dctcoef*)_block;

  914.     stride /= sizeof(pixel);

  915.     for(i=0; i<4; i++){

  916.         pixel v = pix[-1];

  917.         pix[0]= v += block[0];

  918.         pix[1]= v += block[1];

  919.         pix[2]= v += block[2];

  920.         pix[3]= v +  block[3];

  921.         pix+= stride;

  922.         block+= 4;

  923.     }

  924. }

  925. 

  926. static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, const DCTELEM *_block, int stride){

  927.     int i;

  928.     pixel *pix = (pixel*)_pix;

  929.     const dctcoef *block = (const dctcoef*)_block;

  930.     stride /= sizeof(pixel);

  931.     pix -= stride;

  932.     for(i=0; i<8; i++){

  933.         pixel v = pix[0];

  934.         pix[1*stride]= v += block[0];

  935.         pix[2*stride]= v += block[8];

  936.         pix[3*stride]= v += block[16];

  937.         pix[4*stride]= v += block[24];

  938.         pix[5*stride]= v += block[32];

  939.         pix[6*stride]= v += block[40];

  940.         pix[7*stride]= v += block[48];

  941.         pix[8*stride]= v +  block[56];

  942.         pix++;

  943.         block++;

  944.     }

  945. }

  946. 

  947. static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, const DCTELEM *_block, int stride){

  948.     int i;

  949.     pixel *pix = (pixel*)_pix;

  950.     const dctcoef *block = (const dctcoef*)_block;

  951.     stride /= sizeof(pixel);

  952.     for(i=0; i<8; i++){

  953.         pixel v = pix[-1];

  954.         pix[0]= v += block[0];

  955.         pix[1]= v += block[1];

  956.         pix[2]= v += block[2];

  957.         pix[3]= v += block[3];

  958.         pix[4]= v += block[4];

  959.         pix[5]= v += block[5];

  960.         pix[6]= v += block[6];

  961.         pix[7]= v +  block[7];

  962.         pix+= stride;

  963.         block+= 8;

  964.     }

  965. }

  966. 

  967. static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){

  968.     int i;

  969.     for(i=0; i<16; i++)

  970.         FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);

  971. }

  972. 

  973. static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){

  974.     int i;

  975.     for(i=0; i<16; i++)

  976.         FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);

  977. }

  978. 

  979. static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){

  980.     int i;

  981.     for(i=0; i<4; i++)

  982.         FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);

  983. }

  984. 

  985. static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){

  986.     int i;

  987.     for(i=0; i<4; i++)

  988.         FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);

  989. }