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.
43714 Commits
32 Branches
283MB
Tree: 6c180b35c4
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '6c180b35c4'
${ noResults }
FFmpeg/libavcodec/cavsdsp.c

550 lines
23KB
Raw Blame History 

  1. /*

  2.  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.

  3.  *

  4.  * DSP functions

  5.  *

  6.  * Copyright (c) 2006  Stefan Gehrer <stefan.gehrer@gmx.de>

  7.  *

  8.  * This file is part of FFmpeg.

  9.  *

  10.  * FFmpeg is free software; you can redistribute it and/or

  11.  * modify it under the terms of the GNU Lesser General Public

  12.  * License as published by the Free Software Foundation; either

  13.  * version 2.1 of the License, or (at your option) any later version.

  14.  *

  15.  * FFmpeg 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 GNU

  18.  * Lesser General Public License for more details.

  19.  *

  20.  * You should have received a copy of the GNU Lesser General Public

  21.  * License along with FFmpeg; if not, write to the Free Software

  22.  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

  23.  */

  24. 

  25. #include <stdio.h>

  26. #include "dsputil.h"

  27. #include "cavsdsp.h"

  28. #include "libavutil/common.h"

  29. 

  30. /*****************************************************************************

  31.  *

  32.  * in-loop deblocking filter

  33.  *

  34.  ****************************************************************************/

  35. 

  36. #define P2 p0_p[-3*stride]

  37. #define P1 p0_p[-2*stride]

  38. #define P0 p0_p[-1*stride]

  39. #define Q0 p0_p[ 0*stride]

  40. #define Q1 p0_p[ 1*stride]

  41. #define Q2 p0_p[ 2*stride]

  42. 

  43. static inline void loop_filter_l2(uint8_t *p0_p,int stride,int alpha, int beta) {

  44.     int p0 = P0;

  45.     int q0 = Q0;

  46. 

  47.     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {

  48.         int s = p0 + q0 + 2;

  49.         alpha = (alpha>>2) + 2;

  50.         if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {

  51.             P0 = (P1 + p0 + s) >> 2;

  52.             P1 = (2*P1 + s) >> 2;

  53.         } else

  54.             P0 = (2*P1 + s) >> 2;

  55.         if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {

  56.             Q0 = (Q1 + q0 + s) >> 2;

  57.             Q1 = (2*Q1 + s) >> 2;

  58.         } else

  59.             Q0 = (2*Q1 + s) >> 2;

  60.     }

  61. }

  62. 

  63. static inline void loop_filter_l1(uint8_t *p0_p, int stride, int alpha, int beta, int tc) {

  64.     int p0 = P0;

  65.     int q0 = Q0;

  66. 

  67.     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {

  68.         int delta = av_clip(((q0-p0)*3+P1-Q1+4)>>3,-tc, tc);

  69.         P0 = av_clip_uint8(p0+delta);

  70.         Q0 = av_clip_uint8(q0-delta);

  71.         if(abs(P2-p0)<beta) {

  72.             delta = av_clip(((P0-P1)*3+P2-Q0+4)>>3, -tc, tc);

  73.             P1 = av_clip_uint8(P1+delta);

  74.         }

  75.         if(abs(Q2-q0)<beta) {

  76.             delta = av_clip(((Q1-Q0)*3+P0-Q2+4)>>3, -tc, tc);

  77.             Q1 = av_clip_uint8(Q1-delta);

  78.         }

  79.     }

  80. }

  81. 

  82. static inline void loop_filter_c2(uint8_t *p0_p,int stride,int alpha, int beta) {

  83.     int p0 = P0;

  84.     int q0 = Q0;

  85. 

  86.     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {

  87.         int s = p0 + q0 + 2;

  88.         alpha = (alpha>>2) + 2;

  89.         if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {

  90.             P0 = (P1 + p0 + s) >> 2;

  91.         } else

  92.             P0 = (2*P1 + s) >> 2;

  93.         if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {

  94.             Q0 = (Q1 + q0 + s) >> 2;

  95.         } else

  96.             Q0 = (2*Q1 + s) >> 2;

  97.     }

  98. }

  99. 

  100. static inline void loop_filter_c1(uint8_t *p0_p,int stride,int alpha, int beta,

  101.                                   int tc) {

  102.     if(abs(P0-Q0)<alpha && abs(P1-P0)<beta && abs(Q1-Q0)<beta) {

  103.         int delta = av_clip(((Q0-P0)*3+P1-Q1+4)>>3, -tc, tc);

  104.         P0 = av_clip_uint8(P0+delta);

  105.         Q0 = av_clip_uint8(Q0-delta);

  106.     }

  107. }

  108. 

  109. #undef P0

  110. #undef P1

  111. #undef P2

  112. #undef Q0

  113. #undef Q1

  114. #undef Q2

  115. 

  116. static void cavs_filter_lv_c(uint8_t *d, int stride, int alpha, int beta, int tc,

  117.                            int bs1, int bs2) {

  118.     int i;

  119.     if(bs1==2)

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

  121.             loop_filter_l2(d + i*stride,1,alpha,beta);

  122.     else {

  123.         if(bs1)

  124.             for(i=0;i<8;i++)

  125.                 loop_filter_l1(d + i*stride,1,alpha,beta,tc);

  126.         if (bs2)

  127.             for(i=8;i<16;i++)

  128.                 loop_filter_l1(d + i*stride,1,alpha,beta,tc);

  129.     }

  130. }

  131. 

  132. static void cavs_filter_lh_c(uint8_t *d, int stride, int alpha, int beta, int tc,

  133.                            int bs1, int bs2) {

  134.     int i;

  135.     if(bs1==2)

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

  137.             loop_filter_l2(d + i,stride,alpha,beta);

  138.     else {

  139.         if(bs1)

  140.             for(i=0;i<8;i++)

  141.                 loop_filter_l1(d + i,stride,alpha,beta,tc);

  142.         if (bs2)

  143.             for(i=8;i<16;i++)

  144.                 loop_filter_l1(d + i,stride,alpha,beta,tc);

  145.     }

  146. }

  147. 

  148. static void cavs_filter_cv_c(uint8_t *d, int stride, int alpha, int beta, int tc,

  149.                            int bs1, int bs2) {

  150.     int i;

  151.     if(bs1==2)

  152.         for(i=0;i<8;i++)

  153.             loop_filter_c2(d + i*stride,1,alpha,beta);

  154.     else {

  155.         if(bs1)

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

  157.                 loop_filter_c1(d + i*stride,1,alpha,beta,tc);

  158.         if (bs2)

  159.             for(i=4;i<8;i++)

  160.                 loop_filter_c1(d + i*stride,1,alpha,beta,tc);

  161.     }

  162. }

  163. 

  164. static void cavs_filter_ch_c(uint8_t *d, int stride, int alpha, int beta, int tc,

  165.                            int bs1, int bs2) {

  166.     int i;

  167.     if(bs1==2)

  168.         for(i=0;i<8;i++)

  169.             loop_filter_c2(d + i,stride,alpha,beta);

  170.     else {

  171.         if(bs1)

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

  173.                 loop_filter_c1(d + i,stride,alpha,beta,tc);

  174.         if (bs2)

  175.             for(i=4;i<8;i++)

  176.                 loop_filter_c1(d + i,stride,alpha,beta,tc);

  177.     }

  178. }

  179. 

  180. /*****************************************************************************

  181.  *

  182.  * inverse transform

  183.  *

  184.  ****************************************************************************/

  185. 

  186. static void cavs_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride) {

  187.     int i;

  188.     DCTELEM (*src)[8] = (DCTELEM(*)[8])block;

  189.     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;

  190. 

  191.     src[0][0] += 8;

  192. 

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

  194.         const int a0 =  3*src[i][1] - (src[i][7]<<1);

  195.         const int a1 =  3*src[i][3] + (src[i][5]<<1);

  196.         const int a2 =  (src[i][3]<<1) - 3*src[i][5];

  197.         const int a3 =  (src[i][1]<<1) + 3*src[i][7];

  198. 

  199.         const int b4 = ((a0 + a1 + a3)<<1) + a1;

  200.         const int b5 = ((a0 - a1 + a2)<<1) + a0;

  201.         const int b6 = ((a3 - a2 - a1)<<1) + a3;

  202.         const int b7 = ((a0 - a2 - a3)<<1) - a2;

  203. 

  204.         const int a7 = (src[i][2]<<2) - 10*src[i][6];

  205.         const int a6 = (src[i][6]<<2) + 10*src[i][2];

  206.         const int a5 = ((src[i][0] - src[i][4]) << 3) + 4;

  207.         const int a4 = ((src[i][0] + src[i][4]) << 3) + 4;

  208. 

  209.         const int b0 = a4 + a6;

  210.         const int b1 = a5 + a7;

  211.         const int b2 = a5 - a7;

  212.         const int b3 = a4 - a6;

  213. 

  214.         src[i][0] = (b0 + b4) >> 3;

  215.         src[i][1] = (b1 + b5) >> 3;

  216.         src[i][2] = (b2 + b6) >> 3;

  217.         src[i][3] = (b3 + b7) >> 3;

  218.         src[i][4] = (b3 - b7) >> 3;

  219.         src[i][5] = (b2 - b6) >> 3;

  220.         src[i][6] = (b1 - b5) >> 3;

  221.         src[i][7] = (b0 - b4) >> 3;

  222.     }

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

  224.         const int a0 =  3*src[1][i] - (src[7][i]<<1);

  225.         const int a1 =  3*src[3][i] + (src[5][i]<<1);

  226.         const int a2 =  (src[3][i]<<1) - 3*src[5][i];

  227.         const int a3 =  (src[1][i]<<1) + 3*src[7][i];

  228. 

  229.         const int b4 = ((a0 + a1 + a3)<<1) + a1;

  230.         const int b5 = ((a0 - a1 + a2)<<1) + a0;

  231.         const int b6 = ((a3 - a2 - a1)<<1) + a3;

  232.         const int b7 = ((a0 - a2 - a3)<<1) - a2;

  233. 

  234.         const int a7 = (src[2][i]<<2) - 10*src[6][i];

  235.         const int a6 = (src[6][i]<<2) + 10*src[2][i];

  236.         const int a5 = (src[0][i] - src[4][i]) << 3;

  237.         const int a4 = (src[0][i] + src[4][i]) << 3;

  238. 

  239.         const int b0 = a4 + a6;

  240.         const int b1 = a5 + a7;

  241.         const int b2 = a5 - a7;

  242.         const int b3 = a4 - a6;

  243. 

  244.         dst[i + 0*stride] = cm[ dst[i + 0*stride] + ((b0 + b4) >> 7)];

  245.         dst[i + 1*stride] = cm[ dst[i + 1*stride] + ((b1 + b5) >> 7)];

  246.         dst[i + 2*stride] = cm[ dst[i + 2*stride] + ((b2 + b6) >> 7)];

  247.         dst[i + 3*stride] = cm[ dst[i + 3*stride] + ((b3 + b7) >> 7)];

  248.         dst[i + 4*stride] = cm[ dst[i + 4*stride] + ((b3 - b7) >> 7)];

  249.         dst[i + 5*stride] = cm[ dst[i + 5*stride] + ((b2 - b6) >> 7)];

  250.         dst[i + 6*stride] = cm[ dst[i + 6*stride] + ((b1 - b5) >> 7)];

  251.         dst[i + 7*stride] = cm[ dst[i + 7*stride] + ((b0 - b4) >> 7)];

  252.     }

  253. }

  254. 

  255. /*****************************************************************************

  256.  *

  257.  * motion compensation

  258.  *

  259.  ****************************************************************************/

  260. 

  261. #define CAVS_SUBPIX(OPNAME, OP, NAME, A, B, C, D, E, F) \

  262. static void OPNAME ## cavs_filt8_h_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  263.     const int h=8;\

  264.     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\

  265.     int i;\

  266.     for(i=0; i<h; i++)\

  267.     {\

  268.         OP(dst[0], A*src[-2] + B*src[-1] + C*src[0] + D*src[1] + E*src[2] + F*src[3]);\

  269.         OP(dst[1], A*src[-1] + B*src[ 0] + C*src[1] + D*src[2] + E*src[3] + F*src[4]);\

  270.         OP(dst[2], A*src[ 0] + B*src[ 1] + C*src[2] + D*src[3] + E*src[4] + F*src[5]);\

  271.         OP(dst[3], A*src[ 1] + B*src[ 2] + C*src[3] + D*src[4] + E*src[5] + F*src[6]);\

  272.         OP(dst[4], A*src[ 2] + B*src[ 3] + C*src[4] + D*src[5] + E*src[6] + F*src[7]);\

  273.         OP(dst[5], A*src[ 3] + B*src[ 4] + C*src[5] + D*src[6] + E*src[7] + F*src[8]);\

  274.         OP(dst[6], A*src[ 4] + B*src[ 5] + C*src[6] + D*src[7] + E*src[8] + F*src[9]);\

  275.         OP(dst[7], A*src[ 5] + B*src[ 6] + C*src[7] + D*src[8] + E*src[9] + F*src[10]);\

  276.         dst+=dstStride;\

  277.         src+=srcStride;\

  278.     }\

  279. }\

  280. \

  281. static void OPNAME ## cavs_filt8_v_  ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  282.     const int w=8;\

  283.     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\

  284.     int i;\

  285.     for(i=0; i<w; i++)\

  286.     {\

  287.         const int srcB= src[-2*srcStride];\

  288.         const int srcA= src[-1*srcStride];\

  289.         const int src0= src[0 *srcStride];\

  290.         const int src1= src[1 *srcStride];\

  291.         const int src2= src[2 *srcStride];\

  292.         const int src3= src[3 *srcStride];\

  293.         const int src4= src[4 *srcStride];\

  294.         const int src5= src[5 *srcStride];\

  295.         const int src6= src[6 *srcStride];\

  296.         const int src7= src[7 *srcStride];\

  297.         const int src8= src[8 *srcStride];\

  298.         const int src9= src[9 *srcStride];\

  299.         const int src10= src[10 *srcStride];\

  300.         OP(dst[0*dstStride], A*srcB + B*srcA + C*src0 + D*src1 + E*src2 + F*src3);\

  301.         OP(dst[1*dstStride], A*srcA + B*src0 + C*src1 + D*src2 + E*src3 + F*src4);\

  302.         OP(dst[2*dstStride], A*src0 + B*src1 + C*src2 + D*src3 + E*src4 + F*src5);\

  303.         OP(dst[3*dstStride], A*src1 + B*src2 + C*src3 + D*src4 + E*src5 + F*src6);\

  304.         OP(dst[4*dstStride], A*src2 + B*src3 + C*src4 + D*src5 + E*src6 + F*src7);\

  305.         OP(dst[5*dstStride], A*src3 + B*src4 + C*src5 + D*src6 + E*src7 + F*src8);\

  306.         OP(dst[6*dstStride], A*src4 + B*src5 + C*src6 + D*src7 + E*src8 + F*src9);\

  307.         OP(dst[7*dstStride], A*src5 + B*src6 + C*src7 + D*src8 + E*src9 + F*src10);\

  308.         dst++;\

  309.         src++;\

  310.     }\

  311. }\

  312. \

  313. static void OPNAME ## cavs_filt16_v_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  314.     OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\

  315.     OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\

  316.     src += 8*srcStride;\

  317.     dst += 8*dstStride;\

  318.     OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\

  319.     OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\

  320. }\

  321. \

  322. static void OPNAME ## cavs_filt16_h_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  323.     OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\

  324.     OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\

  325.     src += 8*srcStride;\

  326.     dst += 8*dstStride;\

  327.     OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\

  328.     OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\

  329. }\

  330. 

  331. #define CAVS_SUBPIX_HV(OPNAME, OP, NAME, AH, BH, CH, DH, EH, FH, AV, BV, CV, DV, EV, FV, FULL) \

  332. static void OPNAME ## cavs_filt8_hv_ ## NAME(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int srcStride){\

  333.     int16_t temp[8*(8+5)];\

  334.     int16_t *tmp = temp;\

  335.     const int h=8;\

  336.     const int w=8;\

  337.     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\

  338.     int i;\

  339.     src1 -= 2*srcStride;\

  340.     for(i=0; i<h+5; i++)\

  341.     {\

  342.         tmp[0]= AH*src1[-2] + BH*src1[-1] + CH*src1[0] + DH*src1[1] + EH*src1[2] + FH*src1[3];\

  343.         tmp[1]= AH*src1[-1] + BH*src1[ 0] + CH*src1[1] + DH*src1[2] + EH*src1[3] + FH*src1[4];\

  344.         tmp[2]= AH*src1[ 0] + BH*src1[ 1] + CH*src1[2] + DH*src1[3] + EH*src1[4] + FH*src1[5];\

  345.         tmp[3]= AH*src1[ 1] + BH*src1[ 2] + CH*src1[3] + DH*src1[4] + EH*src1[5] + FH*src1[6];\

  346.         tmp[4]= AH*src1[ 2] + BH*src1[ 3] + CH*src1[4] + DH*src1[5] + EH*src1[6] + FH*src1[7];\

  347.         tmp[5]= AH*src1[ 3] + BH*src1[ 4] + CH*src1[5] + DH*src1[6] + EH*src1[7] + FH*src1[8];\

  348.         tmp[6]= AH*src1[ 4] + BH*src1[ 5] + CH*src1[6] + DH*src1[7] + EH*src1[8] + FH*src1[9];\

  349.         tmp[7]= AH*src1[ 5] + BH*src1[ 6] + CH*src1[7] + DH*src1[8] + EH*src1[9] + FH*src1[10];\

  350.         tmp+=8;\

  351.         src1+=srcStride;\

  352.     }\

  353.     if(FULL) {\

  354.       tmp = temp+8*2;                           \

  355.       for(i=0; i<w; i++)                        \

  356.         {                                       \

  357.           const int tmpB= tmp[-2*8];    \

  358.           const int tmpA= tmp[-1*8];    \

  359.           const int tmp0= tmp[0 *8];    \

  360.           const int tmp1= tmp[1 *8];    \

  361.           const int tmp2= tmp[2 *8];    \

  362.           const int tmp3= tmp[3 *8];    \

  363.           const int tmp4= tmp[4 *8];    \

  364.           const int tmp5= tmp[5 *8];    \

  365.           const int tmp6= tmp[6 *8];    \

  366.           const int tmp7= tmp[7 *8];    \

  367.           const int tmp8= tmp[8 *8];    \

  368.           const int tmp9= tmp[9 *8];    \

  369.           const int tmp10=tmp[10*8];                            \

  370.           OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3 + 64*src2[0*srcStride]); \

  371.           OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4 + 64*src2[1*srcStride]); \

  372.           OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5 + 64*src2[2*srcStride]); \

  373.           OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6 + 64*src2[3*srcStride]); \

  374.           OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7 + 64*src2[4*srcStride]); \

  375.           OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8 + 64*src2[5*srcStride]); \

  376.           OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9 + 64*src2[6*srcStride]); \

  377.           OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10 + 64*src2[7*srcStride]); \

  378.           dst++;                                                        \

  379.           tmp++;                                                        \

  380.           src2++;                                                       \

  381.         }                                                               \

  382.     } else {\

  383.       tmp = temp+8*2;                           \

  384.       for(i=0; i<w; i++)                        \

  385.         {                                       \

  386.           const int tmpB= tmp[-2*8];    \

  387.           const int tmpA= tmp[-1*8];    \

  388.           const int tmp0= tmp[0 *8];    \

  389.           const int tmp1= tmp[1 *8];    \

  390.           const int tmp2= tmp[2 *8];    \

  391.           const int tmp3= tmp[3 *8];    \

  392.           const int tmp4= tmp[4 *8];    \

  393.           const int tmp5= tmp[5 *8];    \

  394.           const int tmp6= tmp[6 *8];    \

  395.           const int tmp7= tmp[7 *8];    \

  396.           const int tmp8= tmp[8 *8];    \

  397.           const int tmp9= tmp[9 *8];    \

  398.           const int tmp10=tmp[10*8];                            \

  399.           OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3); \

  400.           OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4); \

  401.           OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5); \

  402.           OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6); \

  403.           OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7); \

  404.           OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8); \

  405.           OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9); \

  406.           OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10); \

  407.           dst++;                                                        \

  408.           tmp++;                                                        \

  409.         }                                                               \

  410.     }\

  411. }\

  412. \

  413. static void OPNAME ## cavs_filt16_hv_ ## NAME(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int srcStride){ \

  414.     OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \

  415.     OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \

  416.     src1 += 8*srcStride;\

  417.     src2 += 8*srcStride;\

  418.     dst += 8*dstStride;\

  419.     OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \

  420.     OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \

  421. }\

  422. 

  423. #define CAVS_MC(OPNAME, SIZE) \

  424. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\

  425.     OPNAME ## cavs_filt ## SIZE ## _h_qpel_l(dst, src, stride, stride);\

  426. }\

  427. \

  428. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\

  429.     OPNAME ## cavs_filt ## SIZE ## _h_hpel(dst, src, stride, stride);\

  430. }\

  431. \

  432. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc30_c(uint8_t *dst, uint8_t *src, int stride){\

  433.     OPNAME ## cavs_filt ## SIZE ## _h_qpel_r(dst, src, stride, stride);\

  434. }\

  435. \

  436. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\

  437.     OPNAME ## cavs_filt ## SIZE ## _v_qpel_l(dst, src, stride, stride);\

  438. }\

  439. \

  440. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\

  441.     OPNAME ## cavs_filt ## SIZE ## _v_hpel(dst, src, stride, stride);\

  442. }\

  443. \

  444. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc03_c(uint8_t *dst, uint8_t *src, int stride){\

  445.     OPNAME ## cavs_filt ## SIZE ## _v_qpel_r(dst, src, stride, stride);\

  446. }\

  447. \

  448. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\

  449.   OPNAME ## cavs_filt ## SIZE ## _hv_jj(dst, src, NULL, stride, stride); \

  450. }\

  451. \

  452. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\

  453.   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src, stride, stride); \

  454. }\

  455. \

  456. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc13_c(uint8_t *dst, uint8_t *src, int stride){\

  457.   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride, stride, stride); \

  458. }\

  459. \

  460. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc31_c(uint8_t *dst, uint8_t *src, int stride){\

  461.   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+1, stride, stride); \

  462. }\

  463. \

  464. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc33_c(uint8_t *dst, uint8_t *src, int stride){\

  465.   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride+1,stride, stride); \

  466. }\

  467. \

  468. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\

  469.   OPNAME ## cavs_filt ## SIZE ## _hv_ff(dst, src, src+stride+1,stride, stride); \

  470. }\

  471. \

  472. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\

  473.   OPNAME ## cavs_filt ## SIZE ## _hv_ii(dst, src, src+stride+1,stride, stride); \

  474. }\

  475. \

  476. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, int stride){\

  477.   OPNAME ## cavs_filt ## SIZE ## _hv_kk(dst, src, src+stride+1,stride, stride); \

  478. }\

  479. \

  480. static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc23_c(uint8_t *dst, uint8_t *src, int stride){\

  481.   OPNAME ## cavs_filt ## SIZE ## _hv_qq(dst, src, src+stride+1,stride, stride); \

  482. }\

  483. 

  484. #define op_put1(a, b)  a = cm[((b)+4)>>3]

  485. #define op_put2(a, b)  a = cm[((b)+64)>>7]

  486. #define op_put3(a, b)  a = cm[((b)+32)>>6]

  487. #define op_put4(a, b)  a = cm[((b)+512)>>10]

  488. #define op_avg1(a, b)  a = ((a)+cm[((b)+4)>>3]   +1)>>1

  489. #define op_avg2(a, b)  a = ((a)+cm[((b)+64)>>7]  +1)>>1

  490. #define op_avg3(a, b)  a = ((a)+cm[((b)+32)>>6]  +1)>>1

  491. #define op_avg4(a, b)  a = ((a)+cm[((b)+512)>>10]+1)>>1

  492. CAVS_SUBPIX(put_   , op_put1, hpel,    0, -1,  5,  5, -1,  0)

  493. CAVS_SUBPIX(put_   , op_put2, qpel_l, -1, -2, 96, 42, -7,  0)

  494. CAVS_SUBPIX(put_   , op_put2, qpel_r,  0, -7, 42, 96, -2, -1)

  495. CAVS_SUBPIX_HV(put_, op_put3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)

  496. CAVS_SUBPIX_HV(put_, op_put4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)

  497. CAVS_SUBPIX_HV(put_, op_put4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)

  498. CAVS_SUBPIX_HV(put_, op_put4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)

  499. CAVS_SUBPIX_HV(put_, op_put4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)

  500. CAVS_SUBPIX_HV(put_, op_put2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)

  501. CAVS_SUBPIX(avg_   , op_avg1, hpel,    0, -1,  5,  5, -1,  0)

  502. CAVS_SUBPIX(avg_   , op_avg2, qpel_l, -1, -2, 96, 42, -7,  0)

  503. CAVS_SUBPIX(avg_   , op_avg2, qpel_r,  0, -7, 42, 96, -2, -1)

  504. CAVS_SUBPIX_HV(avg_, op_avg3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)

  505. CAVS_SUBPIX_HV(avg_, op_avg4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)

  506. CAVS_SUBPIX_HV(avg_, op_avg4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)

  507. CAVS_SUBPIX_HV(avg_, op_avg4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)

  508. CAVS_SUBPIX_HV(avg_, op_avg4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)

  509. CAVS_SUBPIX_HV(avg_, op_avg2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)

  510. CAVS_MC(put_, 8)

  511. CAVS_MC(put_, 16)

  512. CAVS_MC(avg_, 8)

  513. CAVS_MC(avg_, 16)

  514. 

  515. #define ff_put_cavs_qpel8_mc00_c  ff_put_pixels8x8_c

  516. #define ff_avg_cavs_qpel8_mc00_c  ff_avg_pixels8x8_c

  517. #define ff_put_cavs_qpel16_mc00_c ff_put_pixels16x16_c

  518. #define ff_avg_cavs_qpel16_mc00_c ff_avg_pixels16x16_c

  519. 

  520. av_cold void ff_cavsdsp_init(CAVSDSPContext* c, AVCodecContext *avctx) {

  521. #define dspfunc(PFX, IDX, NUM) \

  522.     c->PFX ## _pixels_tab[IDX][ 0] = ff_ ## PFX ## NUM ## _mc00_c; \

  523.     c->PFX ## _pixels_tab[IDX][ 1] = ff_ ## PFX ## NUM ## _mc10_c; \

  524.     c->PFX ## _pixels_tab[IDX][ 2] = ff_ ## PFX ## NUM ## _mc20_c; \

  525.     c->PFX ## _pixels_tab[IDX][ 3] = ff_ ## PFX ## NUM ## _mc30_c; \

  526.     c->PFX ## _pixels_tab[IDX][ 4] = ff_ ## PFX ## NUM ## _mc01_c; \

  527.     c->PFX ## _pixels_tab[IDX][ 5] = ff_ ## PFX ## NUM ## _mc11_c; \

  528.     c->PFX ## _pixels_tab[IDX][ 6] = ff_ ## PFX ## NUM ## _mc21_c; \

  529.     c->PFX ## _pixels_tab[IDX][ 7] = ff_ ## PFX ## NUM ## _mc31_c; \

  530.     c->PFX ## _pixels_tab[IDX][ 8] = ff_ ## PFX ## NUM ## _mc02_c; \

  531.     c->PFX ## _pixels_tab[IDX][ 9] = ff_ ## PFX ## NUM ## _mc12_c; \

  532.     c->PFX ## _pixels_tab[IDX][10] = ff_ ## PFX ## NUM ## _mc22_c; \

  533.     c->PFX ## _pixels_tab[IDX][11] = ff_ ## PFX ## NUM ## _mc32_c; \

  534.     c->PFX ## _pixels_tab[IDX][12] = ff_ ## PFX ## NUM ## _mc03_c; \

  535.     c->PFX ## _pixels_tab[IDX][13] = ff_ ## PFX ## NUM ## _mc13_c; \

  536.     c->PFX ## _pixels_tab[IDX][14] = ff_ ## PFX ## NUM ## _mc23_c; \

  537.     c->PFX ## _pixels_tab[IDX][15] = ff_ ## PFX ## NUM ## _mc33_c

  538.     dspfunc(put_cavs_qpel, 0, 16);

  539.     dspfunc(put_cavs_qpel, 1, 8);

  540.     dspfunc(avg_cavs_qpel, 0, 16);

  541.     dspfunc(avg_cavs_qpel, 1, 8);

  542.     c->cavs_filter_lv = cavs_filter_lv_c;

  543.     c->cavs_filter_lh = cavs_filter_lh_c;

  544.     c->cavs_filter_cv = cavs_filter_cv_c;

  545.     c->cavs_filter_ch = cavs_filter_ch_c;

  546.     c->cavs_idct8_add = cavs_idct8_add_c;

  547. 

  548.     if (HAVE_MMX) ff_cavsdsp_init_mmx(c, avctx);

  549. }