FFmpeg: libavcodec/cavsdsp.c Source File

00001 /*
00002  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.
00003  *
00004  * DSP functions
00005  *
00006  * Copyright (c) 2006  Stefan Gehrer <stefan.gehrer@gmx.de>
00007  *
00008  * This file is part of FFmpeg.
00009  *
00010  * FFmpeg is free software; you can redistribute it and/or
00011  * modify it under the terms of the GNU Lesser General Public
00012  * License as published by the Free Software Foundation; either
00013  * version 2.1 of the License, or (at your option) any later version.
00014  *
00015  * FFmpeg is distributed in the hope that it will be useful,
00016  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00017  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00018  * Lesser General Public License for more details.
00019  *
00020  * You should have received a copy of the GNU Lesser General Public
00021  * License along with FFmpeg; if not, write to the Free Software
00022  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00023  */
00024 
00025 #include <stdio.h>
00026 #include "dsputil.h"
00027 
00028 /*****************************************************************************
00029  *
00030  * in-loop deblocking filter
00031  *
00032  ****************************************************************************/
00033 
00034 #define P2 p0_p[-3*stride]
00035 #define P1 p0_p[-2*stride]
00036 #define P0 p0_p[-1*stride]
00037 #define Q0 p0_p[ 0*stride]
00038 #define Q1 p0_p[ 1*stride]
00039 #define Q2 p0_p[ 2*stride]
00040 
00041 static inline void loop_filter_l2(uint8_t *p0_p,int stride,int alpha, int beta) {
00042     int p0 = P0;
00043     int q0 = Q0;
00044 
00045     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
00046         int s = p0 + q0 + 2;
00047         alpha = (alpha>>2) + 2;
00048         if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
00049             P0 = (P1 + p0 + s) >> 2;
00050             P1 = (2*P1 + s) >> 2;
00051         } else
00052             P0 = (2*P1 + s) >> 2;
00053         if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
00054             Q0 = (Q1 + q0 + s) >> 2;
00055             Q1 = (2*Q1 + s) >> 2;
00056         } else
00057             Q0 = (2*Q1 + s) >> 2;
00058     }
00059 }
00060 
00061 static inline void loop_filter_l1(uint8_t *p0_p, int stride, int alpha, int beta, int tc) {
00062     int p0 = P0;
00063     int q0 = Q0;
00064 
00065     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
00066         int delta = av_clip(((q0-p0)*3+P1-Q1+4)>>3,-tc, tc);
00067         P0 = av_clip_uint8(p0+delta);
00068         Q0 = av_clip_uint8(q0-delta);
00069         if(abs(P2-p0)<beta) {
00070             delta = av_clip(((P0-P1)*3+P2-Q0+4)>>3, -tc, tc);
00071             P1 = av_clip_uint8(P1+delta);
00072         }
00073         if(abs(Q2-q0)<beta) {
00074             delta = av_clip(((Q1-Q0)*3+P0-Q2+4)>>3, -tc, tc);
00075             Q1 = av_clip_uint8(Q1-delta);
00076         }
00077     }
00078 }
00079 
00080 static inline void loop_filter_c2(uint8_t *p0_p,int stride,int alpha, int beta) {
00081     int p0 = P0;
00082     int q0 = Q0;
00083 
00084     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
00085         int s = p0 + q0 + 2;
00086         alpha = (alpha>>2) + 2;
00087         if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
00088             P0 = (P1 + p0 + s) >> 2;
00089         } else
00090             P0 = (2*P1 + s) >> 2;
00091         if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
00092             Q0 = (Q1 + q0 + s) >> 2;
00093         } else
00094             Q0 = (2*Q1 + s) >> 2;
00095     }
00096 }
00097 
00098 static inline void loop_filter_c1(uint8_t *p0_p,int stride,int alpha, int beta,
00099                                   int tc) {
00100     if(abs(P0-Q0)<alpha && abs(P1-P0)<beta && abs(Q1-Q0)<beta) {
00101         int delta = av_clip(((Q0-P0)*3+P1-Q1+4)>>3, -tc, tc);
00102         P0 = av_clip_uint8(P0+delta);
00103         Q0 = av_clip_uint8(Q0-delta);
00104     }
00105 }
00106 
00107 #undef P0
00108 #undef P1
00109 #undef P2
00110 #undef Q0
00111 #undef Q1
00112 #undef Q2
00113 
00114 static void cavs_filter_lv_c(uint8_t *d, int stride, int alpha, int beta, int tc,
00115                            int bs1, int bs2) {
00116     int i;
00117     if(bs1==2)
00118         for(i=0;i<16;i++)
00119             loop_filter_l2(d + i*stride,1,alpha,beta);
00120     else {
00121         if(bs1)
00122             for(i=0;i<8;i++)
00123                 loop_filter_l1(d + i*stride,1,alpha,beta,tc);
00124         if (bs2)
00125             for(i=8;i<16;i++)
00126                 loop_filter_l1(d + i*stride,1,alpha,beta,tc);
00127     }
00128 }
00129 
00130 static void cavs_filter_lh_c(uint8_t *d, int stride, int alpha, int beta, int tc,
00131                            int bs1, int bs2) {
00132     int i;
00133     if(bs1==2)
00134         for(i=0;i<16;i++)
00135             loop_filter_l2(d + i,stride,alpha,beta);
00136     else {
00137         if(bs1)
00138             for(i=0;i<8;i++)
00139                 loop_filter_l1(d + i,stride,alpha,beta,tc);
00140         if (bs2)
00141             for(i=8;i<16;i++)
00142                 loop_filter_l1(d + i,stride,alpha,beta,tc);
00143     }
00144 }
00145 
00146 static void cavs_filter_cv_c(uint8_t *d, int stride, int alpha, int beta, int tc,
00147                            int bs1, int bs2) {
00148     int i;
00149     if(bs1==2)
00150         for(i=0;i<8;i++)
00151             loop_filter_c2(d + i*stride,1,alpha,beta);
00152     else {
00153         if(bs1)
00154             for(i=0;i<4;i++)
00155                 loop_filter_c1(d + i*stride,1,alpha,beta,tc);
00156         if (bs2)
00157             for(i=4;i<8;i++)
00158                 loop_filter_c1(d + i*stride,1,alpha,beta,tc);
00159     }
00160 }
00161 
00162 static void cavs_filter_ch_c(uint8_t *d, int stride, int alpha, int beta, int tc,
00163                            int bs1, int bs2) {
00164     int i;
00165     if(bs1==2)
00166         for(i=0;i<8;i++)
00167             loop_filter_c2(d + i,stride,alpha,beta);
00168     else {
00169         if(bs1)
00170             for(i=0;i<4;i++)
00171                 loop_filter_c1(d + i,stride,alpha,beta,tc);
00172         if (bs2)
00173             for(i=4;i<8;i++)
00174                 loop_filter_c1(d + i,stride,alpha,beta,tc);
00175     }
00176 }
00177 
00178 /*****************************************************************************
00179  *
00180  * inverse transform
00181  *
00182  ****************************************************************************/
00183 
00184 static void cavs_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride) {
00185     int i;
00186     DCTELEM (*src)[8] = (DCTELEM(*)[8])block;
00187     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
00188 
00189     src[0][0] += 8;
00190 
00191     for( i = 0; i < 8; i++ ) {
00192         const int a0 =  3*src[i][1] - (src[i][7]<<1);
00193         const int a1 =  3*src[i][3] + (src[i][5]<<1);
00194         const int a2 =  (src[i][3]<<1) - 3*src[i][5];
00195         const int a3 =  (src[i][1]<<1) + 3*src[i][7];
00196 
00197         const int b4 = ((a0 + a1 + a3)<<1) + a1;
00198         const int b5 = ((a0 - a1 + a2)<<1) + a0;
00199         const int b6 = ((a3 - a2 - a1)<<1) + a3;
00200         const int b7 = ((a0 - a2 - a3)<<1) - a2;
00201 
00202         const int a7 = (src[i][2]<<2) - 10*src[i][6];
00203         const int a6 = (src[i][6]<<2) + 10*src[i][2];
00204         const int a5 = ((src[i][0] - src[i][4]) << 3) + 4;
00205         const int a4 = ((src[i][0] + src[i][4]) << 3) + 4;
00206 
00207         const int b0 = a4 + a6;
00208         const int b1 = a5 + a7;
00209         const int b2 = a5 - a7;
00210         const int b3 = a4 - a6;
00211 
00212         src[i][0] = (b0 + b4) >> 3;
00213         src[i][1] = (b1 + b5) >> 3;
00214         src[i][2] = (b2 + b6) >> 3;
00215         src[i][3] = (b3 + b7) >> 3;
00216         src[i][4] = (b3 - b7) >> 3;
00217         src[i][5] = (b2 - b6) >> 3;
00218         src[i][6] = (b1 - b5) >> 3;
00219         src[i][7] = (b0 - b4) >> 3;
00220     }
00221     for( i = 0; i < 8; i++ ) {
00222         const int a0 =  3*src[1][i] - (src[7][i]<<1);
00223         const int a1 =  3*src[3][i] + (src[5][i]<<1);
00224         const int a2 =  (src[3][i]<<1) - 3*src[5][i];
00225         const int a3 =  (src[1][i]<<1) + 3*src[7][i];
00226 
00227         const int b4 = ((a0 + a1 + a3)<<1) + a1;
00228         const int b5 = ((a0 - a1 + a2)<<1) + a0;
00229         const int b6 = ((a3 - a2 - a1)<<1) + a3;
00230         const int b7 = ((a0 - a2 - a3)<<1) - a2;
00231 
00232         const int a7 = (src[2][i]<<2) - 10*src[6][i];
00233         const int a6 = (src[6][i]<<2) + 10*src[2][i];
00234         const int a5 = (src[0][i] - src[4][i]) << 3;
00235         const int a4 = (src[0][i] + src[4][i]) << 3;
00236 
00237         const int b0 = a4 + a6;
00238         const int b1 = a5 + a7;
00239         const int b2 = a5 - a7;
00240         const int b3 = a4 - a6;
00241 
00242         dst[i + 0*stride] = cm[ dst[i + 0*stride] + ((b0 + b4) >> 7)];
00243         dst[i + 1*stride] = cm[ dst[i + 1*stride] + ((b1 + b5) >> 7)];
00244         dst[i + 2*stride] = cm[ dst[i + 2*stride] + ((b2 + b6) >> 7)];
00245         dst[i + 3*stride] = cm[ dst[i + 3*stride] + ((b3 + b7) >> 7)];
00246         dst[i + 4*stride] = cm[ dst[i + 4*stride] + ((b3 - b7) >> 7)];
00247         dst[i + 5*stride] = cm[ dst[i + 5*stride] + ((b2 - b6) >> 7)];
00248         dst[i + 6*stride] = cm[ dst[i + 6*stride] + ((b1 - b5) >> 7)];
00249         dst[i + 7*stride] = cm[ dst[i + 7*stride] + ((b0 - b4) >> 7)];
00250     }
00251 }
00252 
00253 /*****************************************************************************
00254  *
00255  * motion compensation
00256  *
00257  ****************************************************************************/
00258 
00259 #define CAVS_SUBPIX(OPNAME, OP, NAME, A, B, C, D, E, F) \
00260 static void OPNAME ## cavs_filt8_h_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00261     const int h=8;\
00262     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00263     int i;\
00264     for(i=0; i<h; i++)\
00265     {\
00266         OP(dst[0], A*src[-2] + B*src[-1] + C*src[0] + D*src[1] + E*src[2] + F*src[3]);\
00267         OP(dst[1], A*src[-1] + B*src[ 0] + C*src[1] + D*src[2] + E*src[3] + F*src[4]);\
00268         OP(dst[2], A*src[ 0] + B*src[ 1] + C*src[2] + D*src[3] + E*src[4] + F*src[5]);\
00269         OP(dst[3], A*src[ 1] + B*src[ 2] + C*src[3] + D*src[4] + E*src[5] + F*src[6]);\
00270         OP(dst[4], A*src[ 2] + B*src[ 3] + C*src[4] + D*src[5] + E*src[6] + F*src[7]);\
00271         OP(dst[5], A*src[ 3] + B*src[ 4] + C*src[5] + D*src[6] + E*src[7] + F*src[8]);\
00272         OP(dst[6], A*src[ 4] + B*src[ 5] + C*src[6] + D*src[7] + E*src[8] + F*src[9]);\
00273         OP(dst[7], A*src[ 5] + B*src[ 6] + C*src[7] + D*src[8] + E*src[9] + F*src[10]);\
00274         dst+=dstStride;\
00275         src+=srcStride;\
00276     }\
00277 }\
00278 \
00279 static void OPNAME ## cavs_filt8_v_  ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00280     const int w=8;\
00281     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00282     int i;\
00283     for(i=0; i<w; i++)\
00284     {\
00285         const int srcB= src[-2*srcStride];\
00286         const int srcA= src[-1*srcStride];\
00287         const int src0= src[0 *srcStride];\
00288         const int src1= src[1 *srcStride];\
00289         const int src2= src[2 *srcStride];\
00290         const int src3= src[3 *srcStride];\
00291         const int src4= src[4 *srcStride];\
00292         const int src5= src[5 *srcStride];\
00293         const int src6= src[6 *srcStride];\
00294         const int src7= src[7 *srcStride];\
00295         const int src8= src[8 *srcStride];\
00296         const int src9= src[9 *srcStride];\
00297         const int src10= src[10 *srcStride];\
00298         OP(dst[0*dstStride], A*srcB + B*srcA + C*src0 + D*src1 + E*src2 + F*src3);\
00299         OP(dst[1*dstStride], A*srcA + B*src0 + C*src1 + D*src2 + E*src3 + F*src4);\
00300         OP(dst[2*dstStride], A*src0 + B*src1 + C*src2 + D*src3 + E*src4 + F*src5);\
00301         OP(dst[3*dstStride], A*src1 + B*src2 + C*src3 + D*src4 + E*src5 + F*src6);\
00302         OP(dst[4*dstStride], A*src2 + B*src3 + C*src4 + D*src5 + E*src6 + F*src7);\
00303         OP(dst[5*dstStride], A*src3 + B*src4 + C*src5 + D*src6 + E*src7 + F*src8);\
00304         OP(dst[6*dstStride], A*src4 + B*src5 + C*src6 + D*src7 + E*src8 + F*src9);\
00305         OP(dst[7*dstStride], A*src5 + B*src6 + C*src7 + D*src8 + E*src9 + F*src10);\
00306         dst++;\
00307         src++;\
00308     }\
00309 }\
00310 \
00311 static void OPNAME ## cavs_filt16_v_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00312     OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\
00313     OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
00314     src += 8*srcStride;\
00315     dst += 8*dstStride;\
00316     OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\
00317     OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
00318 }\
00319 \
00320 static void OPNAME ## cavs_filt16_h_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00321     OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\
00322     OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
00323     src += 8*srcStride;\
00324     dst += 8*dstStride;\
00325     OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\
00326     OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
00327 }\
00328 
00329 #define CAVS_SUBPIX_HV(OPNAME, OP, NAME, AH, BH, CH, DH, EH, FH, AV, BV, CV, DV, EV, FV, FULL) \
00330 static void OPNAME ## cavs_filt8_hv_ ## NAME(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int srcStride){\
00331     int16_t temp[8*(8+5)];\
00332     int16_t *tmp = temp;\
00333     const int h=8;\
00334     const int w=8;\
00335     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00336     int i;\
00337     src1 -= 2*srcStride;\
00338     for(i=0; i<h+5; i++)\
00339     {\
00340         tmp[0]= AH*src1[-2] + BH*src1[-1] + CH*src1[0] + DH*src1[1] + EH*src1[2] + FH*src1[3];\
00341         tmp[1]= AH*src1[-1] + BH*src1[ 0] + CH*src1[1] + DH*src1[2] + EH*src1[3] + FH*src1[4];\
00342         tmp[2]= AH*src1[ 0] + BH*src1[ 1] + CH*src1[2] + DH*src1[3] + EH*src1[4] + FH*src1[5];\
00343         tmp[3]= AH*src1[ 1] + BH*src1[ 2] + CH*src1[3] + DH*src1[4] + EH*src1[5] + FH*src1[6];\
00344         tmp[4]= AH*src1[ 2] + BH*src1[ 3] + CH*src1[4] + DH*src1[5] + EH*src1[6] + FH*src1[7];\
00345         tmp[5]= AH*src1[ 3] + BH*src1[ 4] + CH*src1[5] + DH*src1[6] + EH*src1[7] + FH*src1[8];\
00346         tmp[6]= AH*src1[ 4] + BH*src1[ 5] + CH*src1[6] + DH*src1[7] + EH*src1[8] + FH*src1[9];\
00347         tmp[7]= AH*src1[ 5] + BH*src1[ 6] + CH*src1[7] + DH*src1[8] + EH*src1[9] + FH*src1[10];\
00348         tmp+=8;\
00349         src1+=srcStride;\
00350     }\
00351     if(FULL) {\
00352       tmp = temp+8*2;                           \
00353       for(i=0; i<w; i++)                        \
00354         {                                       \
00355           const int tmpB= tmp[-2*8];    \
00356           const int tmpA= tmp[-1*8];    \
00357           const int tmp0= tmp[0 *8];    \
00358           const int tmp1= tmp[1 *8];    \
00359           const int tmp2= tmp[2 *8];    \
00360           const int tmp3= tmp[3 *8];    \
00361           const int tmp4= tmp[4 *8];    \
00362           const int tmp5= tmp[5 *8];    \
00363           const int tmp6= tmp[6 *8];    \
00364           const int tmp7= tmp[7 *8];    \
00365           const int tmp8= tmp[8 *8];    \
00366           const int tmp9= tmp[9 *8];    \
00367           const int tmp10=tmp[10*8];                            \
00368           OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3 + 64*src2[0*srcStride]); \
00369           OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4 + 64*src2[1*srcStride]); \
00370           OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5 + 64*src2[2*srcStride]); \
00371           OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6 + 64*src2[3*srcStride]); \
00372           OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7 + 64*src2[4*srcStride]); \
00373           OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8 + 64*src2[5*srcStride]); \
00374           OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9 + 64*src2[6*srcStride]); \
00375           OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10 + 64*src2[7*srcStride]); \
00376           dst++;                                                        \
00377           tmp++;                                                        \
00378           src2++;                                                       \
00379         }                                                               \
00380     } else {\
00381       tmp = temp+8*2;                           \
00382       for(i=0; i<w; i++)                        \
00383         {                                       \
00384           const int tmpB= tmp[-2*8];    \
00385           const int tmpA= tmp[-1*8];    \
00386           const int tmp0= tmp[0 *8];    \
00387           const int tmp1= tmp[1 *8];    \
00388           const int tmp2= tmp[2 *8];    \
00389           const int tmp3= tmp[3 *8];    \
00390           const int tmp4= tmp[4 *8];    \
00391           const int tmp5= tmp[5 *8];    \
00392           const int tmp6= tmp[6 *8];    \
00393           const int tmp7= tmp[7 *8];    \
00394           const int tmp8= tmp[8 *8];    \
00395           const int tmp9= tmp[9 *8];    \
00396           const int tmp10=tmp[10*8];                            \
00397           OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3); \
00398           OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4); \
00399           OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5); \
00400           OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6); \
00401           OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7); \
00402           OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8); \
00403           OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9); \
00404           OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10); \
00405           dst++;                                                        \
00406           tmp++;                                                        \
00407         }                                                               \
00408     }\
00409 }\
00410 \
00411 static void OPNAME ## cavs_filt16_hv_ ## NAME(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int srcStride){ \
00412     OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \
00413     OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \
00414     src1 += 8*srcStride;\
00415     src2 += 8*srcStride;\
00416     dst += 8*dstStride;\
00417     OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \
00418     OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \
00419 }\
00420 
00421 #define CAVS_MC(OPNAME, SIZE) \
00422 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
00423     OPNAME ## cavs_filt ## SIZE ## _h_qpel_l(dst, src, stride, stride);\
00424 }\
00425 \
00426 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\
00427     OPNAME ## cavs_filt ## SIZE ## _h_hpel(dst, src, stride, stride);\
00428 }\
00429 \
00430 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc30_c(uint8_t *dst, uint8_t *src, int stride){\
00431     OPNAME ## cavs_filt ## SIZE ## _h_qpel_r(dst, src, stride, stride);\
00432 }\
00433 \
00434 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
00435     OPNAME ## cavs_filt ## SIZE ## _v_qpel_l(dst, src, stride, stride);\
00436 }\
00437 \
00438 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\
00439     OPNAME ## cavs_filt ## SIZE ## _v_hpel(dst, src, stride, stride);\
00440 }\
00441 \
00442 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc03_c(uint8_t *dst, uint8_t *src, int stride){\
00443     OPNAME ## cavs_filt ## SIZE ## _v_qpel_r(dst, src, stride, stride);\
00444 }\
00445 \
00446 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
00447   OPNAME ## cavs_filt ## SIZE ## _hv_jj(dst, src, NULL, stride, stride); \
00448 }\
00449 \
00450 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
00451   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src, stride, stride); \
00452 }\
00453 \
00454 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc13_c(uint8_t *dst, uint8_t *src, int stride){\
00455   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride, stride, stride); \
00456 }\
00457 \
00458 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc31_c(uint8_t *dst, uint8_t *src, int stride){\
00459   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+1, stride, stride); \
00460 }\
00461 \
00462 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc33_c(uint8_t *dst, uint8_t *src, int stride){\
00463   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride+1,stride, stride); \
00464 }\
00465 \
00466 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
00467   OPNAME ## cavs_filt ## SIZE ## _hv_ff(dst, src, src+stride+1,stride, stride); \
00468 }\
00469 \
00470 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
00471   OPNAME ## cavs_filt ## SIZE ## _hv_ii(dst, src, src+stride+1,stride, stride); \
00472 }\
00473 \
00474 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, int stride){\
00475   OPNAME ## cavs_filt ## SIZE ## _hv_kk(dst, src, src+stride+1,stride, stride); \
00476 }\
00477 \
00478 static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc23_c(uint8_t *dst, uint8_t *src, int stride){\
00479   OPNAME ## cavs_filt ## SIZE ## _hv_qq(dst, src, src+stride+1,stride, stride); \
00480 }\
00481 
00482 #define op_put1(a, b)  a = cm[((b)+4)>>3]
00483 #define op_put2(a, b)  a = cm[((b)+64)>>7]
00484 #define op_put3(a, b)  a = cm[((b)+32)>>6]
00485 #define op_put4(a, b)  a = cm[((b)+512)>>10]
00486 #define op_avg1(a, b)  a = ((a)+cm[((b)+4)>>3]   +1)>>1
00487 #define op_avg2(a, b)  a = ((a)+cm[((b)+64)>>7]  +1)>>1
00488 #define op_avg3(a, b)  a = ((a)+cm[((b)+32)>>6]  +1)>>1
00489 #define op_avg4(a, b)  a = ((a)+cm[((b)+512)>>10]+1)>>1
00490 CAVS_SUBPIX(put_   , op_put1, hpel,    0, -1,  5,  5, -1,  0)
00491 CAVS_SUBPIX(put_   , op_put2, qpel_l, -1, -2, 96, 42, -7,  0)
00492 CAVS_SUBPIX(put_   , op_put2, qpel_r,  0, -7, 42, 96, -2, -1)
00493 CAVS_SUBPIX_HV(put_, op_put3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)
00494 CAVS_SUBPIX_HV(put_, op_put4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)
00495 CAVS_SUBPIX_HV(put_, op_put4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)
00496 CAVS_SUBPIX_HV(put_, op_put4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)
00497 CAVS_SUBPIX_HV(put_, op_put4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)
00498 CAVS_SUBPIX_HV(put_, op_put2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)
00499 CAVS_SUBPIX(avg_   , op_avg1, hpel,    0, -1,  5,  5, -1,  0)
00500 CAVS_SUBPIX(avg_   , op_avg2, qpel_l, -1, -2, 96, 42, -7,  0)
00501 CAVS_SUBPIX(avg_   , op_avg2, qpel_r,  0, -7, 42, 96, -2, -1)
00502 CAVS_SUBPIX_HV(avg_, op_avg3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)
00503 CAVS_SUBPIX_HV(avg_, op_avg4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)
00504 CAVS_SUBPIX_HV(avg_, op_avg4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)
00505 CAVS_SUBPIX_HV(avg_, op_avg4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)
00506 CAVS_SUBPIX_HV(avg_, op_avg4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)
00507 CAVS_SUBPIX_HV(avg_, op_avg2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)
00508 CAVS_MC(put_, 8)
00509 CAVS_MC(put_, 16)
00510 CAVS_MC(avg_, 8)
00511 CAVS_MC(avg_, 16)
00512 
00513 void ff_put_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride);
00514 void ff_avg_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride);
00515 void ff_put_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride);
00516 void ff_avg_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride);
00517 
00518 void ff_cavsdsp_init(DSPContext* c, AVCodecContext *avctx) {
00519 #define dspfunc(PFX, IDX, NUM) \
00520     c->PFX ## _pixels_tab[IDX][ 0] = ff_ ## PFX ## NUM ## _mc00_c; \
00521     c->PFX ## _pixels_tab[IDX][ 1] = ff_ ## PFX ## NUM ## _mc10_c; \
00522     c->PFX ## _pixels_tab[IDX][ 2] = ff_ ## PFX ## NUM ## _mc20_c; \
00523     c->PFX ## _pixels_tab[IDX][ 3] = ff_ ## PFX ## NUM ## _mc30_c; \
00524     c->PFX ## _pixels_tab[IDX][ 4] = ff_ ## PFX ## NUM ## _mc01_c; \
00525     c->PFX ## _pixels_tab[IDX][ 5] = ff_ ## PFX ## NUM ## _mc11_c; \
00526     c->PFX ## _pixels_tab[IDX][ 6] = ff_ ## PFX ## NUM ## _mc21_c; \
00527     c->PFX ## _pixels_tab[IDX][ 7] = ff_ ## PFX ## NUM ## _mc31_c; \
00528     c->PFX ## _pixels_tab[IDX][ 8] = ff_ ## PFX ## NUM ## _mc02_c; \
00529     c->PFX ## _pixels_tab[IDX][ 9] = ff_ ## PFX ## NUM ## _mc12_c; \
00530     c->PFX ## _pixels_tab[IDX][10] = ff_ ## PFX ## NUM ## _mc22_c; \
00531     c->PFX ## _pixels_tab[IDX][11] = ff_ ## PFX ## NUM ## _mc32_c; \
00532     c->PFX ## _pixels_tab[IDX][12] = ff_ ## PFX ## NUM ## _mc03_c; \
00533     c->PFX ## _pixels_tab[IDX][13] = ff_ ## PFX ## NUM ## _mc13_c; \
00534     c->PFX ## _pixels_tab[IDX][14] = ff_ ## PFX ## NUM ## _mc23_c; \
00535     c->PFX ## _pixels_tab[IDX][15] = ff_ ## PFX ## NUM ## _mc33_c
00536     dspfunc(put_cavs_qpel, 0, 16);
00537     dspfunc(put_cavs_qpel, 1, 8);
00538     dspfunc(avg_cavs_qpel, 0, 16);
00539     dspfunc(avg_cavs_qpel, 1, 8);
00540     c->cavs_filter_lv = cavs_filter_lv_c;
00541     c->cavs_filter_lh = cavs_filter_lh_c;
00542     c->cavs_filter_cv = cavs_filter_cv_c;
00543     c->cavs_filter_ch = cavs_filter_ch_c;
00544     c->cavs_idct8_add = cavs_idct8_add_c;
00545 }