yading@10: /*
yading@10:  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.
yading@10:  *
yading@10:  * DSP functions
yading@10:  *
yading@10:  * Copyright (c) 2006  Stefan Gehrer <stefan.gehrer@gmx.de>
yading@10:  *
yading@10:  * This file is part of FFmpeg.
yading@10:  *
yading@10:  * FFmpeg is free software; you can redistribute it and/or
yading@10:  * modify it under the terms of the GNU Lesser General Public
yading@10:  * License as published by the Free Software Foundation; either
yading@10:  * version 2.1 of the License, or (at your option) any later version.
yading@10:  *
yading@10:  * FFmpeg is distributed in the hope that it will be useful,
yading@10:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
yading@10:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
yading@10:  * Lesser General Public License for more details.
yading@10:  *
yading@10:  * You should have received a copy of the GNU Lesser General Public
yading@10:  * License along with FFmpeg; if not, write to the Free Software
yading@10:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
yading@10:  */
yading@10: 
yading@10: #include <stdio.h>
yading@10: #include "dsputil.h"
yading@10: #include "cavsdsp.h"
yading@10: #include "libavutil/common.h"
yading@10: 
yading@10: /*****************************************************************************
yading@10:  *
yading@10:  * in-loop deblocking filter
yading@10:  *
yading@10:  ****************************************************************************/
yading@10: 
yading@10: #define P2 p0_p[-3*stride]
yading@10: #define P1 p0_p[-2*stride]
yading@10: #define P0 p0_p[-1*stride]
yading@10: #define Q0 p0_p[ 0*stride]
yading@10: #define Q1 p0_p[ 1*stride]
yading@10: #define Q2 p0_p[ 2*stride]
yading@10: 
yading@10: static inline void loop_filter_l2(uint8_t *p0_p,int stride,int alpha, int beta) {
yading@10:     int p0 = P0;
yading@10:     int q0 = Q0;
yading@10: 
yading@10:     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
yading@10:         int s = p0 + q0 + 2;
yading@10:         alpha = (alpha>>2) + 2;
yading@10:         if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
yading@10:             P0 = (P1 + p0 + s) >> 2;
yading@10:             P1 = (2*P1 + s) >> 2;
yading@10:         } else
yading@10:             P0 = (2*P1 + s) >> 2;
yading@10:         if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
yading@10:             Q0 = (Q1 + q0 + s) >> 2;
yading@10:             Q1 = (2*Q1 + s) >> 2;
yading@10:         } else
yading@10:             Q0 = (2*Q1 + s) >> 2;
yading@10:     }
yading@10: }
yading@10: 
yading@10: static inline void loop_filter_l1(uint8_t *p0_p, int stride, int alpha, int beta, int tc) {
yading@10:     int p0 = P0;
yading@10:     int q0 = Q0;
yading@10: 
yading@10:     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
yading@10:         int delta = av_clip(((q0-p0)*3+P1-Q1+4)>>3,-tc, tc);
yading@10:         P0 = av_clip_uint8(p0+delta);
yading@10:         Q0 = av_clip_uint8(q0-delta);
yading@10:         if(abs(P2-p0)<beta) {
yading@10:             delta = av_clip(((P0-P1)*3+P2-Q0+4)>>3, -tc, tc);
yading@10:             P1 = av_clip_uint8(P1+delta);
yading@10:         }
yading@10:         if(abs(Q2-q0)<beta) {
yading@10:             delta = av_clip(((Q1-Q0)*3+P0-Q2+4)>>3, -tc, tc);
yading@10:             Q1 = av_clip_uint8(Q1-delta);
yading@10:         }
yading@10:     }
yading@10: }
yading@10: 
yading@10: static inline void loop_filter_c2(uint8_t *p0_p,int stride,int alpha, int beta) {
yading@10:     int p0 = P0;
yading@10:     int q0 = Q0;
yading@10: 
yading@10:     if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
yading@10:         int s = p0 + q0 + 2;
yading@10:         alpha = (alpha>>2) + 2;
yading@10:         if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
yading@10:             P0 = (P1 + p0 + s) >> 2;
yading@10:         } else
yading@10:             P0 = (2*P1 + s) >> 2;
yading@10:         if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
yading@10:             Q0 = (Q1 + q0 + s) >> 2;
yading@10:         } else
yading@10:             Q0 = (2*Q1 + s) >> 2;
yading@10:     }
yading@10: }
yading@10: 
yading@10: static inline void loop_filter_c1(uint8_t *p0_p,int stride,int alpha, int beta,
yading@10:                                   int tc) {
yading@10:     if(abs(P0-Q0)<alpha && abs(P1-P0)<beta && abs(Q1-Q0)<beta) {
yading@10:         int delta = av_clip(((Q0-P0)*3+P1-Q1+4)>>3, -tc, tc);
yading@10:         P0 = av_clip_uint8(P0+delta);
yading@10:         Q0 = av_clip_uint8(Q0-delta);
yading@10:     }
yading@10: }
yading@10: 
yading@10: #undef P0
yading@10: #undef P1
yading@10: #undef P2
yading@10: #undef Q0
yading@10: #undef Q1
yading@10: #undef Q2
yading@10: 
yading@10: static void cavs_filter_lv_c(uint8_t *d, int stride, int alpha, int beta, int tc,
yading@10:                            int bs1, int bs2) {
yading@10:     int i;
yading@10:     if(bs1==2)
yading@10:         for(i=0;i<16;i++)
yading@10:             loop_filter_l2(d + i*stride,1,alpha,beta);
yading@10:     else {
yading@10:         if(bs1)
yading@10:             for(i=0;i<8;i++)
yading@10:                 loop_filter_l1(d + i*stride,1,alpha,beta,tc);
yading@10:         if (bs2)
yading@10:             for(i=8;i<16;i++)
yading@10:                 loop_filter_l1(d + i*stride,1,alpha,beta,tc);
yading@10:     }
yading@10: }
yading@10: 
yading@10: static void cavs_filter_lh_c(uint8_t *d, int stride, int alpha, int beta, int tc,
yading@10:                            int bs1, int bs2) {
yading@10:     int i;
yading@10:     if(bs1==2)
yading@10:         for(i=0;i<16;i++)
yading@10:             loop_filter_l2(d + i,stride,alpha,beta);
yading@10:     else {
yading@10:         if(bs1)
yading@10:             for(i=0;i<8;i++)
yading@10:                 loop_filter_l1(d + i,stride,alpha,beta,tc);
yading@10:         if (bs2)
yading@10:             for(i=8;i<16;i++)
yading@10:                 loop_filter_l1(d + i,stride,alpha,beta,tc);
yading@10:     }
yading@10: }
yading@10: 
yading@10: static void cavs_filter_cv_c(uint8_t *d, int stride, int alpha, int beta, int tc,
yading@10:                            int bs1, int bs2) {
yading@10:     int i;
yading@10:     if(bs1==2)
yading@10:         for(i=0;i<8;i++)
yading@10:             loop_filter_c2(d + i*stride,1,alpha,beta);
yading@10:     else {
yading@10:         if(bs1)
yading@10:             for(i=0;i<4;i++)
yading@10:                 loop_filter_c1(d + i*stride,1,alpha,beta,tc);
yading@10:         if (bs2)
yading@10:             for(i=4;i<8;i++)
yading@10:                 loop_filter_c1(d + i*stride,1,alpha,beta,tc);
yading@10:     }
yading@10: }
yading@10: 
yading@10: static void cavs_filter_ch_c(uint8_t *d, int stride, int alpha, int beta, int tc,
yading@10:                            int bs1, int bs2) {
yading@10:     int i;
yading@10:     if(bs1==2)
yading@10:         for(i=0;i<8;i++)
yading@10:             loop_filter_c2(d + i,stride,alpha,beta);
yading@10:     else {
yading@10:         if(bs1)
yading@10:             for(i=0;i<4;i++)
yading@10:                 loop_filter_c1(d + i,stride,alpha,beta,tc);
yading@10:         if (bs2)
yading@10:             for(i=4;i<8;i++)
yading@10:                 loop_filter_c1(d + i,stride,alpha,beta,tc);
yading@10:     }
yading@10: }
yading@10: 
yading@10: /*****************************************************************************
yading@10:  *
yading@10:  * inverse transform
yading@10:  *
yading@10:  ****************************************************************************/
yading@10: 
yading@10: static void cavs_idct8_add_c(uint8_t *dst, int16_t *block, int stride) {
yading@10:     int i;
yading@10:     int16_t (*src)[8] = (int16_t(*)[8])block;
yading@10:     const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
yading@10: 
yading@10:     src[0][0] += 8;
yading@10: 
yading@10:     for( i = 0; i < 8; i++ ) {
yading@10:         const int a0 =  3*src[i][1] - (src[i][7]<<1);
yading@10:         const int a1 =  3*src[i][3] + (src[i][5]<<1);
yading@10:         const int a2 =  (src[i][3]<<1) - 3*src[i][5];
yading@10:         const int a3 =  (src[i][1]<<1) + 3*src[i][7];
yading@10: 
yading@10:         const int b4 = ((a0 + a1 + a3)<<1) + a1;
yading@10:         const int b5 = ((a0 - a1 + a2)<<1) + a0;
yading@10:         const int b6 = ((a3 - a2 - a1)<<1) + a3;
yading@10:         const int b7 = ((a0 - a2 - a3)<<1) - a2;
yading@10: 
yading@10:         const int a7 = (src[i][2]<<2) - 10*src[i][6];
yading@10:         const int a6 = (src[i][6]<<2) + 10*src[i][2];
yading@10:         const int a5 = ((src[i][0] - src[i][4]) << 3) + 4;
yading@10:         const int a4 = ((src[i][0] + src[i][4]) << 3) + 4;
yading@10: 
yading@10:         const int b0 = a4 + a6;
yading@10:         const int b1 = a5 + a7;
yading@10:         const int b2 = a5 - a7;
yading@10:         const int b3 = a4 - a6;
yading@10: 
yading@10:         src[i][0] = (b0 + b4) >> 3;
yading@10:         src[i][1] = (b1 + b5) >> 3;
yading@10:         src[i][2] = (b2 + b6) >> 3;
yading@10:         src[i][3] = (b3 + b7) >> 3;
yading@10:         src[i][4] = (b3 - b7) >> 3;
yading@10:         src[i][5] = (b2 - b6) >> 3;
yading@10:         src[i][6] = (b1 - b5) >> 3;
yading@10:         src[i][7] = (b0 - b4) >> 3;
yading@10:     }
yading@10:     for( i = 0; i < 8; i++ ) {
yading@10:         const int a0 =  3*src[1][i] - (src[7][i]<<1);
yading@10:         const int a1 =  3*src[3][i] + (src[5][i]<<1);
yading@10:         const int a2 =  (src[3][i]<<1) - 3*src[5][i];
yading@10:         const int a3 =  (src[1][i]<<1) + 3*src[7][i];
yading@10: 
yading@10:         const int b4 = ((a0 + a1 + a3)<<1) + a1;
yading@10:         const int b5 = ((a0 - a1 + a2)<<1) + a0;
yading@10:         const int b6 = ((a3 - a2 - a1)<<1) + a3;
yading@10:         const int b7 = ((a0 - a2 - a3)<<1) - a2;
yading@10: 
yading@10:         const int a7 = (src[2][i]<<2) - 10*src[6][i];
yading@10:         const int a6 = (src[6][i]<<2) + 10*src[2][i];
yading@10:         const int a5 = (src[0][i] - src[4][i]) << 3;
yading@10:         const int a4 = (src[0][i] + src[4][i]) << 3;
yading@10: 
yading@10:         const int b0 = a4 + a6;
yading@10:         const int b1 = a5 + a7;
yading@10:         const int b2 = a5 - a7;
yading@10:         const int b3 = a4 - a6;
yading@10: 
yading@10:         dst[i + 0*stride] = cm[ dst[i + 0*stride] + ((b0 + b4) >> 7)];
yading@10:         dst[i + 1*stride] = cm[ dst[i + 1*stride] + ((b1 + b5) >> 7)];
yading@10:         dst[i + 2*stride] = cm[ dst[i + 2*stride] + ((b2 + b6) >> 7)];
yading@10:         dst[i + 3*stride] = cm[ dst[i + 3*stride] + ((b3 + b7) >> 7)];
yading@10:         dst[i + 4*stride] = cm[ dst[i + 4*stride] + ((b3 - b7) >> 7)];
yading@10:         dst[i + 5*stride] = cm[ dst[i + 5*stride] + ((b2 - b6) >> 7)];
yading@10:         dst[i + 6*stride] = cm[ dst[i + 6*stride] + ((b1 - b5) >> 7)];
yading@10:         dst[i + 7*stride] = cm[ dst[i + 7*stride] + ((b0 - b4) >> 7)];
yading@10:     }
yading@10: }
yading@10: 
yading@10: /*****************************************************************************
yading@10:  *
yading@10:  * motion compensation
yading@10:  *
yading@10:  ****************************************************************************/
yading@10: 
yading@10: #define CAVS_SUBPIX(OPNAME, OP, NAME, A, B, C, D, E, F) \
yading@10: static void OPNAME ## cavs_filt8_h_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
yading@10:     const int h=8;\
yading@10:     const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
yading@10:     int i;\
yading@10:     for(i=0; i<h; i++)\
yading@10:     {\
yading@10:         OP(dst[0], A*src[-2] + B*src[-1] + C*src[0] + D*src[1] + E*src[2] + F*src[3]);\
yading@10:         OP(dst[1], A*src[-1] + B*src[ 0] + C*src[1] + D*src[2] + E*src[3] + F*src[4]);\
yading@10:         OP(dst[2], A*src[ 0] + B*src[ 1] + C*src[2] + D*src[3] + E*src[4] + F*src[5]);\
yading@10:         OP(dst[3], A*src[ 1] + B*src[ 2] + C*src[3] + D*src[4] + E*src[5] + F*src[6]);\
yading@10:         OP(dst[4], A*src[ 2] + B*src[ 3] + C*src[4] + D*src[5] + E*src[6] + F*src[7]);\
yading@10:         OP(dst[5], A*src[ 3] + B*src[ 4] + C*src[5] + D*src[6] + E*src[7] + F*src[8]);\
yading@10:         OP(dst[6], A*src[ 4] + B*src[ 5] + C*src[6] + D*src[7] + E*src[8] + F*src[9]);\
yading@10:         OP(dst[7], A*src[ 5] + B*src[ 6] + C*src[7] + D*src[8] + E*src[9] + F*src[10]);\
yading@10:         dst+=dstStride;\
yading@10:         src+=srcStride;\
yading@10:     }\
yading@10: }\
yading@10: \
yading@10: static void OPNAME ## cavs_filt8_v_  ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
yading@10:     const int w=8;\
yading@10:     const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
yading@10:     int i;\
yading@10:     for(i=0; i<w; i++)\
yading@10:     {\
yading@10:         const int srcB= src[-2*srcStride];\
yading@10:         const int srcA= src[-1*srcStride];\
yading@10:         const int src0= src[0 *srcStride];\
yading@10:         const int src1= src[1 *srcStride];\
yading@10:         const int src2= src[2 *srcStride];\
yading@10:         const int src3= src[3 *srcStride];\
yading@10:         const int src4= src[4 *srcStride];\
yading@10:         const int src5= src[5 *srcStride];\
yading@10:         const int src6= src[6 *srcStride];\
yading@10:         const int src7= src[7 *srcStride];\
yading@10:         const int src8= src[8 *srcStride];\
yading@10:         const int src9= src[9 *srcStride];\
yading@10:         const int src10= src[10 *srcStride];\
yading@10:         OP(dst[0*dstStride], A*srcB + B*srcA + C*src0 + D*src1 + E*src2 + F*src3);\
yading@10:         OP(dst[1*dstStride], A*srcA + B*src0 + C*src1 + D*src2 + E*src3 + F*src4);\
yading@10:         OP(dst[2*dstStride], A*src0 + B*src1 + C*src2 + D*src3 + E*src4 + F*src5);\
yading@10:         OP(dst[3*dstStride], A*src1 + B*src2 + C*src3 + D*src4 + E*src5 + F*src6);\
yading@10:         OP(dst[4*dstStride], A*src2 + B*src3 + C*src4 + D*src5 + E*src6 + F*src7);\
yading@10:         OP(dst[5*dstStride], A*src3 + B*src4 + C*src5 + D*src6 + E*src7 + F*src8);\
yading@10:         OP(dst[6*dstStride], A*src4 + B*src5 + C*src6 + D*src7 + E*src8 + F*src9);\
yading@10:         OP(dst[7*dstStride], A*src5 + B*src6 + C*src7 + D*src8 + E*src9 + F*src10);\
yading@10:         dst++;\
yading@10:         src++;\
yading@10:     }\
yading@10: }\
yading@10: \
yading@10: static void OPNAME ## cavs_filt16_v_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
yading@10:     OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\
yading@10:     OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
yading@10:     src += 8*srcStride;\
yading@10:     dst += 8*dstStride;\
yading@10:     OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\
yading@10:     OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
yading@10: }\
yading@10: \
yading@10: static void OPNAME ## cavs_filt16_h_ ## NAME(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
yading@10:     OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\
yading@10:     OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
yading@10:     src += 8*srcStride;\
yading@10:     dst += 8*dstStride;\
yading@10:     OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\
yading@10:     OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
yading@10: }\
yading@10: 
yading@10: #define CAVS_SUBPIX_HV(OPNAME, OP, NAME, AH, BH, CH, DH, EH, FH, AV, BV, CV, DV, EV, FV, FULL) \
yading@10: static void OPNAME ## cavs_filt8_hv_ ## NAME(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int srcStride){\
yading@10:     int16_t temp[8*(8+5)];\
yading@10:     int16_t *tmp = temp;\
yading@10:     const int h=8;\
yading@10:     const int w=8;\
yading@10:     const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
yading@10:     int i;\
yading@10:     src1 -= 2*srcStride;\
yading@10:     for(i=0; i<h+5; i++)\
yading@10:     {\
yading@10:         tmp[0]= AH*src1[-2] + BH*src1[-1] + CH*src1[0] + DH*src1[1] + EH*src1[2] + FH*src1[3];\
yading@10:         tmp[1]= AH*src1[-1] + BH*src1[ 0] + CH*src1[1] + DH*src1[2] + EH*src1[3] + FH*src1[4];\
yading@10:         tmp[2]= AH*src1[ 0] + BH*src1[ 1] + CH*src1[2] + DH*src1[3] + EH*src1[4] + FH*src1[5];\
yading@10:         tmp[3]= AH*src1[ 1] + BH*src1[ 2] + CH*src1[3] + DH*src1[4] + EH*src1[5] + FH*src1[6];\
yading@10:         tmp[4]= AH*src1[ 2] + BH*src1[ 3] + CH*src1[4] + DH*src1[5] + EH*src1[6] + FH*src1[7];\
yading@10:         tmp[5]= AH*src1[ 3] + BH*src1[ 4] + CH*src1[5] + DH*src1[6] + EH*src1[7] + FH*src1[8];\
yading@10:         tmp[6]= AH*src1[ 4] + BH*src1[ 5] + CH*src1[6] + DH*src1[7] + EH*src1[8] + FH*src1[9];\
yading@10:         tmp[7]= AH*src1[ 5] + BH*src1[ 6] + CH*src1[7] + DH*src1[8] + EH*src1[9] + FH*src1[10];\
yading@10:         tmp+=8;\
yading@10:         src1+=srcStride;\
yading@10:     }\
yading@10:     if(FULL) {\
yading@10:       tmp = temp+8*2;                           \
yading@10:       for(i=0; i<w; i++)                        \
yading@10:         {                                       \
yading@10:           const int tmpB= tmp[-2*8];    \
yading@10:           const int tmpA= tmp[-1*8];    \
yading@10:           const int tmp0= tmp[0 *8];    \
yading@10:           const int tmp1= tmp[1 *8];    \
yading@10:           const int tmp2= tmp[2 *8];    \
yading@10:           const int tmp3= tmp[3 *8];    \
yading@10:           const int tmp4= tmp[4 *8];    \
yading@10:           const int tmp5= tmp[5 *8];    \
yading@10:           const int tmp6= tmp[6 *8];    \
yading@10:           const int tmp7= tmp[7 *8];    \
yading@10:           const int tmp8= tmp[8 *8];    \
yading@10:           const int tmp9= tmp[9 *8];    \
yading@10:           const int tmp10=tmp[10*8];                            \
yading@10:           OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3 + 64*src2[0*srcStride]); \
yading@10:           OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4 + 64*src2[1*srcStride]); \
yading@10:           OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5 + 64*src2[2*srcStride]); \
yading@10:           OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6 + 64*src2[3*srcStride]); \
yading@10:           OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7 + 64*src2[4*srcStride]); \
yading@10:           OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8 + 64*src2[5*srcStride]); \
yading@10:           OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9 + 64*src2[6*srcStride]); \
yading@10:           OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10 + 64*src2[7*srcStride]); \
yading@10:           dst++;                                                        \
yading@10:           tmp++;                                                        \
yading@10:           src2++;                                                       \
yading@10:         }                                                               \
yading@10:     } else {\
yading@10:       tmp = temp+8*2;                           \
yading@10:       for(i=0; i<w; i++)                        \
yading@10:         {                                       \
yading@10:           const int tmpB= tmp[-2*8];    \
yading@10:           const int tmpA= tmp[-1*8];    \
yading@10:           const int tmp0= tmp[0 *8];    \
yading@10:           const int tmp1= tmp[1 *8];    \
yading@10:           const int tmp2= tmp[2 *8];    \
yading@10:           const int tmp3= tmp[3 *8];    \
yading@10:           const int tmp4= tmp[4 *8];    \
yading@10:           const int tmp5= tmp[5 *8];    \
yading@10:           const int tmp6= tmp[6 *8];    \
yading@10:           const int tmp7= tmp[7 *8];    \
yading@10:           const int tmp8= tmp[8 *8];    \
yading@10:           const int tmp9= tmp[9 *8];    \
yading@10:           const int tmp10=tmp[10*8];                            \
yading@10:           OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3); \
yading@10:           OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4); \
yading@10:           OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5); \
yading@10:           OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6); \
yading@10:           OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7); \
yading@10:           OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8); \
yading@10:           OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9); \
yading@10:           OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10); \
yading@10:           dst++;                                                        \
yading@10:           tmp++;                                                        \
yading@10:         }                                                               \
yading@10:     }\
yading@10: }\
yading@10: \
yading@10: static void OPNAME ## cavs_filt16_hv_ ## NAME(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int srcStride){ \
yading@10:     OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \
yading@10:     OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \
yading@10:     src1 += 8*srcStride;\
yading@10:     src2 += 8*srcStride;\
yading@10:     dst += 8*dstStride;\
yading@10:     OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \
yading@10:     OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \
yading@10: }\
yading@10: 
yading@10: #define CAVS_MC(OPNAME, SIZE) \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:     OPNAME ## cavs_filt ## SIZE ## _h_qpel_l(dst, src, stride, stride);\
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:     OPNAME ## cavs_filt ## SIZE ## _h_hpel(dst, src, stride, stride);\
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc30_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:     OPNAME ## cavs_filt ## SIZE ## _h_qpel_r(dst, src, stride, stride);\
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:     OPNAME ## cavs_filt ## SIZE ## _v_qpel_l(dst, src, stride, stride);\
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:     OPNAME ## cavs_filt ## SIZE ## _v_hpel(dst, src, stride, stride);\
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc03_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:     OPNAME ## cavs_filt ## SIZE ## _v_qpel_r(dst, src, stride, stride);\
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_jj(dst, src, NULL, stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src, stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc13_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride, stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc31_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+1, stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride+1,stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_ff(dst, src, src+stride+1,stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_ii(dst, src, src+stride+1,stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_kk(dst, src, src+stride+1,stride, stride); \
yading@10: }\
yading@10: \
yading@10: static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc23_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
yading@10: {\
yading@10:   OPNAME ## cavs_filt ## SIZE ## _hv_qq(dst, src, src+stride+1,stride, stride); \
yading@10: }\
yading@10: 
yading@10: #define op_put1(a, b)  a = cm[((b)+4)>>3]
yading@10: #define op_put2(a, b)  a = cm[((b)+64)>>7]
yading@10: #define op_put3(a, b)  a = cm[((b)+32)>>6]
yading@10: #define op_put4(a, b)  a = cm[((b)+512)>>10]
yading@10: #define op_avg1(a, b)  a = ((a)+cm[((b)+4)>>3]   +1)>>1
yading@10: #define op_avg2(a, b)  a = ((a)+cm[((b)+64)>>7]  +1)>>1
yading@10: #define op_avg3(a, b)  a = ((a)+cm[((b)+32)>>6]  +1)>>1
yading@10: #define op_avg4(a, b)  a = ((a)+cm[((b)+512)>>10]+1)>>1
yading@10: CAVS_SUBPIX(put_   , op_put1, hpel,    0, -1,  5,  5, -1,  0)
yading@10: CAVS_SUBPIX(put_   , op_put2, qpel_l, -1, -2, 96, 42, -7,  0)
yading@10: CAVS_SUBPIX(put_   , op_put2, qpel_r,  0, -7, 42, 96, -2, -1)
yading@10: CAVS_SUBPIX_HV(put_, op_put3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)
yading@10: CAVS_SUBPIX_HV(put_, op_put4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)
yading@10: CAVS_SUBPIX_HV(put_, op_put4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)
yading@10: CAVS_SUBPIX_HV(put_, op_put4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)
yading@10: CAVS_SUBPIX_HV(put_, op_put4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)
yading@10: CAVS_SUBPIX_HV(put_, op_put2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)
yading@10: CAVS_SUBPIX(avg_   , op_avg1, hpel,    0, -1,  5,  5, -1,  0)
yading@10: CAVS_SUBPIX(avg_   , op_avg2, qpel_l, -1, -2, 96, 42, -7,  0)
yading@10: CAVS_SUBPIX(avg_   , op_avg2, qpel_r,  0, -7, 42, 96, -2, -1)
yading@10: CAVS_SUBPIX_HV(avg_, op_avg3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)
yading@10: CAVS_SUBPIX_HV(avg_, op_avg4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)
yading@10: CAVS_SUBPIX_HV(avg_, op_avg4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)
yading@10: CAVS_SUBPIX_HV(avg_, op_avg4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)
yading@10: CAVS_SUBPIX_HV(avg_, op_avg4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)
yading@10: CAVS_SUBPIX_HV(avg_, op_avg2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)
yading@10: CAVS_MC(put_, 8)
yading@10: CAVS_MC(put_, 16)
yading@10: CAVS_MC(avg_, 8)
yading@10: CAVS_MC(avg_, 16)
yading@10: 
yading@10: #define ff_put_cavs_qpel8_mc00_c  ff_put_pixels8x8_c
yading@10: #define ff_avg_cavs_qpel8_mc00_c  ff_avg_pixels8x8_c
yading@10: #define ff_put_cavs_qpel16_mc00_c ff_put_pixels16x16_c
yading@10: #define ff_avg_cavs_qpel16_mc00_c ff_avg_pixels16x16_c
yading@10: 
yading@10: av_cold void ff_cavsdsp_init(CAVSDSPContext* c, AVCodecContext *avctx) {
yading@10: #define dspfunc(PFX, IDX, NUM) \
yading@10:     c->PFX ## _pixels_tab[IDX][ 0] = ff_ ## PFX ## NUM ## _mc00_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 1] = ff_ ## PFX ## NUM ## _mc10_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 2] = ff_ ## PFX ## NUM ## _mc20_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 3] = ff_ ## PFX ## NUM ## _mc30_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 4] = ff_ ## PFX ## NUM ## _mc01_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 5] = ff_ ## PFX ## NUM ## _mc11_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 6] = ff_ ## PFX ## NUM ## _mc21_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 7] = ff_ ## PFX ## NUM ## _mc31_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 8] = ff_ ## PFX ## NUM ## _mc02_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][ 9] = ff_ ## PFX ## NUM ## _mc12_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][10] = ff_ ## PFX ## NUM ## _mc22_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][11] = ff_ ## PFX ## NUM ## _mc32_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][12] = ff_ ## PFX ## NUM ## _mc03_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][13] = ff_ ## PFX ## NUM ## _mc13_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][14] = ff_ ## PFX ## NUM ## _mc23_c; \
yading@10:     c->PFX ## _pixels_tab[IDX][15] = ff_ ## PFX ## NUM ## _mc33_c
yading@10:     dspfunc(put_cavs_qpel, 0, 16);
yading@10:     dspfunc(put_cavs_qpel, 1, 8);
yading@10:     dspfunc(avg_cavs_qpel, 0, 16);
yading@10:     dspfunc(avg_cavs_qpel, 1, 8);
yading@10:     c->cavs_filter_lv = cavs_filter_lv_c;
yading@10:     c->cavs_filter_lh = cavs_filter_lh_c;
yading@10:     c->cavs_filter_cv = cavs_filter_cv_c;
yading@10:     c->cavs_filter_ch = cavs_filter_ch_c;
yading@10:     c->cavs_idct8_add = cavs_idct8_add_c;
yading@10:     c->idct_perm = FF_NO_IDCT_PERM;
yading@10: 
yading@10:     if (ARCH_X86)
yading@10:         ff_cavsdsp_init_x86(c, avctx);
yading@10: }