FFmpeg: libavcodec/dsputil_template.c Source File

00001 /*
00002  * DSP utils
00003  * Copyright (c) 2000, 2001 Fabrice Bellard
00004  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
00005  *
00006  * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
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 
00030 #include "bit_depth_template.c"
00031 
00032 static inline void FUNC(copy_block2)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
00033 {
00034     int i;
00035     for(i=0; i<h; i++)
00036     {
00037         AV_WN2P(dst   , AV_RN2P(src   ));
00038         dst+=dstStride;
00039         src+=srcStride;
00040     }
00041 }
00042 
00043 static inline void FUNC(copy_block4)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
00044 {
00045     int i;
00046     for(i=0; i<h; i++)
00047     {
00048         AV_WN4P(dst   , AV_RN4P(src   ));
00049         dst+=dstStride;
00050         src+=srcStride;
00051     }
00052 }
00053 
00054 static inline void FUNC(copy_block8)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
00055 {
00056     int i;
00057     for(i=0; i<h; i++)
00058     {
00059         AV_WN4P(dst                , AV_RN4P(src                ));
00060         AV_WN4P(dst+4*sizeof(pixel), AV_RN4P(src+4*sizeof(pixel)));
00061         dst+=dstStride;
00062         src+=srcStride;
00063     }
00064 }
00065 
00066 static inline void FUNC(copy_block16)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
00067 {
00068     int i;
00069     for(i=0; i<h; i++)
00070     {
00071         AV_WN4P(dst                 , AV_RN4P(src                 ));
00072         AV_WN4P(dst+ 4*sizeof(pixel), AV_RN4P(src+ 4*sizeof(pixel)));
00073         AV_WN4P(dst+ 8*sizeof(pixel), AV_RN4P(src+ 8*sizeof(pixel)));
00074         AV_WN4P(dst+12*sizeof(pixel), AV_RN4P(src+12*sizeof(pixel)));
00075         dst+=dstStride;
00076         src+=srcStride;
00077     }
00078 }
00079 
00080 /* draw the edges of width 'w' of an image of size width, height */
00081 //FIXME check that this is ok for mpeg4 interlaced
00082 static void FUNCC(draw_edges)(uint8_t *p_buf, int p_wrap, int width, int height, int w, int h, int sides)
00083 {
00084     pixel *buf = (pixel*)p_buf;
00085     int wrap = p_wrap / sizeof(pixel);
00086     pixel *ptr, *last_line;
00087     int i;
00088 
00089     /* left and right */
00090     ptr = buf;
00091     for(i=0;i<height;i++) {
00092 #if BIT_DEPTH > 8
00093         int j;
00094         for (j = 0; j < w; j++) {
00095             ptr[j-w] = ptr[0];
00096             ptr[j+width] = ptr[width-1];
00097         }
00098 #else
00099         memset(ptr - w, ptr[0], w);
00100         memset(ptr + width, ptr[width-1], w);
00101 #endif
00102         ptr += wrap;
00103     }
00104 
00105     /* top and bottom + corners */
00106     buf -= w;
00107     last_line = buf + (height - 1) * wrap;
00108     if (sides & EDGE_TOP)
00109         for(i = 0; i < h; i++)
00110             memcpy(buf - (i + 1) * wrap, buf, (width + w + w) * sizeof(pixel)); // top
00111     if (sides & EDGE_BOTTOM)
00112         for (i = 0; i < h; i++)
00113             memcpy(last_line + (i + 1) * wrap, last_line, (width + w + w) * sizeof(pixel)); // bottom
00114 }
00115 
00128 void FUNC(ff_emulated_edge_mc)(uint8_t *buf, const uint8_t *src, int linesize, int block_w, int block_h,
00129                                     int src_x, int src_y, int w, int h){
00130     int x, y;
00131     int start_y, start_x, end_y, end_x;
00132 
00133     if(!w || !h)
00134         return;
00135 
00136     if(src_y>= h){
00137         src-= src_y*linesize;
00138         src+= (h-1)*linesize;
00139         src_y=h-1;
00140     }else if(src_y<=-block_h){
00141         src-= src_y*linesize;
00142         src+= (1-block_h)*linesize;
00143         src_y=1-block_h;
00144     }
00145     if(src_x>= w){
00146         src+= (w-1-src_x)*sizeof(pixel);
00147         src_x=w-1;
00148     }else if(src_x<=-block_w){
00149         src+= (1-block_w-src_x)*sizeof(pixel);
00150         src_x=1-block_w;
00151     }
00152 
00153     start_y= FFMAX(0, -src_y);
00154     start_x= FFMAX(0, -src_x);
00155     end_y= FFMIN(block_h, h-src_y);
00156     end_x= FFMIN(block_w, w-src_x);
00157     av_assert2(start_y < end_y && block_h);
00158     av_assert2(start_x < end_x && block_w);
00159 
00160     w    = end_x - start_x;
00161     src += start_y*linesize + start_x*sizeof(pixel);
00162     buf += start_x*sizeof(pixel);
00163 
00164     //top
00165     for(y=0; y<start_y; y++){
00166         memcpy(buf, src, w*sizeof(pixel));
00167         buf += linesize;
00168     }
00169 
00170     // copy existing part
00171     for(; y<end_y; y++){
00172         memcpy(buf, src, w*sizeof(pixel));
00173         src += linesize;
00174         buf += linesize;
00175     }
00176 
00177     //bottom
00178     src -= linesize;
00179     for(; y<block_h; y++){
00180         memcpy(buf, src, w*sizeof(pixel));
00181         buf += linesize;
00182     }
00183 
00184     buf -= block_h * linesize + start_x*sizeof(pixel);
00185     while (block_h--){
00186         pixel *bufp = (pixel*)buf;
00187        //left
00188         for(x=0; x<start_x; x++){
00189             bufp[x] = bufp[start_x];
00190         }
00191 
00192        //right
00193         for(x=end_x; x<block_w; x++){
00194             bufp[x] = bufp[end_x - 1];
00195         }
00196         buf += linesize;
00197     }
00198 }
00199 
00200 #define DCTELEM_FUNCS(dctcoef, suffix)                                  \
00201 static void FUNCC(get_pixels ## suffix)(DCTELEM *av_restrict _block,    \
00202                                         const uint8_t *_pixels,         \
00203                                         int line_size)                  \
00204 {                                                                       \
00205     const pixel *pixels = (const pixel *) _pixels;                      \
00206     dctcoef *av_restrict block = (dctcoef *) _block;                    \
00207     int i;                                                              \
00208                                                                         \
00209     /* read the pixels */                                               \
00210     for(i=0;i<8;i++) {                                                  \
00211         block[0] = pixels[0];                                           \
00212         block[1] = pixels[1];                                           \
00213         block[2] = pixels[2];                                           \
00214         block[3] = pixels[3];                                           \
00215         block[4] = pixels[4];                                           \
00216         block[5] = pixels[5];                                           \
00217         block[6] = pixels[6];                                           \
00218         block[7] = pixels[7];                                           \
00219         pixels += line_size / sizeof(pixel);                            \
00220         block += 8;                                                     \
00221     }                                                                   \
00222 }                                                                       \
00223                                                                         \
00224 static void FUNCC(add_pixels8 ## suffix)(uint8_t *av_restrict _pixels,  \
00225                                          DCTELEM *_block,               \
00226                                          int line_size)                 \
00227 {                                                                       \
00228     int i;                                                              \
00229     pixel *av_restrict pixels = (pixel *av_restrict)_pixels;            \
00230     dctcoef *block = (dctcoef*)_block;                                  \
00231     line_size /= sizeof(pixel);                                         \
00232                                                                         \
00233     for(i=0;i<8;i++) {                                                  \
00234         pixels[0] += block[0];                                          \
00235         pixels[1] += block[1];                                          \
00236         pixels[2] += block[2];                                          \
00237         pixels[3] += block[3];                                          \
00238         pixels[4] += block[4];                                          \
00239         pixels[5] += block[5];                                          \
00240         pixels[6] += block[6];                                          \
00241         pixels[7] += block[7];                                          \
00242         pixels += line_size;                                            \
00243         block += 8;                                                     \
00244     }                                                                   \
00245 }                                                                       \
00246                                                                         \
00247 static void FUNCC(add_pixels4 ## suffix)(uint8_t *av_restrict _pixels,  \
00248                                          DCTELEM *_block,               \
00249                                          int line_size)                 \
00250 {                                                                       \
00251     int i;                                                              \
00252     pixel *av_restrict pixels = (pixel *av_restrict)_pixels;            \
00253     dctcoef *block = (dctcoef*)_block;                                  \
00254     line_size /= sizeof(pixel);                                         \
00255                                                                         \
00256     for(i=0;i<4;i++) {                                                  \
00257         pixels[0] += block[0];                                          \
00258         pixels[1] += block[1];                                          \
00259         pixels[2] += block[2];                                          \
00260         pixels[3] += block[3];                                          \
00261         pixels += line_size;                                            \
00262         block += 4;                                                     \
00263     }                                                                   \
00264 }                                                                       \
00265                                                                         \
00266 static void FUNCC(clear_block ## suffix)(DCTELEM *block)                \
00267 {                                                                       \
00268     memset(block, 0, sizeof(dctcoef)*64);                               \
00269 }                                                                       \
00270                                                                         \
00271                                                                      \
00274 static void FUNCC(clear_blocks ## suffix)(DCTELEM *blocks)              \
00275 {                                                                       \
00276     memset(blocks, 0, sizeof(dctcoef)*6*64);                            \
00277 }
00278 
00279 DCTELEM_FUNCS(DCTELEM, _16)
00280 #if BIT_DEPTH > 8
00281 DCTELEM_FUNCS(dctcoef, _32)
00282 #endif
00283 
00284 #define PIXOP2(OPNAME, OP) \
00285 static void FUNCC(OPNAME ## _pixels2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00286     int i;\
00287     for(i=0; i<h; i++){\
00288         OP(*((pixel2*)(block  )), AV_RN2P(pixels  ));\
00289         pixels+=line_size;\
00290         block +=line_size;\
00291     }\
00292 }\
00293 static void FUNCC(OPNAME ## _pixels4)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00294     int i;\
00295     for(i=0; i<h; i++){\
00296         OP(*((pixel4*)(block  )), AV_RN4P(pixels  ));\
00297         pixels+=line_size;\
00298         block +=line_size;\
00299     }\
00300 }\
00301 static void FUNCC(OPNAME ## _pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00302     int i;\
00303     for(i=0; i<h; i++){\
00304         OP(*((pixel4*)(block                )), AV_RN4P(pixels                ));\
00305         OP(*((pixel4*)(block+4*sizeof(pixel))), AV_RN4P(pixels+4*sizeof(pixel)));\
00306         pixels+=line_size;\
00307         block +=line_size;\
00308     }\
00309 }\
00310 static inline void FUNCC(OPNAME ## _no_rnd_pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00311     FUNCC(OPNAME ## _pixels8)(block, pixels, line_size, h);\
00312 }\
00313 \
00314 static inline void FUNC(OPNAME ## _no_rnd_pixels8_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00315                                                 int src_stride1, int src_stride2, int h){\
00316     int i;\
00317     for(i=0; i<h; i++){\
00318         pixel4 a,b;\
00319         a= AV_RN4P(&src1[i*src_stride1  ]);\
00320         b= AV_RN4P(&src2[i*src_stride2  ]);\
00321         OP(*((pixel4*)&dst[i*dst_stride  ]), no_rnd_avg_pixel4(a, b));\
00322         a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\
00323         b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\
00324         OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), no_rnd_avg_pixel4(a, b));\
00325     }\
00326 }\
00327 \
00328 static inline void FUNC(OPNAME ## _pixels8_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00329                                                 int src_stride1, int src_stride2, int h){\
00330     int i;\
00331     for(i=0; i<h; i++){\
00332         pixel4 a,b;\
00333         a= AV_RN4P(&src1[i*src_stride1  ]);\
00334         b= AV_RN4P(&src2[i*src_stride2  ]);\
00335         OP(*((pixel4*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\
00336         a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\
00337         b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\
00338         OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), rnd_avg_pixel4(a, b));\
00339     }\
00340 }\
00341 \
00342 static inline void FUNC(OPNAME ## _pixels4_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00343                                                 int src_stride1, int src_stride2, int h){\
00344     int i;\
00345     for(i=0; i<h; i++){\
00346         pixel4 a,b;\
00347         a= AV_RN4P(&src1[i*src_stride1  ]);\
00348         b= AV_RN4P(&src2[i*src_stride2  ]);\
00349         OP(*((pixel4*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\
00350     }\
00351 }\
00352 \
00353 static inline void FUNC(OPNAME ## _pixels2_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00354                                                 int src_stride1, int src_stride2, int h){\
00355     int i;\
00356     for(i=0; i<h; i++){\
00357         pixel4 a,b;\
00358         a= AV_RN2P(&src1[i*src_stride1  ]);\
00359         b= AV_RN2P(&src2[i*src_stride2  ]);\
00360         OP(*((pixel2*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\
00361     }\
00362 }\
00363 \
00364 static inline void FUNC(OPNAME ## _pixels16_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00365                                                 int src_stride1, int src_stride2, int h){\
00366     FUNC(OPNAME ## _pixels8_l2)(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\
00367     FUNC(OPNAME ## _pixels8_l2)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, h);\
00368 }\
00369 \
00370 static inline void FUNC(OPNAME ## _no_rnd_pixels16_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00371                                                 int src_stride1, int src_stride2, int h){\
00372     FUNC(OPNAME ## _no_rnd_pixels8_l2)(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\
00373     FUNC(OPNAME ## _no_rnd_pixels8_l2)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, h);\
00374 }\
00375 \
00376 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00377     FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00378 }\
00379 \
00380 static inline void FUNCC(OPNAME ## _pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00381     FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00382 }\
00383 \
00384 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00385     FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00386 }\
00387 \
00388 static inline void FUNCC(OPNAME ## _pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00389     FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00390 }\
00391 \
00392 static inline void FUNC(OPNAME ## _pixels8_l4)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint8_t *src3, const uint8_t *src4,\
00393                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00394     /* FIXME HIGH BIT DEPTH */\
00395     int i;\
00396     for(i=0; i<h; i++){\
00397         uint32_t a, b, c, d, l0, l1, h0, h1;\
00398         a= AV_RN32(&src1[i*src_stride1]);\
00399         b= AV_RN32(&src2[i*src_stride2]);\
00400         c= AV_RN32(&src3[i*src_stride3]);\
00401         d= AV_RN32(&src4[i*src_stride4]);\
00402         l0=  (a&0x03030303UL)\
00403            + (b&0x03030303UL)\
00404            + 0x02020202UL;\
00405         h0= ((a&0xFCFCFCFCUL)>>2)\
00406           + ((b&0xFCFCFCFCUL)>>2);\
00407         l1=  (c&0x03030303UL)\
00408            + (d&0x03030303UL);\
00409         h1= ((c&0xFCFCFCFCUL)>>2)\
00410           + ((d&0xFCFCFCFCUL)>>2);\
00411         OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00412         a= AV_RN32(&src1[i*src_stride1+4]);\
00413         b= AV_RN32(&src2[i*src_stride2+4]);\
00414         c= AV_RN32(&src3[i*src_stride3+4]);\
00415         d= AV_RN32(&src4[i*src_stride4+4]);\
00416         l0=  (a&0x03030303UL)\
00417            + (b&0x03030303UL)\
00418            + 0x02020202UL;\
00419         h0= ((a&0xFCFCFCFCUL)>>2)\
00420           + ((b&0xFCFCFCFCUL)>>2);\
00421         l1=  (c&0x03030303UL)\
00422            + (d&0x03030303UL);\
00423         h1= ((c&0xFCFCFCFCUL)>>2)\
00424           + ((d&0xFCFCFCFCUL)>>2);\
00425         OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00426     }\
00427 }\
00428 \
00429 static inline void FUNCC(OPNAME ## _pixels4_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00430     FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00431 }\
00432 \
00433 static inline void FUNCC(OPNAME ## _pixels4_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00434     FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00435 }\
00436 \
00437 static inline void FUNCC(OPNAME ## _pixels2_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00438     FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00439 }\
00440 \
00441 static inline void FUNCC(OPNAME ## _pixels2_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00442     FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00443 }\
00444 \
00445 static inline void FUNC(OPNAME ## _no_rnd_pixels8_l4)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint8_t *src3, const uint8_t *src4,\
00446                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00447     /* FIXME HIGH BIT DEPTH*/\
00448     int i;\
00449     for(i=0; i<h; i++){\
00450         uint32_t a, b, c, d, l0, l1, h0, h1;\
00451         a= AV_RN32(&src1[i*src_stride1]);\
00452         b= AV_RN32(&src2[i*src_stride2]);\
00453         c= AV_RN32(&src3[i*src_stride3]);\
00454         d= AV_RN32(&src4[i*src_stride4]);\
00455         l0=  (a&0x03030303UL)\
00456            + (b&0x03030303UL)\
00457            + 0x01010101UL;\
00458         h0= ((a&0xFCFCFCFCUL)>>2)\
00459           + ((b&0xFCFCFCFCUL)>>2);\
00460         l1=  (c&0x03030303UL)\
00461            + (d&0x03030303UL);\
00462         h1= ((c&0xFCFCFCFCUL)>>2)\
00463           + ((d&0xFCFCFCFCUL)>>2);\
00464         OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00465         a= AV_RN32(&src1[i*src_stride1+4]);\
00466         b= AV_RN32(&src2[i*src_stride2+4]);\
00467         c= AV_RN32(&src3[i*src_stride3+4]);\
00468         d= AV_RN32(&src4[i*src_stride4+4]);\
00469         l0=  (a&0x03030303UL)\
00470            + (b&0x03030303UL)\
00471            + 0x01010101UL;\
00472         h0= ((a&0xFCFCFCFCUL)>>2)\
00473           + ((b&0xFCFCFCFCUL)>>2);\
00474         l1=  (c&0x03030303UL)\
00475            + (d&0x03030303UL);\
00476         h1= ((c&0xFCFCFCFCUL)>>2)\
00477           + ((d&0xFCFCFCFCUL)>>2);\
00478         OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00479     }\
00480 }\
00481 static inline void FUNC(OPNAME ## _pixels16_l4)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint8_t *src3, const uint8_t *src4,\
00482                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00483     FUNC(OPNAME ## _pixels8_l4)(dst  , src1  , src2  , src3  , src4  , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
00484     FUNC(OPNAME ## _pixels8_l4)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), src3+8*sizeof(pixel), src4+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
00485 }\
00486 static inline void FUNC(OPNAME ## _no_rnd_pixels16_l4)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint8_t *src3, const uint8_t *src4,\
00487                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00488     FUNC(OPNAME ## _no_rnd_pixels8_l4)(dst  , src1  , src2  , src3  , src4  , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
00489     FUNC(OPNAME ## _no_rnd_pixels8_l4)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), src3+8*sizeof(pixel), src4+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
00490 }\
00491 \
00492 static inline void FUNCC(OPNAME ## _pixels2_xy2)(uint8_t *p_block, const uint8_t *p_pixels, int line_size, int h)\
00493 {\
00494         int i, a0, b0, a1, b1;\
00495         pixel *block = (pixel*)p_block;\
00496         const pixel *pixels = (const pixel*)p_pixels;\
00497         line_size >>= sizeof(pixel)-1;\
00498         a0= pixels[0];\
00499         b0= pixels[1] + 2;\
00500         a0 += b0;\
00501         b0 += pixels[2];\
00502 \
00503         pixels+=line_size;\
00504         for(i=0; i<h; i+=2){\
00505             a1= pixels[0];\
00506             b1= pixels[1];\
00507             a1 += b1;\
00508             b1 += pixels[2];\
00509 \
00510             block[0]= (a1+a0)>>2; /* FIXME non put */\
00511             block[1]= (b1+b0)>>2;\
00512 \
00513             pixels+=line_size;\
00514             block +=line_size;\
00515 \
00516             a0= pixels[0];\
00517             b0= pixels[1] + 2;\
00518             a0 += b0;\
00519             b0 += pixels[2];\
00520 \
00521             block[0]= (a1+a0)>>2;\
00522             block[1]= (b1+b0)>>2;\
00523             pixels+=line_size;\
00524             block +=line_size;\
00525         }\
00526 }\
00527 \
00528 static inline void FUNCC(OPNAME ## _pixels4_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
00529 {\
00530         /* FIXME HIGH BIT DEPTH */\
00531         int i;\
00532         const uint32_t a= AV_RN32(pixels  );\
00533         const uint32_t b= AV_RN32(pixels+1);\
00534         uint32_t l0=  (a&0x03030303UL)\
00535                     + (b&0x03030303UL)\
00536                     + 0x02020202UL;\
00537         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
00538                    + ((b&0xFCFCFCFCUL)>>2);\
00539         uint32_t l1,h1;\
00540 \
00541         pixels+=line_size;\
00542         for(i=0; i<h; i+=2){\
00543             uint32_t a= AV_RN32(pixels  );\
00544             uint32_t b= AV_RN32(pixels+1);\
00545             l1=  (a&0x03030303UL)\
00546                + (b&0x03030303UL);\
00547             h1= ((a&0xFCFCFCFCUL)>>2)\
00548               + ((b&0xFCFCFCFCUL)>>2);\
00549             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00550             pixels+=line_size;\
00551             block +=line_size;\
00552             a= AV_RN32(pixels  );\
00553             b= AV_RN32(pixels+1);\
00554             l0=  (a&0x03030303UL)\
00555                + (b&0x03030303UL)\
00556                + 0x02020202UL;\
00557             h0= ((a&0xFCFCFCFCUL)>>2)\
00558               + ((b&0xFCFCFCFCUL)>>2);\
00559             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00560             pixels+=line_size;\
00561             block +=line_size;\
00562         }\
00563 }\
00564 \
00565 static inline void FUNCC(OPNAME ## _pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
00566 {\
00567     /* FIXME HIGH BIT DEPTH */\
00568     int j;\
00569     for(j=0; j<2; j++){\
00570         int i;\
00571         const uint32_t a= AV_RN32(pixels  );\
00572         const uint32_t b= AV_RN32(pixels+1);\
00573         uint32_t l0=  (a&0x03030303UL)\
00574                     + (b&0x03030303UL)\
00575                     + 0x02020202UL;\
00576         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
00577                    + ((b&0xFCFCFCFCUL)>>2);\
00578         uint32_t l1,h1;\
00579 \
00580         pixels+=line_size;\
00581         for(i=0; i<h; i+=2){\
00582             uint32_t a= AV_RN32(pixels  );\
00583             uint32_t b= AV_RN32(pixels+1);\
00584             l1=  (a&0x03030303UL)\
00585                + (b&0x03030303UL);\
00586             h1= ((a&0xFCFCFCFCUL)>>2)\
00587               + ((b&0xFCFCFCFCUL)>>2);\
00588             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00589             pixels+=line_size;\
00590             block +=line_size;\
00591             a= AV_RN32(pixels  );\
00592             b= AV_RN32(pixels+1);\
00593             l0=  (a&0x03030303UL)\
00594                + (b&0x03030303UL)\
00595                + 0x02020202UL;\
00596             h0= ((a&0xFCFCFCFCUL)>>2)\
00597               + ((b&0xFCFCFCFCUL)>>2);\
00598             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00599             pixels+=line_size;\
00600             block +=line_size;\
00601         }\
00602         pixels+=4-line_size*(h+1);\
00603         block +=4-line_size*h;\
00604     }\
00605 }\
00606 \
00607 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
00608 {\
00609     /* FIXME HIGH BIT DEPTH */\
00610     int j;\
00611     for(j=0; j<2; j++){\
00612         int i;\
00613         const uint32_t a= AV_RN32(pixels  );\
00614         const uint32_t b= AV_RN32(pixels+1);\
00615         uint32_t l0=  (a&0x03030303UL)\
00616                     + (b&0x03030303UL)\
00617                     + 0x01010101UL;\
00618         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
00619                    + ((b&0xFCFCFCFCUL)>>2);\
00620         uint32_t l1,h1;\
00621 \
00622         pixels+=line_size;\
00623         for(i=0; i<h; i+=2){\
00624             uint32_t a= AV_RN32(pixels  );\
00625             uint32_t b= AV_RN32(pixels+1);\
00626             l1=  (a&0x03030303UL)\
00627                + (b&0x03030303UL);\
00628             h1= ((a&0xFCFCFCFCUL)>>2)\
00629               + ((b&0xFCFCFCFCUL)>>2);\
00630             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00631             pixels+=line_size;\
00632             block +=line_size;\
00633             a= AV_RN32(pixels  );\
00634             b= AV_RN32(pixels+1);\
00635             l0=  (a&0x03030303UL)\
00636                + (b&0x03030303UL)\
00637                + 0x01010101UL;\
00638             h0= ((a&0xFCFCFCFCUL)>>2)\
00639               + ((b&0xFCFCFCFCUL)>>2);\
00640             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00641             pixels+=line_size;\
00642             block +=line_size;\
00643         }\
00644         pixels+=4-line_size*(h+1);\
00645         block +=4-line_size*h;\
00646     }\
00647 }\
00648 \
00649 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16)    , FUNCC(OPNAME ## _pixels8)    , 8*sizeof(pixel))\
00650 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_x2) , FUNCC(OPNAME ## _pixels8_x2) , 8*sizeof(pixel))\
00651 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_y2) , FUNCC(OPNAME ## _pixels8_y2) , 8*sizeof(pixel))\
00652 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_xy2), FUNCC(OPNAME ## _pixels8_xy2), 8*sizeof(pixel))\
00653 av_unused CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16)    , FUNCC(OPNAME ## _pixels8) , 8*sizeof(pixel))\
00654 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_x2) , FUNCC(OPNAME ## _no_rnd_pixels8_x2) , 8*sizeof(pixel))\
00655 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_y2) , FUNCC(OPNAME ## _no_rnd_pixels8_y2) , 8*sizeof(pixel))\
00656 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_xy2), FUNCC(OPNAME ## _no_rnd_pixels8_xy2), 8*sizeof(pixel))\
00657 
00658 #define op_avg(a, b) a = rnd_avg_pixel4(a, b)
00659 #define op_put(a, b) a = b
00660 
00661 PIXOP2(avg, op_avg)
00662 PIXOP2(put, op_put)
00663 #undef op_avg
00664 #undef op_put
00665 
00666 #define put_no_rnd_pixels8_c  put_pixels8_c
00667 #define put_no_rnd_pixels16_c put_pixels16_c
00668 
00669 static void FUNCC(put_no_rnd_pixels16_l2)(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){
00670     FUNC(put_no_rnd_pixels16_l2)(dst, a, b, stride, stride, stride, h);
00671 }
00672 
00673 static void FUNCC(put_no_rnd_pixels8_l2)(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){
00674     FUNC(put_no_rnd_pixels8_l2)(dst, a, b, stride, stride, stride, h);
00675 }
00676 
00677 #define H264_CHROMA_MC(OPNAME, OP)\
00678 static void FUNCC(OPNAME ## h264_chroma_mc2)(uint8_t *p_dst/*align 8*/, uint8_t *p_src/*align 1*/, int stride, int h, int x, int y){\
00679     pixel *dst = (pixel*)p_dst;\
00680     pixel *src = (pixel*)p_src;\
00681     const int A=(8-x)*(8-y);\
00682     const int B=(  x)*(8-y);\
00683     const int C=(8-x)*(  y);\
00684     const int D=(  x)*(  y);\
00685     int i;\
00686     stride >>= sizeof(pixel)-1;\
00687     \
00688     av_assert2(x<8 && y<8 && x>=0 && y>=0);\
00689 \
00690     if(D){\
00691         for(i=0; i<h; i++){\
00692             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
00693             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
00694             dst+= stride;\
00695             src+= stride;\
00696         }\
00697     }else{\
00698         const int E= B+C;\
00699         const int step= C ? stride : 1;\
00700         for(i=0; i<h; i++){\
00701             OP(dst[0], (A*src[0] + E*src[step+0]));\
00702             OP(dst[1], (A*src[1] + E*src[step+1]));\
00703             dst+= stride;\
00704             src+= stride;\
00705         }\
00706     }\
00707 }\
00708 \
00709 static void FUNCC(OPNAME ## h264_chroma_mc4)(uint8_t *p_dst/*align 8*/, uint8_t *p_src/*align 1*/, int stride, int h, int x, int y){\
00710     pixel *dst = (pixel*)p_dst;\
00711     pixel *src = (pixel*)p_src;\
00712     const int A=(8-x)*(8-y);\
00713     const int B=(  x)*(8-y);\
00714     const int C=(8-x)*(  y);\
00715     const int D=(  x)*(  y);\
00716     int i;\
00717     stride >>= sizeof(pixel)-1;\
00718     \
00719     av_assert2(x<8 && y<8 && x>=0 && y>=0);\
00720 \
00721     if(D){\
00722         for(i=0; i<h; i++){\
00723             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
00724             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
00725             OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\
00726             OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\
00727             dst+= stride;\
00728             src+= stride;\
00729         }\
00730     }else{\
00731         const int E= B+C;\
00732         const int step= C ? stride : 1;\
00733         for(i=0; i<h; i++){\
00734             OP(dst[0], (A*src[0] + E*src[step+0]));\
00735             OP(dst[1], (A*src[1] + E*src[step+1]));\
00736             OP(dst[2], (A*src[2] + E*src[step+2]));\
00737             OP(dst[3], (A*src[3] + E*src[step+3]));\
00738             dst+= stride;\
00739             src+= stride;\
00740         }\
00741     }\
00742 }\
00743 \
00744 static void FUNCC(OPNAME ## h264_chroma_mc8)(uint8_t *p_dst/*align 8*/, uint8_t *p_src/*align 1*/, int stride, int h, int x, int y){\
00745     pixel *dst = (pixel*)p_dst;\
00746     pixel *src = (pixel*)p_src;\
00747     const int A=(8-x)*(8-y);\
00748     const int B=(  x)*(8-y);\
00749     const int C=(8-x)*(  y);\
00750     const int D=(  x)*(  y);\
00751     int i;\
00752     stride >>= sizeof(pixel)-1;\
00753     \
00754     av_assert2(x<8 && y<8 && x>=0 && y>=0);\
00755 \
00756     if(D){\
00757         for(i=0; i<h; i++){\
00758             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
00759             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
00760             OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\
00761             OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\
00762             OP(dst[4], (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5]));\
00763             OP(dst[5], (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6]));\
00764             OP(dst[6], (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7]));\
00765             OP(dst[7], (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8]));\
00766             dst+= stride;\
00767             src+= stride;\
00768         }\
00769     }else{\
00770         const int E= B+C;\
00771         const int step= C ? stride : 1;\
00772         for(i=0; i<h; i++){\
00773             OP(dst[0], (A*src[0] + E*src[step+0]));\
00774             OP(dst[1], (A*src[1] + E*src[step+1]));\
00775             OP(dst[2], (A*src[2] + E*src[step+2]));\
00776             OP(dst[3], (A*src[3] + E*src[step+3]));\
00777             OP(dst[4], (A*src[4] + E*src[step+4]));\
00778             OP(dst[5], (A*src[5] + E*src[step+5]));\
00779             OP(dst[6], (A*src[6] + E*src[step+6]));\
00780             OP(dst[7], (A*src[7] + E*src[step+7]));\
00781             dst+= stride;\
00782             src+= stride;\
00783         }\
00784     }\
00785 }
00786 
00787 #define op_avg(a, b) a = (((a)+(((b) + 32)>>6)+1)>>1)
00788 #define op_put(a, b) a = (((b) + 32)>>6)
00789 
00790 H264_CHROMA_MC(put_       , op_put)
00791 H264_CHROMA_MC(avg_       , op_avg)
00792 #undef op_avg
00793 #undef op_put
00794 
00795 #define H264_LOWPASS(OPNAME, OP, OP2) \
00796 static av_unused void FUNC(OPNAME ## h264_qpel2_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00797     const int h=2;\
00798     INIT_CLIP\
00799     int i;\
00800     pixel *dst = (pixel*)p_dst;\
00801     pixel *src = (pixel*)p_src;\
00802     dstStride >>= sizeof(pixel)-1;\
00803     srcStride >>= sizeof(pixel)-1;\
00804     for(i=0; i<h; i++)\
00805     {\
00806         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
00807         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
00808         dst+=dstStride;\
00809         src+=srcStride;\
00810     }\
00811 }\
00812 \
00813 static av_unused void FUNC(OPNAME ## h264_qpel2_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00814     const int w=2;\
00815     INIT_CLIP\
00816     int i;\
00817     pixel *dst = (pixel*)p_dst;\
00818     pixel *src = (pixel*)p_src;\
00819     dstStride >>= sizeof(pixel)-1;\
00820     srcStride >>= sizeof(pixel)-1;\
00821     for(i=0; i<w; i++)\
00822     {\
00823         const int srcB= src[-2*srcStride];\
00824         const int srcA= src[-1*srcStride];\
00825         const int src0= src[0 *srcStride];\
00826         const int src1= src[1 *srcStride];\
00827         const int src2= src[2 *srcStride];\
00828         const int src3= src[3 *srcStride];\
00829         const int src4= src[4 *srcStride];\
00830         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
00831         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
00832         dst++;\
00833         src++;\
00834     }\
00835 }\
00836 \
00837 static av_unused void FUNC(OPNAME ## h264_qpel2_hv_lowpass)(uint8_t *p_dst, pixeltmp *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\
00838     const int h=2;\
00839     const int w=2;\
00840     const int pad = (BIT_DEPTH == 10) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
00841     INIT_CLIP\
00842     int i;\
00843     pixel *dst = (pixel*)p_dst;\
00844     pixel *src = (pixel*)p_src;\
00845     dstStride >>= sizeof(pixel)-1;\
00846     srcStride >>= sizeof(pixel)-1;\
00847     src -= 2*srcStride;\
00848     for(i=0; i<h+5; i++)\
00849     {\
00850         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
00851         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
00852         tmp+=tmpStride;\
00853         src+=srcStride;\
00854     }\
00855     tmp -= tmpStride*(h+5-2);\
00856     for(i=0; i<w; i++)\
00857     {\
00858         const int tmpB= tmp[-2*tmpStride] - pad;\
00859         const int tmpA= tmp[-1*tmpStride] - pad;\
00860         const int tmp0= tmp[0 *tmpStride] - pad;\
00861         const int tmp1= tmp[1 *tmpStride] - pad;\
00862         const int tmp2= tmp[2 *tmpStride] - pad;\
00863         const int tmp3= tmp[3 *tmpStride] - pad;\
00864         const int tmp4= tmp[4 *tmpStride] - pad;\
00865         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
00866         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
00867         dst++;\
00868         tmp++;\
00869     }\
00870 }\
00871 static void FUNC(OPNAME ## h264_qpel4_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00872     const int h=4;\
00873     INIT_CLIP\
00874     int i;\
00875     pixel *dst = (pixel*)p_dst;\
00876     pixel *src = (pixel*)p_src;\
00877     dstStride >>= sizeof(pixel)-1;\
00878     srcStride >>= sizeof(pixel)-1;\
00879     for(i=0; i<h; i++)\
00880     {\
00881         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
00882         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
00883         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]));\
00884         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]));\
00885         dst+=dstStride;\
00886         src+=srcStride;\
00887     }\
00888 }\
00889 \
00890 static void FUNC(OPNAME ## h264_qpel4_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00891     const int w=4;\
00892     INIT_CLIP\
00893     int i;\
00894     pixel *dst = (pixel*)p_dst;\
00895     pixel *src = (pixel*)p_src;\
00896     dstStride >>= sizeof(pixel)-1;\
00897     srcStride >>= sizeof(pixel)-1;\
00898     for(i=0; i<w; i++)\
00899     {\
00900         const int srcB= src[-2*srcStride];\
00901         const int srcA= src[-1*srcStride];\
00902         const int src0= src[0 *srcStride];\
00903         const int src1= src[1 *srcStride];\
00904         const int src2= src[2 *srcStride];\
00905         const int src3= src[3 *srcStride];\
00906         const int src4= src[4 *srcStride];\
00907         const int src5= src[5 *srcStride];\
00908         const int src6= src[6 *srcStride];\
00909         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
00910         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
00911         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
00912         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
00913         dst++;\
00914         src++;\
00915     }\
00916 }\
00917 \
00918 static void FUNC(OPNAME ## h264_qpel4_hv_lowpass)(uint8_t *p_dst, pixeltmp *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\
00919     const int h=4;\
00920     const int w=4;\
00921     const int pad = (BIT_DEPTH == 10) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
00922     INIT_CLIP\
00923     int i;\
00924     pixel *dst = (pixel*)p_dst;\
00925     pixel *src = (pixel*)p_src;\
00926     dstStride >>= sizeof(pixel)-1;\
00927     srcStride >>= sizeof(pixel)-1;\
00928     src -= 2*srcStride;\
00929     for(i=0; i<h+5; i++)\
00930     {\
00931         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
00932         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
00933         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]) + pad;\
00934         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]) + pad;\
00935         tmp+=tmpStride;\
00936         src+=srcStride;\
00937     }\
00938     tmp -= tmpStride*(h+5-2);\
00939     for(i=0; i<w; i++)\
00940     {\
00941         const int tmpB= tmp[-2*tmpStride] - pad;\
00942         const int tmpA= tmp[-1*tmpStride] - pad;\
00943         const int tmp0= tmp[0 *tmpStride] - pad;\
00944         const int tmp1= tmp[1 *tmpStride] - pad;\
00945         const int tmp2= tmp[2 *tmpStride] - pad;\
00946         const int tmp3= tmp[3 *tmpStride] - pad;\
00947         const int tmp4= tmp[4 *tmpStride] - pad;\
00948         const int tmp5= tmp[5 *tmpStride] - pad;\
00949         const int tmp6= tmp[6 *tmpStride] - pad;\
00950         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
00951         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
00952         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
00953         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
00954         dst++;\
00955         tmp++;\
00956     }\
00957 }\
00958 \
00959 static void FUNC(OPNAME ## h264_qpel8_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00960     const int h=8;\
00961     INIT_CLIP\
00962     int i;\
00963     pixel *dst = (pixel*)p_dst;\
00964     pixel *src = (pixel*)p_src;\
00965     dstStride >>= sizeof(pixel)-1;\
00966     srcStride >>= sizeof(pixel)-1;\
00967     for(i=0; i<h; i++)\
00968     {\
00969         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]));\
00970         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]));\
00971         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]));\
00972         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]));\
00973         OP(dst[4], (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]));\
00974         OP(dst[5], (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]));\
00975         OP(dst[6], (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]));\
00976         OP(dst[7], (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]));\
00977         dst+=dstStride;\
00978         src+=srcStride;\
00979     }\
00980 }\
00981 \
00982 static void FUNC(OPNAME ## h264_qpel8_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00983     const int w=8;\
00984     INIT_CLIP\
00985     int i;\
00986     pixel *dst = (pixel*)p_dst;\
00987     pixel *src = (pixel*)p_src;\
00988     dstStride >>= sizeof(pixel)-1;\
00989     srcStride >>= sizeof(pixel)-1;\
00990     for(i=0; i<w; i++)\
00991     {\
00992         const int srcB= src[-2*srcStride];\
00993         const int srcA= src[-1*srcStride];\
00994         const int src0= src[0 *srcStride];\
00995         const int src1= src[1 *srcStride];\
00996         const int src2= src[2 *srcStride];\
00997         const int src3= src[3 *srcStride];\
00998         const int src4= src[4 *srcStride];\
00999         const int src5= src[5 *srcStride];\
01000         const int src6= src[6 *srcStride];\
01001         const int src7= src[7 *srcStride];\
01002         const int src8= src[8 *srcStride];\
01003         const int src9= src[9 *srcStride];\
01004         const int src10=src[10*srcStride];\
01005         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
01006         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
01007         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
01008         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
01009         OP(dst[4*dstStride], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\
01010         OP(dst[5*dstStride], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\
01011         OP(dst[6*dstStride], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\
01012         OP(dst[7*dstStride], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\
01013         dst++;\
01014         src++;\
01015     }\
01016 }\
01017 \
01018 static void FUNC(OPNAME ## h264_qpel8_hv_lowpass)(uint8_t *p_dst, pixeltmp *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\
01019     const int h=8;\
01020     const int w=8;\
01021     const int pad = (BIT_DEPTH == 10) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
01022     INIT_CLIP\
01023     int i;\
01024     pixel *dst = (pixel*)p_dst;\
01025     pixel *src = (pixel*)p_src;\
01026     dstStride >>= sizeof(pixel)-1;\
01027     srcStride >>= sizeof(pixel)-1;\
01028     src -= 2*srcStride;\
01029     for(i=0; i<h+5; i++)\
01030     {\
01031         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]) + pad;\
01032         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]) + pad;\
01033         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]) + pad;\
01034         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]) + pad;\
01035         tmp[4]= (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]) + pad;\
01036         tmp[5]= (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]) + pad;\
01037         tmp[6]= (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]) + pad;\
01038         tmp[7]= (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]) + pad;\
01039         tmp+=tmpStride;\
01040         src+=srcStride;\
01041     }\
01042     tmp -= tmpStride*(h+5-2);\
01043     for(i=0; i<w; i++)\
01044     {\
01045         const int tmpB= tmp[-2*tmpStride] - pad;\
01046         const int tmpA= tmp[-1*tmpStride] - pad;\
01047         const int tmp0= tmp[0 *tmpStride] - pad;\
01048         const int tmp1= tmp[1 *tmpStride] - pad;\
01049         const int tmp2= tmp[2 *tmpStride] - pad;\
01050         const int tmp3= tmp[3 *tmpStride] - pad;\
01051         const int tmp4= tmp[4 *tmpStride] - pad;\
01052         const int tmp5= tmp[5 *tmpStride] - pad;\
01053         const int tmp6= tmp[6 *tmpStride] - pad;\
01054         const int tmp7= tmp[7 *tmpStride] - pad;\
01055         const int tmp8= tmp[8 *tmpStride] - pad;\
01056         const int tmp9= tmp[9 *tmpStride] - pad;\
01057         const int tmp10=tmp[10*tmpStride] - pad;\
01058         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
01059         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
01060         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
01061         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
01062         OP2(dst[4*dstStride], (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));\
01063         OP2(dst[5*dstStride], (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));\
01064         OP2(dst[6*dstStride], (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));\
01065         OP2(dst[7*dstStride], (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));\
01066         dst++;\
01067         tmp++;\
01068     }\
01069 }\
01070 \
01071 static void FUNC(OPNAME ## h264_qpel16_v_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
01072     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\
01073     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01074     src += 8*srcStride;\
01075     dst += 8*dstStride;\
01076     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\
01077     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01078 }\
01079 \
01080 static void FUNC(OPNAME ## h264_qpel16_h_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
01081     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\
01082     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01083     src += 8*srcStride;\
01084     dst += 8*dstStride;\
01085     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\
01086     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01087 }\
01088 \
01089 static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, pixeltmp *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
01090     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\
01091     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
01092     src += 8*srcStride;\
01093     dst += 8*dstStride;\
01094     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\
01095     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
01096 }\
01097 
01098 #define H264_MC(OPNAME, SIZE) \
01099 static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\
01100     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\
01101 }\
01102 \
01103 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\
01104     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01105     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
01106     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\
01107 }\
01108 \
01109 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\
01110     FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\
01111 }\
01112 \
01113 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\
01114     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01115     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
01116     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\
01117 }\
01118 \
01119 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\
01120     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01121     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01122     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01123     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01124     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01125     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01126 }\
01127 \
01128 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\
01129     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01130     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01131     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01132     FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\
01133 }\
01134 \
01135 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\
01136     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01137     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01138     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01139     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01140     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01141     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01142 }\
01143 \
01144 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(uint8_t *dst, uint8_t *src, int stride){\
01145     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01146     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01147     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01148     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01149     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
01150     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01151     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01152     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01153 }\
01154 \
01155 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\
01156     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01157     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01158     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01159     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01160     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
01161     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
01162     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01163     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01164 }\
01165 \
01166 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\
01167     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01168     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01169     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01170     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01171     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
01172     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01173     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01174     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01175 }\
01176 \
01177 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\
01178     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01179     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01180     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01181     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01182     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
01183     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
01184     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01185     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01186 }\
01187 \
01188 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\
01189     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01190     FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\
01191 }\
01192 \
01193 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\
01194     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01195     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01196     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01197     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
01198     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01199     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01200 }\
01201 \
01202 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\
01203     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01204     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01205     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01206     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
01207     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01208     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01209 }\
01210 \
01211 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\
01212     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01213     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01214     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01215     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01216     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01217     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01218     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01219     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01220     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01221 }\
01222 \
01223 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\
01224     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01225     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01226     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01227     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01228     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01229     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
01230     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01231     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01232     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01233 }\
01234 
01235 #define op_avg(a, b)  a = (((a)+CLIP(((b) + 16)>>5)+1)>>1)
01236 //#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)
01237 #define op_put(a, b)  a = CLIP(((b) + 16)>>5)
01238 #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)
01239 #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)
01240 
01241 H264_LOWPASS(put_       , op_put, op2_put)
01242 H264_LOWPASS(avg_       , op_avg, op2_avg)
01243 H264_MC(put_, 2)
01244 H264_MC(put_, 4)
01245 H264_MC(put_, 8)
01246 H264_MC(put_, 16)
01247 H264_MC(avg_, 4)
01248 H264_MC(avg_, 8)
01249 H264_MC(avg_, 16)
01250 
01251 #undef op_avg
01252 #undef op_put
01253 #undef op2_avg
01254 #undef op2_put
01255 
01256 #if BIT_DEPTH == 8
01257 #   define put_h264_qpel8_mc00_8_c  ff_put_pixels8x8_8_c
01258 #   define avg_h264_qpel8_mc00_8_c  ff_avg_pixels8x8_8_c
01259 #   define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c
01260 #   define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c
01261 #elif BIT_DEPTH == 9
01262 #   define put_h264_qpel8_mc00_9_c  ff_put_pixels8x8_9_c
01263 #   define avg_h264_qpel8_mc00_9_c  ff_avg_pixels8x8_9_c
01264 #   define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c
01265 #   define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c
01266 #elif BIT_DEPTH == 10
01267 #   define put_h264_qpel8_mc00_10_c  ff_put_pixels8x8_10_c
01268 #   define avg_h264_qpel8_mc00_10_c  ff_avg_pixels8x8_10_c
01269 #   define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c
01270 #   define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c
01271 #elif BIT_DEPTH == 12
01272 #   define put_h264_qpel8_mc00_12_c  ff_put_pixels8x8_12_c
01273 #   define avg_h264_qpel8_mc00_12_c  ff_avg_pixels8x8_12_c
01274 #   define put_h264_qpel16_mc00_12_c ff_put_pixels16x16_12_c
01275 #   define avg_h264_qpel16_mc00_12_c ff_avg_pixels16x16_12_c
01276 #elif BIT_DEPTH == 14
01277 #   define put_h264_qpel8_mc00_14_c  ff_put_pixels8x8_14_c
01278 #   define avg_h264_qpel8_mc00_14_c  ff_avg_pixels8x8_14_c
01279 #   define put_h264_qpel16_mc00_14_c ff_put_pixels16x16_14_c
01280 #   define avg_h264_qpel16_mc00_14_c ff_avg_pixels16x16_14_c
01281 #endif
01282 
01283 void FUNCC(ff_put_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {
01284     FUNCC(put_pixels8)(dst, src, stride, 8);
01285 }
01286 void FUNCC(ff_avg_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {
01287     FUNCC(avg_pixels8)(dst, src, stride, 8);
01288 }
01289 void FUNCC(ff_put_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {
01290     FUNCC(put_pixels16)(dst, src, stride, 16);
01291 }
01292 void FUNCC(ff_avg_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {
01293     FUNCC(avg_pixels16)(dst, src, stride, 16);
01294 }
01295