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(src_y>= h){
00134         src+= (h-1-src_y)*linesize;
00135         src_y=h-1;
00136     }else if(src_y<=-block_h){
00137         src+= (1-block_h-src_y)*linesize;
00138         src_y=1-block_h;
00139     }
00140     if(src_x>= w){
00141         src+= (w-1-src_x)*sizeof(pixel);
00142         src_x=w-1;
00143     }else if(src_x<=-block_w){
00144         src+= (1-block_w-src_x)*sizeof(pixel);
00145         src_x=1-block_w;
00146     }
00147 
00148     start_y= FFMAX(0, -src_y);
00149     start_x= FFMAX(0, -src_x);
00150     end_y= FFMIN(block_h, h-src_y);
00151     end_x= FFMIN(block_w, w-src_x);
00152     assert(start_y < end_y && block_h);
00153     assert(start_x < end_x && block_w);
00154 
00155     w    = end_x - start_x;
00156     src += start_y*linesize + start_x*sizeof(pixel);
00157     buf += start_x*sizeof(pixel);
00158 
00159     //top
00160     for(y=0; y<start_y; y++){
00161         memcpy(buf, src, w*sizeof(pixel));
00162         buf += linesize;
00163     }
00164 
00165     // copy existing part
00166     for(; y<end_y; y++){
00167         memcpy(buf, src, w*sizeof(pixel));
00168         src += linesize;
00169         buf += linesize;
00170     }
00171 
00172     //bottom
00173     src -= linesize;
00174     for(; y<block_h; y++){
00175         memcpy(buf, src, w*sizeof(pixel));
00176         buf += linesize;
00177     }
00178 
00179     buf -= block_h * linesize + start_x*sizeof(pixel);
00180     while (block_h--){
00181         pixel *bufp = (pixel*)buf;
00182        //left
00183         for(x=0; x<start_x; x++){
00184             bufp[x] = bufp[start_x];
00185         }
00186 
00187        //right
00188         for(x=end_x; x<block_w; x++){
00189             bufp[x] = bufp[end_x - 1];
00190         }
00191         buf += linesize;
00192     }
00193 }
00194 
00195 #define DCTELEM_FUNCS(dctcoef, suffix)                                  \
00196 static void FUNCC(get_pixels ## suffix)(DCTELEM *restrict _block,       \
00197                                         const uint8_t *_pixels,         \
00198                                         int line_size)                  \
00199 {                                                                       \
00200     const pixel *pixels = (const pixel *) _pixels;                      \
00201     dctcoef *restrict block = (dctcoef *) _block;                       \
00202     int i;                                                              \
00203                                                                         \
00204     /* read the pixels */                                               \
00205     for(i=0;i<8;i++) {                                                  \
00206         block[0] = pixels[0];                                           \
00207         block[1] = pixels[1];                                           \
00208         block[2] = pixels[2];                                           \
00209         block[3] = pixels[3];                                           \
00210         block[4] = pixels[4];                                           \
00211         block[5] = pixels[5];                                           \
00212         block[6] = pixels[6];                                           \
00213         block[7] = pixels[7];                                           \
00214         pixels += line_size / sizeof(pixel);                            \
00215         block += 8;                                                     \
00216     }                                                                   \
00217 }                                                                       \
00218                                                                         \
00219 static void FUNCC(add_pixels8 ## suffix)(uint8_t *restrict _pixels,     \
00220                                          DCTELEM *_block,               \
00221                                          int line_size)                 \
00222 {                                                                       \
00223     int i;                                                              \
00224     pixel *restrict pixels = (pixel *restrict)_pixels;                  \
00225     dctcoef *block = (dctcoef*)_block;                                  \
00226     line_size /= sizeof(pixel);                                         \
00227                                                                         \
00228     for(i=0;i<8;i++) {                                                  \
00229         pixels[0] += block[0];                                          \
00230         pixels[1] += block[1];                                          \
00231         pixels[2] += block[2];                                          \
00232         pixels[3] += block[3];                                          \
00233         pixels[4] += block[4];                                          \
00234         pixels[5] += block[5];                                          \
00235         pixels[6] += block[6];                                          \
00236         pixels[7] += block[7];                                          \
00237         pixels += line_size;                                            \
00238         block += 8;                                                     \
00239     }                                                                   \
00240 }                                                                       \
00241                                                                         \
00242 static void FUNCC(add_pixels4 ## suffix)(uint8_t *restrict _pixels,     \
00243                                          DCTELEM *_block,               \
00244                                          int line_size)                 \
00245 {                                                                       \
00246     int i;                                                              \
00247     pixel *restrict pixels = (pixel *restrict)_pixels;                  \
00248     dctcoef *block = (dctcoef*)_block;                                  \
00249     line_size /= sizeof(pixel);                                         \
00250                                                                         \
00251     for(i=0;i<4;i++) {                                                  \
00252         pixels[0] += block[0];                                          \
00253         pixels[1] += block[1];                                          \
00254         pixels[2] += block[2];                                          \
00255         pixels[3] += block[3];                                          \
00256         pixels += line_size;                                            \
00257         block += 4;                                                     \
00258     }                                                                   \
00259 }                                                                       \
00260                                                                         \
00261 static void FUNCC(clear_block ## suffix)(DCTELEM *block)                \
00262 {                                                                       \
00263     memset(block, 0, sizeof(dctcoef)*64);                               \
00264 }                                                                       \
00265                                                                         \
00266                                                                      \
00269 static void FUNCC(clear_blocks ## suffix)(DCTELEM *blocks)              \
00270 {                                                                       \
00271     memset(blocks, 0, sizeof(dctcoef)*6*64);                            \
00272 }
00273 
00274 DCTELEM_FUNCS(DCTELEM, _16)
00275 #if BIT_DEPTH > 8
00276 DCTELEM_FUNCS(dctcoef, _32)
00277 #endif
00278 
00279 #define PIXOP2(OPNAME, OP) \
00280 static void FUNCC(OPNAME ## _pixels2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00281     int i;\
00282     for(i=0; i<h; i++){\
00283         OP(*((pixel2*)(block  )), AV_RN2P(pixels  ));\
00284         pixels+=line_size;\
00285         block +=line_size;\
00286     }\
00287 }\
00288 static void FUNCC(OPNAME ## _pixels4)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00289     int i;\
00290     for(i=0; i<h; i++){\
00291         OP(*((pixel4*)(block  )), AV_RN4P(pixels  ));\
00292         pixels+=line_size;\
00293         block +=line_size;\
00294     }\
00295 }\
00296 static void FUNCC(OPNAME ## _pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00297     int i;\
00298     for(i=0; i<h; i++){\
00299         OP(*((pixel4*)(block                )), AV_RN4P(pixels                ));\
00300         OP(*((pixel4*)(block+4*sizeof(pixel))), AV_RN4P(pixels+4*sizeof(pixel)));\
00301         pixels+=line_size;\
00302         block +=line_size;\
00303     }\
00304 }\
00305 static inline void FUNCC(OPNAME ## _no_rnd_pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00306     FUNCC(OPNAME ## _pixels8)(block, pixels, line_size, h);\
00307 }\
00308 \
00309 static inline void FUNC(OPNAME ## _no_rnd_pixels8_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00310                                                 int src_stride1, int src_stride2, int h){\
00311     int i;\
00312     for(i=0; i<h; i++){\
00313         pixel4 a,b;\
00314         a= AV_RN4P(&src1[i*src_stride1  ]);\
00315         b= AV_RN4P(&src2[i*src_stride2  ]);\
00316         OP(*((pixel4*)&dst[i*dst_stride  ]), no_rnd_avg_pixel4(a, b));\
00317         a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\
00318         b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\
00319         OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), no_rnd_avg_pixel4(a, b));\
00320     }\
00321 }\
00322 \
00323 static inline void FUNC(OPNAME ## _pixels8_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00324                                                 int src_stride1, int src_stride2, int h){\
00325     int i;\
00326     for(i=0; i<h; i++){\
00327         pixel4 a,b;\
00328         a= AV_RN4P(&src1[i*src_stride1  ]);\
00329         b= AV_RN4P(&src2[i*src_stride2  ]);\
00330         OP(*((pixel4*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\
00331         a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\
00332         b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\
00333         OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), rnd_avg_pixel4(a, b));\
00334     }\
00335 }\
00336 \
00337 static inline void FUNC(OPNAME ## _pixels4_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00338                                                 int src_stride1, int src_stride2, int h){\
00339     int i;\
00340     for(i=0; i<h; i++){\
00341         pixel4 a,b;\
00342         a= AV_RN4P(&src1[i*src_stride1  ]);\
00343         b= AV_RN4P(&src2[i*src_stride2  ]);\
00344         OP(*((pixel4*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\
00345     }\
00346 }\
00347 \
00348 static inline void FUNC(OPNAME ## _pixels2_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00349                                                 int src_stride1, int src_stride2, int h){\
00350     int i;\
00351     for(i=0; i<h; i++){\
00352         pixel4 a,b;\
00353         a= AV_RN2P(&src1[i*src_stride1  ]);\
00354         b= AV_RN2P(&src2[i*src_stride2  ]);\
00355         OP(*((pixel2*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\
00356     }\
00357 }\
00358 \
00359 static inline void FUNC(OPNAME ## _pixels16_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00360                                                 int src_stride1, int src_stride2, int h){\
00361     FUNC(OPNAME ## _pixels8_l2)(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\
00362     FUNC(OPNAME ## _pixels8_l2)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, h);\
00363 }\
00364 \
00365 static inline void FUNC(OPNAME ## _no_rnd_pixels16_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
00366                                                 int src_stride1, int src_stride2, int h){\
00367     FUNC(OPNAME ## _no_rnd_pixels8_l2)(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\
00368     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);\
00369 }\
00370 \
00371 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00372     FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00373 }\
00374 \
00375 static inline void FUNCC(OPNAME ## _pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00376     FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00377 }\
00378 \
00379 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00380     FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00381 }\
00382 \
00383 static inline void FUNCC(OPNAME ## _pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00384     FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00385 }\
00386 \
00387 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,\
00388                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00389     /* FIXME HIGH BIT DEPTH */\
00390     int i;\
00391     for(i=0; i<h; i++){\
00392         uint32_t a, b, c, d, l0, l1, h0, h1;\
00393         a= AV_RN32(&src1[i*src_stride1]);\
00394         b= AV_RN32(&src2[i*src_stride2]);\
00395         c= AV_RN32(&src3[i*src_stride3]);\
00396         d= AV_RN32(&src4[i*src_stride4]);\
00397         l0=  (a&0x03030303UL)\
00398            + (b&0x03030303UL)\
00399            + 0x02020202UL;\
00400         h0= ((a&0xFCFCFCFCUL)>>2)\
00401           + ((b&0xFCFCFCFCUL)>>2);\
00402         l1=  (c&0x03030303UL)\
00403            + (d&0x03030303UL);\
00404         h1= ((c&0xFCFCFCFCUL)>>2)\
00405           + ((d&0xFCFCFCFCUL)>>2);\
00406         OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00407         a= AV_RN32(&src1[i*src_stride1+4]);\
00408         b= AV_RN32(&src2[i*src_stride2+4]);\
00409         c= AV_RN32(&src3[i*src_stride3+4]);\
00410         d= AV_RN32(&src4[i*src_stride4+4]);\
00411         l0=  (a&0x03030303UL)\
00412            + (b&0x03030303UL)\
00413            + 0x02020202UL;\
00414         h0= ((a&0xFCFCFCFCUL)>>2)\
00415           + ((b&0xFCFCFCFCUL)>>2);\
00416         l1=  (c&0x03030303UL)\
00417            + (d&0x03030303UL);\
00418         h1= ((c&0xFCFCFCFCUL)>>2)\
00419           + ((d&0xFCFCFCFCUL)>>2);\
00420         OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00421     }\
00422 }\
00423 \
00424 static inline void FUNCC(OPNAME ## _pixels4_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00425     FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00426 }\
00427 \
00428 static inline void FUNCC(OPNAME ## _pixels4_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00429     FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00430 }\
00431 \
00432 static inline void FUNCC(OPNAME ## _pixels2_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00433     FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
00434 }\
00435 \
00436 static inline void FUNCC(OPNAME ## _pixels2_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
00437     FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
00438 }\
00439 \
00440 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,\
00441                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00442     /* FIXME HIGH BIT DEPTH*/\
00443     int i;\
00444     for(i=0; i<h; i++){\
00445         uint32_t a, b, c, d, l0, l1, h0, h1;\
00446         a= AV_RN32(&src1[i*src_stride1]);\
00447         b= AV_RN32(&src2[i*src_stride2]);\
00448         c= AV_RN32(&src3[i*src_stride3]);\
00449         d= AV_RN32(&src4[i*src_stride4]);\
00450         l0=  (a&0x03030303UL)\
00451            + (b&0x03030303UL)\
00452            + 0x01010101UL;\
00453         h0= ((a&0xFCFCFCFCUL)>>2)\
00454           + ((b&0xFCFCFCFCUL)>>2);\
00455         l1=  (c&0x03030303UL)\
00456            + (d&0x03030303UL);\
00457         h1= ((c&0xFCFCFCFCUL)>>2)\
00458           + ((d&0xFCFCFCFCUL)>>2);\
00459         OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00460         a= AV_RN32(&src1[i*src_stride1+4]);\
00461         b= AV_RN32(&src2[i*src_stride2+4]);\
00462         c= AV_RN32(&src3[i*src_stride3+4]);\
00463         d= AV_RN32(&src4[i*src_stride4+4]);\
00464         l0=  (a&0x03030303UL)\
00465            + (b&0x03030303UL)\
00466            + 0x01010101UL;\
00467         h0= ((a&0xFCFCFCFCUL)>>2)\
00468           + ((b&0xFCFCFCFCUL)>>2);\
00469         l1=  (c&0x03030303UL)\
00470            + (d&0x03030303UL);\
00471         h1= ((c&0xFCFCFCFCUL)>>2)\
00472           + ((d&0xFCFCFCFCUL)>>2);\
00473         OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00474     }\
00475 }\
00476 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,\
00477                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00478     FUNC(OPNAME ## _pixels8_l4)(dst  , src1  , src2  , src3  , src4  , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
00479     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);\
00480 }\
00481 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,\
00482                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
00483     FUNC(OPNAME ## _no_rnd_pixels8_l4)(dst  , src1  , src2  , src3  , src4  , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
00484     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);\
00485 }\
00486 \
00487 static inline void FUNCC(OPNAME ## _pixels2_xy2)(uint8_t *p_block, const uint8_t *p_pixels, int line_size, int h)\
00488 {\
00489         int i, a0, b0, a1, b1;\
00490         pixel *block = (pixel*)p_block;\
00491         const pixel *pixels = (const pixel*)p_pixels;\
00492         line_size >>= sizeof(pixel)-1;\
00493         a0= pixels[0];\
00494         b0= pixels[1] + 2;\
00495         a0 += b0;\
00496         b0 += pixels[2];\
00497 \
00498         pixels+=line_size;\
00499         for(i=0; i<h; i+=2){\
00500             a1= pixels[0];\
00501             b1= pixels[1];\
00502             a1 += b1;\
00503             b1 += pixels[2];\
00504 \
00505             block[0]= (a1+a0)>>2; /* FIXME non put */\
00506             block[1]= (b1+b0)>>2;\
00507 \
00508             pixels+=line_size;\
00509             block +=line_size;\
00510 \
00511             a0= pixels[0];\
00512             b0= pixels[1] + 2;\
00513             a0 += b0;\
00514             b0 += pixels[2];\
00515 \
00516             block[0]= (a1+a0)>>2;\
00517             block[1]= (b1+b0)>>2;\
00518             pixels+=line_size;\
00519             block +=line_size;\
00520         }\
00521 }\
00522 \
00523 static inline void FUNCC(OPNAME ## _pixels4_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
00524 {\
00525         /* FIXME HIGH BIT DEPTH */\
00526         int i;\
00527         const uint32_t a= AV_RN32(pixels  );\
00528         const uint32_t b= AV_RN32(pixels+1);\
00529         uint32_t l0=  (a&0x03030303UL)\
00530                     + (b&0x03030303UL)\
00531                     + 0x02020202UL;\
00532         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
00533                    + ((b&0xFCFCFCFCUL)>>2);\
00534         uint32_t l1,h1;\
00535 \
00536         pixels+=line_size;\
00537         for(i=0; i<h; i+=2){\
00538             uint32_t a= AV_RN32(pixels  );\
00539             uint32_t b= AV_RN32(pixels+1);\
00540             l1=  (a&0x03030303UL)\
00541                + (b&0x03030303UL);\
00542             h1= ((a&0xFCFCFCFCUL)>>2)\
00543               + ((b&0xFCFCFCFCUL)>>2);\
00544             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00545             pixels+=line_size;\
00546             block +=line_size;\
00547             a= AV_RN32(pixels  );\
00548             b= AV_RN32(pixels+1);\
00549             l0=  (a&0x03030303UL)\
00550                + (b&0x03030303UL)\
00551                + 0x02020202UL;\
00552             h0= ((a&0xFCFCFCFCUL)>>2)\
00553               + ((b&0xFCFCFCFCUL)>>2);\
00554             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00555             pixels+=line_size;\
00556             block +=line_size;\
00557         }\
00558 }\
00559 \
00560 static inline void FUNCC(OPNAME ## _pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
00561 {\
00562     /* FIXME HIGH BIT DEPTH */\
00563     int j;\
00564     for(j=0; j<2; j++){\
00565         int i;\
00566         const uint32_t a= AV_RN32(pixels  );\
00567         const uint32_t b= AV_RN32(pixels+1);\
00568         uint32_t l0=  (a&0x03030303UL)\
00569                     + (b&0x03030303UL)\
00570                     + 0x02020202UL;\
00571         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
00572                    + ((b&0xFCFCFCFCUL)>>2);\
00573         uint32_t l1,h1;\
00574 \
00575         pixels+=line_size;\
00576         for(i=0; i<h; i+=2){\
00577             uint32_t a= AV_RN32(pixels  );\
00578             uint32_t b= AV_RN32(pixels+1);\
00579             l1=  (a&0x03030303UL)\
00580                + (b&0x03030303UL);\
00581             h1= ((a&0xFCFCFCFCUL)>>2)\
00582               + ((b&0xFCFCFCFCUL)>>2);\
00583             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00584             pixels+=line_size;\
00585             block +=line_size;\
00586             a= AV_RN32(pixels  );\
00587             b= AV_RN32(pixels+1);\
00588             l0=  (a&0x03030303UL)\
00589                + (b&0x03030303UL)\
00590                + 0x02020202UL;\
00591             h0= ((a&0xFCFCFCFCUL)>>2)\
00592               + ((b&0xFCFCFCFCUL)>>2);\
00593             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00594             pixels+=line_size;\
00595             block +=line_size;\
00596         }\
00597         pixels+=4-line_size*(h+1);\
00598         block +=4-line_size*h;\
00599     }\
00600 }\
00601 \
00602 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
00603 {\
00604     /* FIXME HIGH BIT DEPTH */\
00605     int j;\
00606     for(j=0; j<2; j++){\
00607         int i;\
00608         const uint32_t a= AV_RN32(pixels  );\
00609         const uint32_t b= AV_RN32(pixels+1);\
00610         uint32_t l0=  (a&0x03030303UL)\
00611                     + (b&0x03030303UL)\
00612                     + 0x01010101UL;\
00613         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
00614                    + ((b&0xFCFCFCFCUL)>>2);\
00615         uint32_t l1,h1;\
00616 \
00617         pixels+=line_size;\
00618         for(i=0; i<h; i+=2){\
00619             uint32_t a= AV_RN32(pixels  );\
00620             uint32_t b= AV_RN32(pixels+1);\
00621             l1=  (a&0x03030303UL)\
00622                + (b&0x03030303UL);\
00623             h1= ((a&0xFCFCFCFCUL)>>2)\
00624               + ((b&0xFCFCFCFCUL)>>2);\
00625             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00626             pixels+=line_size;\
00627             block +=line_size;\
00628             a= AV_RN32(pixels  );\
00629             b= AV_RN32(pixels+1);\
00630             l0=  (a&0x03030303UL)\
00631                + (b&0x03030303UL)\
00632                + 0x01010101UL;\
00633             h0= ((a&0xFCFCFCFCUL)>>2)\
00634               + ((b&0xFCFCFCFCUL)>>2);\
00635             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
00636             pixels+=line_size;\
00637             block +=line_size;\
00638         }\
00639         pixels+=4-line_size*(h+1);\
00640         block +=4-line_size*h;\
00641     }\
00642 }\
00643 \
00644 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16)    , FUNCC(OPNAME ## _pixels8)    , 8*sizeof(pixel))\
00645 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_x2) , FUNCC(OPNAME ## _pixels8_x2) , 8*sizeof(pixel))\
00646 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_y2) , FUNCC(OPNAME ## _pixels8_y2) , 8*sizeof(pixel))\
00647 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_xy2), FUNCC(OPNAME ## _pixels8_xy2), 8*sizeof(pixel))\
00648 av_unused CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16)    , FUNCC(OPNAME ## _pixels8) , 8*sizeof(pixel))\
00649 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_x2) , FUNCC(OPNAME ## _no_rnd_pixels8_x2) , 8*sizeof(pixel))\
00650 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_y2) , FUNCC(OPNAME ## _no_rnd_pixels8_y2) , 8*sizeof(pixel))\
00651 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_xy2), FUNCC(OPNAME ## _no_rnd_pixels8_xy2), 8*sizeof(pixel))\
00652 
00653 #define op_avg(a, b) a = rnd_avg_pixel4(a, b)
00654 #define op_put(a, b) a = b
00655 
00656 PIXOP2(avg, op_avg)
00657 PIXOP2(put, op_put)
00658 #undef op_avg
00659 #undef op_put
00660 
00661 #define put_no_rnd_pixels8_c  put_pixels8_c
00662 #define put_no_rnd_pixels16_c put_pixels16_c
00663 
00664 static void FUNCC(put_no_rnd_pixels16_l2)(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){
00665     FUNC(put_no_rnd_pixels16_l2)(dst, a, b, stride, stride, stride, h);
00666 }
00667 
00668 static void FUNCC(put_no_rnd_pixels8_l2)(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){
00669     FUNC(put_no_rnd_pixels8_l2)(dst, a, b, stride, stride, stride, h);
00670 }
00671 
00672 #define H264_CHROMA_MC(OPNAME, OP)\
00673 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){\
00674     pixel *dst = (pixel*)p_dst;\
00675     pixel *src = (pixel*)p_src;\
00676     const int A=(8-x)*(8-y);\
00677     const int B=(  x)*(8-y);\
00678     const int C=(8-x)*(  y);\
00679     const int D=(  x)*(  y);\
00680     int i;\
00681     stride >>= sizeof(pixel)-1;\
00682     \
00683     assert(x<8 && y<8 && x>=0 && y>=0);\
00684 \
00685     if(D){\
00686         for(i=0; i<h; i++){\
00687             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
00688             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
00689             dst+= stride;\
00690             src+= stride;\
00691         }\
00692     }else{\
00693         const int E= B+C;\
00694         const int step= C ? stride : 1;\
00695         for(i=0; i<h; i++){\
00696             OP(dst[0], (A*src[0] + E*src[step+0]));\
00697             OP(dst[1], (A*src[1] + E*src[step+1]));\
00698             dst+= stride;\
00699             src+= stride;\
00700         }\
00701     }\
00702 }\
00703 \
00704 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){\
00705     pixel *dst = (pixel*)p_dst;\
00706     pixel *src = (pixel*)p_src;\
00707     const int A=(8-x)*(8-y);\
00708     const int B=(  x)*(8-y);\
00709     const int C=(8-x)*(  y);\
00710     const int D=(  x)*(  y);\
00711     int i;\
00712     stride >>= sizeof(pixel)-1;\
00713     \
00714     assert(x<8 && y<8 && x>=0 && y>=0);\
00715 \
00716     if(D){\
00717         for(i=0; i<h; i++){\
00718             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
00719             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
00720             OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\
00721             OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\
00722             dst+= stride;\
00723             src+= stride;\
00724         }\
00725     }else{\
00726         const int E= B+C;\
00727         const int step= C ? stride : 1;\
00728         for(i=0; i<h; i++){\
00729             OP(dst[0], (A*src[0] + E*src[step+0]));\
00730             OP(dst[1], (A*src[1] + E*src[step+1]));\
00731             OP(dst[2], (A*src[2] + E*src[step+2]));\
00732             OP(dst[3], (A*src[3] + E*src[step+3]));\
00733             dst+= stride;\
00734             src+= stride;\
00735         }\
00736     }\
00737 }\
00738 \
00739 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){\
00740     pixel *dst = (pixel*)p_dst;\
00741     pixel *src = (pixel*)p_src;\
00742     const int A=(8-x)*(8-y);\
00743     const int B=(  x)*(8-y);\
00744     const int C=(8-x)*(  y);\
00745     const int D=(  x)*(  y);\
00746     int i;\
00747     stride >>= sizeof(pixel)-1;\
00748     \
00749     assert(x<8 && y<8 && x>=0 && y>=0);\
00750 \
00751     if(D){\
00752         for(i=0; i<h; i++){\
00753             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
00754             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
00755             OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\
00756             OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\
00757             OP(dst[4], (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5]));\
00758             OP(dst[5], (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6]));\
00759             OP(dst[6], (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7]));\
00760             OP(dst[7], (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8]));\
00761             dst+= stride;\
00762             src+= stride;\
00763         }\
00764     }else{\
00765         const int E= B+C;\
00766         const int step= C ? stride : 1;\
00767         for(i=0; i<h; i++){\
00768             OP(dst[0], (A*src[0] + E*src[step+0]));\
00769             OP(dst[1], (A*src[1] + E*src[step+1]));\
00770             OP(dst[2], (A*src[2] + E*src[step+2]));\
00771             OP(dst[3], (A*src[3] + E*src[step+3]));\
00772             OP(dst[4], (A*src[4] + E*src[step+4]));\
00773             OP(dst[5], (A*src[5] + E*src[step+5]));\
00774             OP(dst[6], (A*src[6] + E*src[step+6]));\
00775             OP(dst[7], (A*src[7] + E*src[step+7]));\
00776             dst+= stride;\
00777             src+= stride;\
00778         }\
00779     }\
00780 }
00781 
00782 #define op_avg(a, b) a = (((a)+(((b) + 32)>>6)+1)>>1)
00783 #define op_put(a, b) a = (((b) + 32)>>6)
00784 
00785 H264_CHROMA_MC(put_       , op_put)
00786 H264_CHROMA_MC(avg_       , op_avg)
00787 #undef op_avg
00788 #undef op_put
00789 
00790 #define H264_LOWPASS(OPNAME, OP, OP2) \
00791 static av_unused void FUNC(OPNAME ## h264_qpel2_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00792     const int h=2;\
00793     INIT_CLIP\
00794     int i;\
00795     pixel *dst = (pixel*)p_dst;\
00796     pixel *src = (pixel*)p_src;\
00797     dstStride >>= sizeof(pixel)-1;\
00798     srcStride >>= sizeof(pixel)-1;\
00799     for(i=0; i<h; i++)\
00800     {\
00801         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
00802         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
00803         dst+=dstStride;\
00804         src+=srcStride;\
00805     }\
00806 }\
00807 \
00808 static av_unused void FUNC(OPNAME ## h264_qpel2_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00809     const int w=2;\
00810     INIT_CLIP\
00811     int i;\
00812     pixel *dst = (pixel*)p_dst;\
00813     pixel *src = (pixel*)p_src;\
00814     dstStride >>= sizeof(pixel)-1;\
00815     srcStride >>= sizeof(pixel)-1;\
00816     for(i=0; i<w; i++)\
00817     {\
00818         const int srcB= src[-2*srcStride];\
00819         const int srcA= src[-1*srcStride];\
00820         const int src0= src[0 *srcStride];\
00821         const int src1= src[1 *srcStride];\
00822         const int src2= src[2 *srcStride];\
00823         const int src3= src[3 *srcStride];\
00824         const int src4= src[4 *srcStride];\
00825         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
00826         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
00827         dst++;\
00828         src++;\
00829     }\
00830 }\
00831 \
00832 static av_unused void FUNC(OPNAME ## h264_qpel2_hv_lowpass)(uint8_t *p_dst, int16_t *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\
00833     const int h=2;\
00834     const int w=2;\
00835     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
00836     INIT_CLIP\
00837     int i;\
00838     pixel *dst = (pixel*)p_dst;\
00839     pixel *src = (pixel*)p_src;\
00840     dstStride >>= sizeof(pixel)-1;\
00841     srcStride >>= sizeof(pixel)-1;\
00842     src -= 2*srcStride;\
00843     for(i=0; i<h+5; i++)\
00844     {\
00845         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
00846         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
00847         tmp+=tmpStride;\
00848         src+=srcStride;\
00849     }\
00850     tmp -= tmpStride*(h+5-2);\
00851     for(i=0; i<w; i++)\
00852     {\
00853         const int tmpB= tmp[-2*tmpStride] - pad;\
00854         const int tmpA= tmp[-1*tmpStride] - pad;\
00855         const int tmp0= tmp[0 *tmpStride] - pad;\
00856         const int tmp1= tmp[1 *tmpStride] - pad;\
00857         const int tmp2= tmp[2 *tmpStride] - pad;\
00858         const int tmp3= tmp[3 *tmpStride] - pad;\
00859         const int tmp4= tmp[4 *tmpStride] - pad;\
00860         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
00861         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
00862         dst++;\
00863         tmp++;\
00864     }\
00865 }\
00866 static void FUNC(OPNAME ## h264_qpel4_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00867     const int h=4;\
00868     INIT_CLIP\
00869     int i;\
00870     pixel *dst = (pixel*)p_dst;\
00871     pixel *src = (pixel*)p_src;\
00872     dstStride >>= sizeof(pixel)-1;\
00873     srcStride >>= sizeof(pixel)-1;\
00874     for(i=0; i<h; i++)\
00875     {\
00876         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
00877         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
00878         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]));\
00879         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]));\
00880         dst+=dstStride;\
00881         src+=srcStride;\
00882     }\
00883 }\
00884 \
00885 static void FUNC(OPNAME ## h264_qpel4_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00886     const int w=4;\
00887     INIT_CLIP\
00888     int i;\
00889     pixel *dst = (pixel*)p_dst;\
00890     pixel *src = (pixel*)p_src;\
00891     dstStride >>= sizeof(pixel)-1;\
00892     srcStride >>= sizeof(pixel)-1;\
00893     for(i=0; i<w; i++)\
00894     {\
00895         const int srcB= src[-2*srcStride];\
00896         const int srcA= src[-1*srcStride];\
00897         const int src0= src[0 *srcStride];\
00898         const int src1= src[1 *srcStride];\
00899         const int src2= src[2 *srcStride];\
00900         const int src3= src[3 *srcStride];\
00901         const int src4= src[4 *srcStride];\
00902         const int src5= src[5 *srcStride];\
00903         const int src6= src[6 *srcStride];\
00904         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
00905         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
00906         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
00907         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
00908         dst++;\
00909         src++;\
00910     }\
00911 }\
00912 \
00913 static void FUNC(OPNAME ## h264_qpel4_hv_lowpass)(uint8_t *p_dst, int16_t *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\
00914     const int h=4;\
00915     const int w=4;\
00916     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
00917     INIT_CLIP\
00918     int i;\
00919     pixel *dst = (pixel*)p_dst;\
00920     pixel *src = (pixel*)p_src;\
00921     dstStride >>= sizeof(pixel)-1;\
00922     srcStride >>= sizeof(pixel)-1;\
00923     src -= 2*srcStride;\
00924     for(i=0; i<h+5; i++)\
00925     {\
00926         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
00927         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
00928         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]) + pad;\
00929         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]) + pad;\
00930         tmp+=tmpStride;\
00931         src+=srcStride;\
00932     }\
00933     tmp -= tmpStride*(h+5-2);\
00934     for(i=0; i<w; i++)\
00935     {\
00936         const int tmpB= tmp[-2*tmpStride] - pad;\
00937         const int tmpA= tmp[-1*tmpStride] - pad;\
00938         const int tmp0= tmp[0 *tmpStride] - pad;\
00939         const int tmp1= tmp[1 *tmpStride] - pad;\
00940         const int tmp2= tmp[2 *tmpStride] - pad;\
00941         const int tmp3= tmp[3 *tmpStride] - pad;\
00942         const int tmp4= tmp[4 *tmpStride] - pad;\
00943         const int tmp5= tmp[5 *tmpStride] - pad;\
00944         const int tmp6= tmp[6 *tmpStride] - pad;\
00945         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
00946         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
00947         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
00948         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
00949         dst++;\
00950         tmp++;\
00951     }\
00952 }\
00953 \
00954 static void FUNC(OPNAME ## h264_qpel8_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00955     const int h=8;\
00956     INIT_CLIP\
00957     int i;\
00958     pixel *dst = (pixel*)p_dst;\
00959     pixel *src = (pixel*)p_src;\
00960     dstStride >>= sizeof(pixel)-1;\
00961     srcStride >>= sizeof(pixel)-1;\
00962     for(i=0; i<h; i++)\
00963     {\
00964         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]));\
00965         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]));\
00966         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]));\
00967         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]));\
00968         OP(dst[4], (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]));\
00969         OP(dst[5], (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]));\
00970         OP(dst[6], (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]));\
00971         OP(dst[7], (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]));\
00972         dst+=dstStride;\
00973         src+=srcStride;\
00974     }\
00975 }\
00976 \
00977 static void FUNC(OPNAME ## h264_qpel8_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\
00978     const int w=8;\
00979     INIT_CLIP\
00980     int i;\
00981     pixel *dst = (pixel*)p_dst;\
00982     pixel *src = (pixel*)p_src;\
00983     dstStride >>= sizeof(pixel)-1;\
00984     srcStride >>= sizeof(pixel)-1;\
00985     for(i=0; i<w; i++)\
00986     {\
00987         const int srcB= src[-2*srcStride];\
00988         const int srcA= src[-1*srcStride];\
00989         const int src0= src[0 *srcStride];\
00990         const int src1= src[1 *srcStride];\
00991         const int src2= src[2 *srcStride];\
00992         const int src3= src[3 *srcStride];\
00993         const int src4= src[4 *srcStride];\
00994         const int src5= src[5 *srcStride];\
00995         const int src6= src[6 *srcStride];\
00996         const int src7= src[7 *srcStride];\
00997         const int src8= src[8 *srcStride];\
00998         const int src9= src[9 *srcStride];\
00999         const int src10=src[10*srcStride];\
01000         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
01001         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
01002         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
01003         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
01004         OP(dst[4*dstStride], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\
01005         OP(dst[5*dstStride], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\
01006         OP(dst[6*dstStride], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\
01007         OP(dst[7*dstStride], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\
01008         dst++;\
01009         src++;\
01010     }\
01011 }\
01012 \
01013 static void FUNC(OPNAME ## h264_qpel8_hv_lowpass)(uint8_t *p_dst, int16_t *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\
01014     const int h=8;\
01015     const int w=8;\
01016     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
01017     INIT_CLIP\
01018     int i;\
01019     pixel *dst = (pixel*)p_dst;\
01020     pixel *src = (pixel*)p_src;\
01021     dstStride >>= sizeof(pixel)-1;\
01022     srcStride >>= sizeof(pixel)-1;\
01023     src -= 2*srcStride;\
01024     for(i=0; i<h+5; i++)\
01025     {\
01026         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]) + pad;\
01027         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]) + pad;\
01028         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]) + pad;\
01029         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]) + pad;\
01030         tmp[4]= (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]) + pad;\
01031         tmp[5]= (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]) + pad;\
01032         tmp[6]= (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]) + pad;\
01033         tmp[7]= (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]) + pad;\
01034         tmp+=tmpStride;\
01035         src+=srcStride;\
01036     }\
01037     tmp -= tmpStride*(h+5-2);\
01038     for(i=0; i<w; i++)\
01039     {\
01040         const int tmpB= tmp[-2*tmpStride] - pad;\
01041         const int tmpA= tmp[-1*tmpStride] - pad;\
01042         const int tmp0= tmp[0 *tmpStride] - pad;\
01043         const int tmp1= tmp[1 *tmpStride] - pad;\
01044         const int tmp2= tmp[2 *tmpStride] - pad;\
01045         const int tmp3= tmp[3 *tmpStride] - pad;\
01046         const int tmp4= tmp[4 *tmpStride] - pad;\
01047         const int tmp5= tmp[5 *tmpStride] - pad;\
01048         const int tmp6= tmp[6 *tmpStride] - pad;\
01049         const int tmp7= tmp[7 *tmpStride] - pad;\
01050         const int tmp8= tmp[8 *tmpStride] - pad;\
01051         const int tmp9= tmp[9 *tmpStride] - pad;\
01052         const int tmp10=tmp[10*tmpStride] - pad;\
01053         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
01054         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
01055         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
01056         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
01057         OP2(dst[4*dstStride], (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));\
01058         OP2(dst[5*dstStride], (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));\
01059         OP2(dst[6*dstStride], (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));\
01060         OP2(dst[7*dstStride], (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));\
01061         dst++;\
01062         tmp++;\
01063     }\
01064 }\
01065 \
01066 static void FUNC(OPNAME ## h264_qpel16_v_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
01067     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\
01068     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01069     src += 8*srcStride;\
01070     dst += 8*dstStride;\
01071     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\
01072     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01073 }\
01074 \
01075 static void FUNC(OPNAME ## h264_qpel16_h_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
01076     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\
01077     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01078     src += 8*srcStride;\
01079     dst += 8*dstStride;\
01080     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\
01081     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
01082 }\
01083 \
01084 static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
01085     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\
01086     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
01087     src += 8*srcStride;\
01088     dst += 8*dstStride;\
01089     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\
01090     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
01091 }\
01092 
01093 #define H264_MC(OPNAME, SIZE) \
01094 static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\
01095     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\
01096 }\
01097 \
01098 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\
01099     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01100     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
01101     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\
01102 }\
01103 \
01104 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\
01105     FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\
01106 }\
01107 \
01108 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\
01109     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01110     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
01111     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\
01112 }\
01113 \
01114 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\
01115     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01116     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01117     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01118     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01119     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01120     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01121 }\
01122 \
01123 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\
01124     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01125     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01126     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01127     FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\
01128 }\
01129 \
01130 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\
01131     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01132     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01133     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
01134     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01135     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01136     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01137 }\
01138 \
01139 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(uint8_t *dst, uint8_t *src, int stride){\
01140     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01141     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01142     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01143     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01144     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
01145     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01146     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01147     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01148 }\
01149 \
01150 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\
01151     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01152     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01153     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01154     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01155     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
01156     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
01157     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01158     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01159 }\
01160 \
01161 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\
01162     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01163     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01164     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01165     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01166     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
01167     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01168     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01169     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01170 }\
01171 \
01172 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\
01173     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01174     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01175     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01176     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01177     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
01178     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
01179     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01180     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01181 }\
01182 \
01183 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\
01184     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01185     FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\
01186 }\
01187 \
01188 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\
01189     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01190     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01191     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01192     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
01193     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01194     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01195 }\
01196 \
01197 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\
01198     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01199     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
01200     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01201     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
01202     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01203     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01204 }\
01205 \
01206 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\
01207     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01208     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01209     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01210     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01211     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01212     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
01213     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01214     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01215     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01216 }\
01217 \
01218 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\
01219     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
01220     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
01221     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
01222     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
01223     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
01224     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
01225     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
01226     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
01227     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
01228 }\
01229 
01230 #define op_avg(a, b)  a = (((a)+CLIP(((b) + 16)>>5)+1)>>1)
01231 //#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)
01232 #define op_put(a, b)  a = CLIP(((b) + 16)>>5)
01233 #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)
01234 #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)
01235 
01236 H264_LOWPASS(put_       , op_put, op2_put)
01237 H264_LOWPASS(avg_       , op_avg, op2_avg)
01238 H264_MC(put_, 2)
01239 H264_MC(put_, 4)
01240 H264_MC(put_, 8)
01241 H264_MC(put_, 16)
01242 H264_MC(avg_, 4)
01243 H264_MC(avg_, 8)
01244 H264_MC(avg_, 16)
01245 
01246 #undef op_avg
01247 #undef op_put
01248 #undef op2_avg
01249 #undef op2_put
01250 
01251 #if BIT_DEPTH == 8
01252 #   define put_h264_qpel8_mc00_8_c  ff_put_pixels8x8_8_c
01253 #   define avg_h264_qpel8_mc00_8_c  ff_avg_pixels8x8_8_c
01254 #   define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c
01255 #   define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c
01256 #elif BIT_DEPTH == 9
01257 #   define put_h264_qpel8_mc00_9_c  ff_put_pixels8x8_9_c
01258 #   define avg_h264_qpel8_mc00_9_c  ff_avg_pixels8x8_9_c
01259 #   define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c
01260 #   define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c
01261 #elif BIT_DEPTH == 10
01262 #   define put_h264_qpel8_mc00_10_c  ff_put_pixels8x8_10_c
01263 #   define avg_h264_qpel8_mc00_10_c  ff_avg_pixels8x8_10_c
01264 #   define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c
01265 #   define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c
01266 #endif
01267 
01268 void FUNCC(ff_put_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {
01269     FUNCC(put_pixels8)(dst, src, stride, 8);
01270 }
01271 void FUNCC(ff_avg_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {
01272     FUNCC(avg_pixels8)(dst, src, stride, 8);
01273 }
01274 void FUNCC(ff_put_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {
01275     FUNCC(put_pixels16)(dst, src, stride, 16);
01276 }
01277 void FUNCC(ff_avg_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {
01278     FUNCC(avg_pixels16)(dst, src, stride, 16);
01279 }
01280