falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
46919 Commits
32 Branches
286MB
Tree: 07a866282f
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '07a866282f'
${ noResults }
FFmpeg/libavcodec/dsputil_template.c

1293 lines
56KB
Raw Blame History 

  1. /*

  2.  * DSP utils

  3.  * Copyright (c) 2000, 2001 Fabrice Bellard

  4.  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>

  5.  *

  6.  * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>

  7.  *

  8.  * This file is part of FFmpeg.

  9.  *

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

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

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

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

  14.  *

  15.  * FFmpeg is distributed in the hope that it will be useful,

  16.  * but WITHOUT ANY WARRANTY; without even the implied warranty of

  17.  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU

  18.  * Lesser General Public License for more details.

  19.  *

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

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

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

  23.  */

  24. 

  25. /**

  26.  * @file

  27.  * DSP utils

  28.  */

  29. 

  30. #include "bit_depth_template.c"

  31. 

  32. static inline void FUNC(copy_block2)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)

  33. {

  34.     int i;

  35.     for(i=0; i<h; i++)

  36.     {

  37.         AV_WN2P(dst   , AV_RN2P(src   ));

  38.         dst+=dstStride;

  39.         src+=srcStride;

  40.     }

  41. }

  42. 

  43. static inline void FUNC(copy_block4)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)

  44. {

  45.     int i;

  46.     for(i=0; i<h; i++)

  47.     {

  48.         AV_WN4P(dst   , AV_RN4P(src   ));

  49.         dst+=dstStride;

  50.         src+=srcStride;

  51.     }

  52. }

  53. 

  54. static inline void FUNC(copy_block8)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)

  55. {

  56.     int i;

  57.     for(i=0; i<h; i++)

  58.     {

  59.         AV_WN4P(dst                , AV_RN4P(src                ));

  60.         AV_WN4P(dst+4*sizeof(pixel), AV_RN4P(src+4*sizeof(pixel)));

  61.         dst+=dstStride;

  62.         src+=srcStride;

  63.     }

  64. }

  65. 

  66. static inline void FUNC(copy_block16)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)

  67. {

  68.     int i;

  69.     for(i=0; i<h; i++)

  70.     {

  71.         AV_WN4P(dst                 , AV_RN4P(src                 ));

  72.         AV_WN4P(dst+ 4*sizeof(pixel), AV_RN4P(src+ 4*sizeof(pixel)));

  73.         AV_WN4P(dst+ 8*sizeof(pixel), AV_RN4P(src+ 8*sizeof(pixel)));

  74.         AV_WN4P(dst+12*sizeof(pixel), AV_RN4P(src+12*sizeof(pixel)));

  75.         dst+=dstStride;

  76.         src+=srcStride;

  77.     }

  78. }

  79. 

  80. /* draw the edges of width 'w' of an image of size width, height */

  81. //FIXME check that this is ok for mpeg4 interlaced

  82. static void FUNCC(draw_edges)(uint8_t *p_buf, int p_wrap, int width, int height, int w, int h, int sides)

  83. {

  84.     pixel *buf = (pixel*)p_buf;

  85.     int wrap = p_wrap / sizeof(pixel);

  86.     pixel *ptr, *last_line;

  87.     int i;

  88. 

  89.     /* left and right */

  90.     ptr = buf;

  91.     for(i=0;i<height;i++) {

  92. #if BIT_DEPTH > 8

  93.         int j;

  94.         for (j = 0; j < w; j++) {

  95.             ptr[j-w] = ptr[0];

  96.             ptr[j+width] = ptr[width-1];

  97.         }

  98. #else

  99.         memset(ptr - w, ptr[0], w);

  100.         memset(ptr + width, ptr[width-1], w);

  101. #endif

  102.         ptr += wrap;

  103.     }

  104. 

  105.     /* top and bottom + corners */

  106.     buf -= w;

  107.     last_line = buf + (height - 1) * wrap;

  108.     if (sides & EDGE_TOP)

  109.         for(i = 0; i < h; i++)

  110.             memcpy(buf - (i + 1) * wrap, buf, (width + w + w) * sizeof(pixel)); // top

  111.     if (sides & EDGE_BOTTOM)

  112.         for (i = 0; i < h; i++)

  113.             memcpy(last_line + (i + 1) * wrap, last_line, (width + w + w) * sizeof(pixel)); // bottom

  114. }

  115. 

  116. /**

  117.  * Copy a rectangular area of samples to a temporary buffer and replicate the border samples.

  118.  * @param buf destination buffer

  119.  * @param src source buffer

  120.  * @param linesize number of bytes between 2 vertically adjacent samples in both the source and destination buffers

  121.  * @param block_w width of block

  122.  * @param block_h height of block

  123.  * @param src_x x coordinate of the top left sample of the block in the source buffer

  124.  * @param src_y y coordinate of the top left sample of the block in the source buffer

  125.  * @param w width of the source buffer

  126.  * @param h height of the source buffer

  127.  */

  128. void FUNC(ff_emulated_edge_mc)(uint8_t *buf, const uint8_t *src, int linesize, int block_w, int block_h,

  129.                                     int src_x, int src_y, int w, int h){

  130.     int x, y;

  131.     int start_y, start_x, end_y, end_x;

  132. 

  133.     if(src_y>= h){

  134.         src-= src_y*linesize;

  135.         src+= (h-1)*linesize;

  136.         src_y=h-1;

  137.     }else if(src_y<=-block_h){

  138.         src-= src_y*linesize;

  139.         src+= (1-block_h)*linesize;

  140.         src_y=1-block_h;

  141.     }

  142.     if(src_x>= w){

  143.         src+= (w-1-src_x)*sizeof(pixel);

  144.         src_x=w-1;

  145.     }else if(src_x<=-block_w){

  146.         src+= (1-block_w-src_x)*sizeof(pixel);

  147.         src_x=1-block_w;

  148.     }

  149. 

  150.     start_y= FFMAX(0, -src_y);

  151.     start_x= FFMAX(0, -src_x);

  152.     end_y= FFMIN(block_h, h-src_y);

  153.     end_x= FFMIN(block_w, w-src_x);

  154.     av_assert2(start_y < end_y && block_h);

  155.     av_assert2(start_x < end_x && block_w);

  156. 

  157.     w    = end_x - start_x;

  158.     src += start_y*linesize + start_x*sizeof(pixel);

  159.     buf += start_x*sizeof(pixel);

  160. 

  161.     //top

  162.     for(y=0; y<start_y; y++){

  163.         memcpy(buf, src, w*sizeof(pixel));

  164.         buf += linesize;

  165.     }

  166. 

  167.     // copy existing part

  168.     for(; y<end_y; y++){

  169.         memcpy(buf, src, w*sizeof(pixel));

  170.         src += linesize;

  171.         buf += linesize;

  172.     }

  173. 

  174.     //bottom

  175.     src -= linesize;

  176.     for(; y<block_h; y++){

  177.         memcpy(buf, src, w*sizeof(pixel));

  178.         buf += linesize;

  179.     }

  180. 

  181.     buf -= block_h * linesize + start_x*sizeof(pixel);

  182.     while (block_h--){

  183.         pixel *bufp = (pixel*)buf;

  184.        //left

  185.         for(x=0; x<start_x; x++){

  186.             bufp[x] = bufp[start_x];

  187.         }

  188. 

  189.        //right

  190.         for(x=end_x; x<block_w; x++){

  191.             bufp[x] = bufp[end_x - 1];

  192.         }

  193.         buf += linesize;

  194.     }

  195. }

  196. 

  197. #define DCTELEM_FUNCS(dctcoef, suffix)                                  \

  198. static void FUNCC(get_pixels ## suffix)(DCTELEM *av_restrict _block,    \

  199.                                         const uint8_t *_pixels,         \

  200.                                         int line_size)                  \

  201. {                                                                       \

  202.     const pixel *pixels = (const pixel *) _pixels;                      \

  203.     dctcoef *av_restrict block = (dctcoef *) _block;                    \

  204.     int i;                                                              \

  205.                                                                         \

  206.     /* read the pixels */                                               \

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

  208.         block[0] = pixels[0];                                           \

  209.         block[1] = pixels[1];                                           \

  210.         block[2] = pixels[2];                                           \

  211.         block[3] = pixels[3];                                           \

  212.         block[4] = pixels[4];                                           \

  213.         block[5] = pixels[5];                                           \

  214.         block[6] = pixels[6];                                           \

  215.         block[7] = pixels[7];                                           \

  216.         pixels += line_size / sizeof(pixel);                            \

  217.         block += 8;                                                     \

  218.     }                                                                   \

  219. }                                                                       \

  220.                                                                         \

  221. static void FUNCC(add_pixels8 ## suffix)(uint8_t *av_restrict _pixels,  \

  222.                                          DCTELEM *_block,               \

  223.                                          int line_size)                 \

  224. {                                                                       \

  225.     int i;                                                              \

  226.     pixel *av_restrict pixels = (pixel *av_restrict)_pixels;            \

  227.     dctcoef *block = (dctcoef*)_block;                                  \

  228.     line_size /= sizeof(pixel);                                         \

  229.                                                                         \

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

  231.         pixels[0] += block[0];                                          \

  232.         pixels[1] += block[1];                                          \

  233.         pixels[2] += block[2];                                          \

  234.         pixels[3] += block[3];                                          \

  235.         pixels[4] += block[4];                                          \

  236.         pixels[5] += block[5];                                          \

  237.         pixels[6] += block[6];                                          \

  238.         pixels[7] += block[7];                                          \

  239.         pixels += line_size;                                            \

  240.         block += 8;                                                     \

  241.     }                                                                   \

  242. }                                                                       \

  243.                                                                         \

  244. static void FUNCC(add_pixels4 ## suffix)(uint8_t *av_restrict _pixels,  \

  245.                                          DCTELEM *_block,               \

  246.                                          int line_size)                 \

  247. {                                                                       \

  248.     int i;                                                              \

  249.     pixel *av_restrict pixels = (pixel *av_restrict)_pixels;            \

  250.     dctcoef *block = (dctcoef*)_block;                                  \

  251.     line_size /= sizeof(pixel);                                         \

  252.                                                                         \

  253.     for(i=0;i<4;i++) {                                                  \

  254.         pixels[0] += block[0];                                          \

  255.         pixels[1] += block[1];                                          \

  256.         pixels[2] += block[2];                                          \

  257.         pixels[3] += block[3];                                          \

  258.         pixels += line_size;                                            \

  259.         block += 4;                                                     \

  260.     }                                                                   \

  261. }                                                                       \

  262.                                                                         \

  263. static void FUNCC(clear_block ## suffix)(DCTELEM *block)                \

  264. {                                                                       \

  265.     memset(block, 0, sizeof(dctcoef)*64);                               \

  266. }                                                                       \

  267.                                                                         \

  268. /**                                                                     \

  269.  * memset(blocks, 0, sizeof(DCTELEM)*6*64)                              \

  270.  */                                                                     \

  271. static void FUNCC(clear_blocks ## suffix)(DCTELEM *blocks)              \

  272. {                                                                       \

  273.     memset(blocks, 0, sizeof(dctcoef)*6*64);                            \

  274. }

  275. 

  276. DCTELEM_FUNCS(DCTELEM, _16)

  277. #if BIT_DEPTH > 8

  278. DCTELEM_FUNCS(dctcoef, _32)

  279. #endif

  280. 

  281. #define PIXOP2(OPNAME, OP) \

  282. static void FUNCC(OPNAME ## _pixels2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  283.     int i;\

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

  285.         OP(*((pixel2*)(block  )), AV_RN2P(pixels  ));\

  286.         pixels+=line_size;\

  287.         block +=line_size;\

  288.     }\

  289. }\

  290. static void FUNCC(OPNAME ## _pixels4)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  291.     int i;\

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

  293.         OP(*((pixel4*)(block  )), AV_RN4P(pixels  ));\

  294.         pixels+=line_size;\

  295.         block +=line_size;\

  296.     }\

  297. }\

  298. static void FUNCC(OPNAME ## _pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  299.     int i;\

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

  301.         OP(*((pixel4*)(block                )), AV_RN4P(pixels                ));\

  302.         OP(*((pixel4*)(block+4*sizeof(pixel))), AV_RN4P(pixels+4*sizeof(pixel)));\

  303.         pixels+=line_size;\

  304.         block +=line_size;\

  305.     }\

  306. }\

  307. static inline void FUNCC(OPNAME ## _no_rnd_pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  308.     FUNCC(OPNAME ## _pixels8)(block, pixels, line_size, h);\

  309. }\

  310. \

  311. static inline void FUNC(OPNAME ## _no_rnd_pixels8_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \

  312.                                                 int src_stride1, int src_stride2, int h){\

  313.     int i;\

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

  315.         pixel4 a,b;\

  316.         a= AV_RN4P(&src1[i*src_stride1  ]);\

  317.         b= AV_RN4P(&src2[i*src_stride2  ]);\

  318.         OP(*((pixel4*)&dst[i*dst_stride  ]), no_rnd_avg_pixel4(a, b));\

  319.         a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\

  320.         b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\

  321.         OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), no_rnd_avg_pixel4(a, b));\

  322.     }\

  323. }\

  324. \

  325. static inline void FUNC(OPNAME ## _pixels8_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \

  326.                                                 int src_stride1, int src_stride2, int h){\

  327.     int i;\

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

  329.         pixel4 a,b;\

  330.         a= AV_RN4P(&src1[i*src_stride1  ]);\

  331.         b= AV_RN4P(&src2[i*src_stride2  ]);\

  332.         OP(*((pixel4*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\

  333.         a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\

  334.         b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\

  335.         OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), rnd_avg_pixel4(a, b));\

  336.     }\

  337. }\

  338. \

  339. static inline void FUNC(OPNAME ## _pixels4_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \

  340.                                                 int src_stride1, int src_stride2, int h){\

  341.     int i;\

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

  343.         pixel4 a,b;\

  344.         a= AV_RN4P(&src1[i*src_stride1  ]);\

  345.         b= AV_RN4P(&src2[i*src_stride2  ]);\

  346.         OP(*((pixel4*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\

  347.     }\

  348. }\

  349. \

  350. static inline void FUNC(OPNAME ## _pixels2_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \

  351.                                                 int src_stride1, int src_stride2, int h){\

  352.     int i;\

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

  354.         pixel4 a,b;\

  355.         a= AV_RN2P(&src1[i*src_stride1  ]);\

  356.         b= AV_RN2P(&src2[i*src_stride2  ]);\

  357.         OP(*((pixel2*)&dst[i*dst_stride  ]), rnd_avg_pixel4(a, b));\

  358.     }\

  359. }\

  360. \

  361. static inline void FUNC(OPNAME ## _pixels16_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \

  362.                                                 int src_stride1, int src_stride2, int h){\

  363.     FUNC(OPNAME ## _pixels8_l2)(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\

  364.     FUNC(OPNAME ## _pixels8_l2)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, h);\

  365. }\

  366. \

  367. static inline void FUNC(OPNAME ## _no_rnd_pixels16_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \

  368.                                                 int src_stride1, int src_stride2, int h){\

  369.     FUNC(OPNAME ## _no_rnd_pixels8_l2)(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\

  370.     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);\

  371. }\

  372. \

  373. static inline void FUNCC(OPNAME ## _no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  374.     FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\

  375. }\

  376. \

  377. static inline void FUNCC(OPNAME ## _pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  378.     FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\

  379. }\

  380. \

  381. static inline void FUNCC(OPNAME ## _no_rnd_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  382.     FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\

  383. }\

  384. \

  385. static inline void FUNCC(OPNAME ## _pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  386.     FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\

  387. }\

  388. \

  389. 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,\

  390.                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\

  391.     /* FIXME HIGH BIT DEPTH */\

  392.     int i;\

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

  394.         uint32_t a, b, c, d, l0, l1, h0, h1;\

  395.         a= AV_RN32(&src1[i*src_stride1]);\

  396.         b= AV_RN32(&src2[i*src_stride2]);\

  397.         c= AV_RN32(&src3[i*src_stride3]);\

  398.         d= AV_RN32(&src4[i*src_stride4]);\

  399.         l0=  (a&0x03030303UL)\

  400.            + (b&0x03030303UL)\

  401.            + 0x02020202UL;\

  402.         h0= ((a&0xFCFCFCFCUL)>>2)\

  403.           + ((b&0xFCFCFCFCUL)>>2);\

  404.         l1=  (c&0x03030303UL)\

  405.            + (d&0x03030303UL);\

  406.         h1= ((c&0xFCFCFCFCUL)>>2)\

  407.           + ((d&0xFCFCFCFCUL)>>2);\

  408.         OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  409.         a= AV_RN32(&src1[i*src_stride1+4]);\

  410.         b= AV_RN32(&src2[i*src_stride2+4]);\

  411.         c= AV_RN32(&src3[i*src_stride3+4]);\

  412.         d= AV_RN32(&src4[i*src_stride4+4]);\

  413.         l0=  (a&0x03030303UL)\

  414.            + (b&0x03030303UL)\

  415.            + 0x02020202UL;\

  416.         h0= ((a&0xFCFCFCFCUL)>>2)\

  417.           + ((b&0xFCFCFCFCUL)>>2);\

  418.         l1=  (c&0x03030303UL)\

  419.            + (d&0x03030303UL);\

  420.         h1= ((c&0xFCFCFCFCUL)>>2)\

  421.           + ((d&0xFCFCFCFCUL)>>2);\

  422.         OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  423.     }\

  424. }\

  425. \

  426. static inline void FUNCC(OPNAME ## _pixels4_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  427.     FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\

  428. }\

  429. \

  430. static inline void FUNCC(OPNAME ## _pixels4_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  431.     FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\

  432. }\

  433. \

  434. static inline void FUNCC(OPNAME ## _pixels2_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  435.     FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\

  436. }\

  437. \

  438. static inline void FUNCC(OPNAME ## _pixels2_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\

  439.     FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\

  440. }\

  441. \

  442. 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,\

  443.                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\

  444.     /* FIXME HIGH BIT DEPTH*/\

  445.     int i;\

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

  447.         uint32_t a, b, c, d, l0, l1, h0, h1;\

  448.         a= AV_RN32(&src1[i*src_stride1]);\

  449.         b= AV_RN32(&src2[i*src_stride2]);\

  450.         c= AV_RN32(&src3[i*src_stride3]);\

  451.         d= AV_RN32(&src4[i*src_stride4]);\

  452.         l0=  (a&0x03030303UL)\

  453.            + (b&0x03030303UL)\

  454.            + 0x01010101UL;\

  455.         h0= ((a&0xFCFCFCFCUL)>>2)\

  456.           + ((b&0xFCFCFCFCUL)>>2);\

  457.         l1=  (c&0x03030303UL)\

  458.            + (d&0x03030303UL);\

  459.         h1= ((c&0xFCFCFCFCUL)>>2)\

  460.           + ((d&0xFCFCFCFCUL)>>2);\

  461.         OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  462.         a= AV_RN32(&src1[i*src_stride1+4]);\

  463.         b= AV_RN32(&src2[i*src_stride2+4]);\

  464.         c= AV_RN32(&src3[i*src_stride3+4]);\

  465.         d= AV_RN32(&src4[i*src_stride4+4]);\

  466.         l0=  (a&0x03030303UL)\

  467.            + (b&0x03030303UL)\

  468.            + 0x01010101UL;\

  469.         h0= ((a&0xFCFCFCFCUL)>>2)\

  470.           + ((b&0xFCFCFCFCUL)>>2);\

  471.         l1=  (c&0x03030303UL)\

  472.            + (d&0x03030303UL);\

  473.         h1= ((c&0xFCFCFCFCUL)>>2)\

  474.           + ((d&0xFCFCFCFCUL)>>2);\

  475.         OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  476.     }\

  477. }\

  478. 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,\

  479.                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\

  480.     FUNC(OPNAME ## _pixels8_l4)(dst  , src1  , src2  , src3  , src4  , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\

  481.     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);\

  482. }\

  483. 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,\

  484.                  int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\

  485.     FUNC(OPNAME ## _no_rnd_pixels8_l4)(dst  , src1  , src2  , src3  , src4  , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\

  486.     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);\

  487. }\

  488. \

  489. static inline void FUNCC(OPNAME ## _pixels2_xy2)(uint8_t *p_block, const uint8_t *p_pixels, int line_size, int h)\

  490. {\

  491.         int i, a0, b0, a1, b1;\

  492.         pixel *block = (pixel*)p_block;\

  493.         const pixel *pixels = (const pixel*)p_pixels;\

  494.         line_size >>= sizeof(pixel)-1;\

  495.         a0= pixels[0];\

  496.         b0= pixels[1] + 2;\

  497.         a0 += b0;\

  498.         b0 += pixels[2];\

  499. \

  500.         pixels+=line_size;\

  501.         for(i=0; i<h; i+=2){\

  502.             a1= pixels[0];\

  503.             b1= pixels[1];\

  504.             a1 += b1;\

  505.             b1 += pixels[2];\

  506. \

  507.             block[0]= (a1+a0)>>2; /* FIXME non put */\

  508.             block[1]= (b1+b0)>>2;\

  509. \

  510.             pixels+=line_size;\

  511.             block +=line_size;\

  512. \

  513.             a0= pixels[0];\

  514.             b0= pixels[1] + 2;\

  515.             a0 += b0;\

  516.             b0 += pixels[2];\

  517. \

  518.             block[0]= (a1+a0)>>2;\

  519.             block[1]= (b1+b0)>>2;\

  520.             pixels+=line_size;\

  521.             block +=line_size;\

  522.         }\

  523. }\

  524. \

  525. static inline void FUNCC(OPNAME ## _pixels4_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\

  526. {\

  527.         /* FIXME HIGH BIT DEPTH */\

  528.         int i;\

  529.         const uint32_t a= AV_RN32(pixels  );\

  530.         const uint32_t b= AV_RN32(pixels+1);\

  531.         uint32_t l0=  (a&0x03030303UL)\

  532.                     + (b&0x03030303UL)\

  533.                     + 0x02020202UL;\

  534.         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\

  535.                    + ((b&0xFCFCFCFCUL)>>2);\

  536.         uint32_t l1,h1;\

  537. \

  538.         pixels+=line_size;\

  539.         for(i=0; i<h; i+=2){\

  540.             uint32_t a= AV_RN32(pixels  );\

  541.             uint32_t b= AV_RN32(pixels+1);\

  542.             l1=  (a&0x03030303UL)\

  543.                + (b&0x03030303UL);\

  544.             h1= ((a&0xFCFCFCFCUL)>>2)\

  545.               + ((b&0xFCFCFCFCUL)>>2);\

  546.             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  547.             pixels+=line_size;\

  548.             block +=line_size;\

  549.             a= AV_RN32(pixels  );\

  550.             b= AV_RN32(pixels+1);\

  551.             l0=  (a&0x03030303UL)\

  552.                + (b&0x03030303UL)\

  553.                + 0x02020202UL;\

  554.             h0= ((a&0xFCFCFCFCUL)>>2)\

  555.               + ((b&0xFCFCFCFCUL)>>2);\

  556.             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  557.             pixels+=line_size;\

  558.             block +=line_size;\

  559.         }\

  560. }\

  561. \

  562. static inline void FUNCC(OPNAME ## _pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\

  563. {\

  564.     /* FIXME HIGH BIT DEPTH */\

  565.     int j;\

  566.     for(j=0; j<2; j++){\

  567.         int i;\

  568.         const uint32_t a= AV_RN32(pixels  );\

  569.         const uint32_t b= AV_RN32(pixels+1);\

  570.         uint32_t l0=  (a&0x03030303UL)\

  571.                     + (b&0x03030303UL)\

  572.                     + 0x02020202UL;\

  573.         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\

  574.                    + ((b&0xFCFCFCFCUL)>>2);\

  575.         uint32_t l1,h1;\

  576. \

  577.         pixels+=line_size;\

  578.         for(i=0; i<h; i+=2){\

  579.             uint32_t a= AV_RN32(pixels  );\

  580.             uint32_t b= AV_RN32(pixels+1);\

  581.             l1=  (a&0x03030303UL)\

  582.                + (b&0x03030303UL);\

  583.             h1= ((a&0xFCFCFCFCUL)>>2)\

  584.               + ((b&0xFCFCFCFCUL)>>2);\

  585.             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  586.             pixels+=line_size;\

  587.             block +=line_size;\

  588.             a= AV_RN32(pixels  );\

  589.             b= AV_RN32(pixels+1);\

  590.             l0=  (a&0x03030303UL)\

  591.                + (b&0x03030303UL)\

  592.                + 0x02020202UL;\

  593.             h0= ((a&0xFCFCFCFCUL)>>2)\

  594.               + ((b&0xFCFCFCFCUL)>>2);\

  595.             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  596.             pixels+=line_size;\

  597.             block +=line_size;\

  598.         }\

  599.         pixels+=4-line_size*(h+1);\

  600.         block +=4-line_size*h;\

  601.     }\

  602. }\

  603. \

  604. static inline void FUNCC(OPNAME ## _no_rnd_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)\

  605. {\

  606.     /* FIXME HIGH BIT DEPTH */\

  607.     int j;\

  608.     for(j=0; j<2; j++){\

  609.         int i;\

  610.         const uint32_t a= AV_RN32(pixels  );\

  611.         const uint32_t b= AV_RN32(pixels+1);\

  612.         uint32_t l0=  (a&0x03030303UL)\

  613.                     + (b&0x03030303UL)\

  614.                     + 0x01010101UL;\

  615.         uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\

  616.                    + ((b&0xFCFCFCFCUL)>>2);\

  617.         uint32_t l1,h1;\

  618. \

  619.         pixels+=line_size;\

  620.         for(i=0; i<h; i+=2){\

  621.             uint32_t a= AV_RN32(pixels  );\

  622.             uint32_t b= AV_RN32(pixels+1);\

  623.             l1=  (a&0x03030303UL)\

  624.                + (b&0x03030303UL);\

  625.             h1= ((a&0xFCFCFCFCUL)>>2)\

  626.               + ((b&0xFCFCFCFCUL)>>2);\

  627.             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  628.             pixels+=line_size;\

  629.             block +=line_size;\

  630.             a= AV_RN32(pixels  );\

  631.             b= AV_RN32(pixels+1);\

  632.             l0=  (a&0x03030303UL)\

  633.                + (b&0x03030303UL)\

  634.                + 0x01010101UL;\

  635.             h0= ((a&0xFCFCFCFCUL)>>2)\

  636.               + ((b&0xFCFCFCFCUL)>>2);\

  637.             OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\

  638.             pixels+=line_size;\

  639.             block +=line_size;\

  640.         }\

  641.         pixels+=4-line_size*(h+1);\

  642.         block +=4-line_size*h;\

  643.     }\

  644. }\

  645. \

  646. CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16)    , FUNCC(OPNAME ## _pixels8)    , 8*sizeof(pixel))\

  647. CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_x2) , FUNCC(OPNAME ## _pixels8_x2) , 8*sizeof(pixel))\

  648. CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_y2) , FUNCC(OPNAME ## _pixels8_y2) , 8*sizeof(pixel))\

  649. CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_xy2), FUNCC(OPNAME ## _pixels8_xy2), 8*sizeof(pixel))\

  650. av_unused CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16)    , FUNCC(OPNAME ## _pixels8) , 8*sizeof(pixel))\

  651. CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_x2) , FUNCC(OPNAME ## _no_rnd_pixels8_x2) , 8*sizeof(pixel))\

  652. CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_y2) , FUNCC(OPNAME ## _no_rnd_pixels8_y2) , 8*sizeof(pixel))\

  653. CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_xy2), FUNCC(OPNAME ## _no_rnd_pixels8_xy2), 8*sizeof(pixel))\

  654. 

  655. #define op_avg(a, b) a = rnd_avg_pixel4(a, b)

  656. #define op_put(a, b) a = b

  657. 

  658. PIXOP2(avg, op_avg)

  659. PIXOP2(put, op_put)

  660. #undef op_avg

  661. #undef op_put

  662. 

  663. #define put_no_rnd_pixels8_c  put_pixels8_c

  664. #define put_no_rnd_pixels16_c put_pixels16_c

  665. 

  666. static void FUNCC(put_no_rnd_pixels16_l2)(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){

  667.     FUNC(put_no_rnd_pixels16_l2)(dst, a, b, stride, stride, stride, h);

  668. }

  669. 

  670. static void FUNCC(put_no_rnd_pixels8_l2)(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){

  671.     FUNC(put_no_rnd_pixels8_l2)(dst, a, b, stride, stride, stride, h);

  672. }

  673. 

  674. #define H264_CHROMA_MC(OPNAME, OP)\

  675. 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){\

  676.     pixel *dst = (pixel*)p_dst;\

  677.     pixel *src = (pixel*)p_src;\

  678.     const int A=(8-x)*(8-y);\

  679.     const int B=(  x)*(8-y);\

  680.     const int C=(8-x)*(  y);\

  681.     const int D=(  x)*(  y);\

  682.     int i;\

  683.     stride >>= sizeof(pixel)-1;\

  684.     \

  685.     av_assert2(x<8 && y<8 && x>=0 && y>=0);\

  686. \

  687.     if(D){\

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

  689.             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\

  690.             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\

  691.             dst+= stride;\

  692.             src+= stride;\

  693.         }\

  694.     }else{\

  695.         const int E= B+C;\

  696.         const int step= C ? stride : 1;\

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

  698.             OP(dst[0], (A*src[0] + E*src[step+0]));\

  699.             OP(dst[1], (A*src[1] + E*src[step+1]));\

  700.             dst+= stride;\

  701.             src+= stride;\

  702.         }\

  703.     }\

  704. }\

  705. \

  706. 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){\

  707.     pixel *dst = (pixel*)p_dst;\

  708.     pixel *src = (pixel*)p_src;\

  709.     const int A=(8-x)*(8-y);\

  710.     const int B=(  x)*(8-y);\

  711.     const int C=(8-x)*(  y);\

  712.     const int D=(  x)*(  y);\

  713.     int i;\

  714.     stride >>= sizeof(pixel)-1;\

  715.     \

  716.     av_assert2(x<8 && y<8 && x>=0 && y>=0);\

  717. \

  718.     if(D){\

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

  720.             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\

  721.             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\

  722.             OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\

  723.             OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\

  724.             dst+= stride;\

  725.             src+= stride;\

  726.         }\

  727.     }else{\

  728.         const int E= B+C;\

  729.         const int step= C ? stride : 1;\

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

  731.             OP(dst[0], (A*src[0] + E*src[step+0]));\

  732.             OP(dst[1], (A*src[1] + E*src[step+1]));\

  733.             OP(dst[2], (A*src[2] + E*src[step+2]));\

  734.             OP(dst[3], (A*src[3] + E*src[step+3]));\

  735.             dst+= stride;\

  736.             src+= stride;\

  737.         }\

  738.     }\

  739. }\

  740. \

  741. 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){\

  742.     pixel *dst = (pixel*)p_dst;\

  743.     pixel *src = (pixel*)p_src;\

  744.     const int A=(8-x)*(8-y);\

  745.     const int B=(  x)*(8-y);\

  746.     const int C=(8-x)*(  y);\

  747.     const int D=(  x)*(  y);\

  748.     int i;\

  749.     stride >>= sizeof(pixel)-1;\

  750.     \

  751.     av_assert2(x<8 && y<8 && x>=0 && y>=0);\

  752. \

  753.     if(D){\

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

  755.             OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\

  756.             OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\

  757.             OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\

  758.             OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\

  759.             OP(dst[4], (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5]));\

  760.             OP(dst[5], (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6]));\

  761.             OP(dst[6], (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7]));\

  762.             OP(dst[7], (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8]));\

  763.             dst+= stride;\

  764.             src+= stride;\

  765.         }\

  766.     }else{\

  767.         const int E= B+C;\

  768.         const int step= C ? stride : 1;\

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

  770.             OP(dst[0], (A*src[0] + E*src[step+0]));\

  771.             OP(dst[1], (A*src[1] + E*src[step+1]));\

  772.             OP(dst[2], (A*src[2] + E*src[step+2]));\

  773.             OP(dst[3], (A*src[3] + E*src[step+3]));\

  774.             OP(dst[4], (A*src[4] + E*src[step+4]));\

  775.             OP(dst[5], (A*src[5] + E*src[step+5]));\

  776.             OP(dst[6], (A*src[6] + E*src[step+6]));\

  777.             OP(dst[7], (A*src[7] + E*src[step+7]));\

  778.             dst+= stride;\

  779.             src+= stride;\

  780.         }\

  781.     }\

  782. }

  783. 

  784. #define op_avg(a, b) a = (((a)+(((b) + 32)>>6)+1)>>1)

  785. #define op_put(a, b) a = (((b) + 32)>>6)

  786. 

  787. H264_CHROMA_MC(put_       , op_put)

  788. H264_CHROMA_MC(avg_       , op_avg)

  789. #undef op_avg

  790. #undef op_put

  791. 

  792. #define H264_LOWPASS(OPNAME, OP, OP2) \

  793. static av_unused void FUNC(OPNAME ## h264_qpel2_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\

  794.     const int h=2;\

  795.     INIT_CLIP\

  796.     int i;\

  797.     pixel *dst = (pixel*)p_dst;\

  798.     pixel *src = (pixel*)p_src;\

  799.     dstStride >>= sizeof(pixel)-1;\

  800.     srcStride >>= sizeof(pixel)-1;\

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

  802.     {\

  803.         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\

  804.         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\

  805.         dst+=dstStride;\

  806.         src+=srcStride;\

  807.     }\

  808. }\

  809. \

  810. static av_unused void FUNC(OPNAME ## h264_qpel2_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\

  811.     const int w=2;\

  812.     INIT_CLIP\

  813.     int i;\

  814.     pixel *dst = (pixel*)p_dst;\

  815.     pixel *src = (pixel*)p_src;\

  816.     dstStride >>= sizeof(pixel)-1;\

  817.     srcStride >>= sizeof(pixel)-1;\

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

  819.     {\

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

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

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

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

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

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

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

  827.         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\

  828.         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\

  829.         dst++;\

  830.         src++;\

  831.     }\

  832. }\

  833. \

  834. 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){\

  835.     const int h=2;\

  836.     const int w=2;\

  837.     const int pad = (BIT_DEPTH == 10) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\

  838.     INIT_CLIP\

  839.     int i;\

  840.     pixel *dst = (pixel*)p_dst;\

  841.     pixel *src = (pixel*)p_src;\

  842.     dstStride >>= sizeof(pixel)-1;\

  843.     srcStride >>= sizeof(pixel)-1;\

  844.     src -= 2*srcStride;\

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

  846.     {\

  847.         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\

  848.         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\

  849.         tmp+=tmpStride;\

  850.         src+=srcStride;\

  851.     }\

  852.     tmp -= tmpStride*(h+5-2);\

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

  854.     {\

  855.         const int tmpB= tmp[-2*tmpStride] - pad;\

  856.         const int tmpA= tmp[-1*tmpStride] - pad;\

  857.         const int tmp0= tmp[0 *tmpStride] - pad;\

  858.         const int tmp1= tmp[1 *tmpStride] - pad;\

  859.         const int tmp2= tmp[2 *tmpStride] - pad;\

  860.         const int tmp3= tmp[3 *tmpStride] - pad;\

  861.         const int tmp4= tmp[4 *tmpStride] - pad;\

  862.         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\

  863.         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\

  864.         dst++;\

  865.         tmp++;\

  866.     }\

  867. }\

  868. static void FUNC(OPNAME ## h264_qpel4_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\

  869.     const int h=4;\

  870.     INIT_CLIP\

  871.     int i;\

  872.     pixel *dst = (pixel*)p_dst;\

  873.     pixel *src = (pixel*)p_src;\

  874.     dstStride >>= sizeof(pixel)-1;\

  875.     srcStride >>= sizeof(pixel)-1;\

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

  877.     {\

  878.         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\

  879.         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\

  880.         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]));\

  881.         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]));\

  882.         dst+=dstStride;\

  883.         src+=srcStride;\

  884.     }\

  885. }\

  886. \

  887. static void FUNC(OPNAME ## h264_qpel4_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\

  888.     const int w=4;\

  889.     INIT_CLIP\

  890.     int i;\

  891.     pixel *dst = (pixel*)p_dst;\

  892.     pixel *src = (pixel*)p_src;\

  893.     dstStride >>= sizeof(pixel)-1;\

  894.     srcStride >>= sizeof(pixel)-1;\

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

  896.     {\

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

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

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

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

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

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

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

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

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

  906.         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\

  907.         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\

  908.         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\

  909.         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\

  910.         dst++;\

  911.         src++;\

  912.     }\

  913. }\

  914. \

  915. static void FUNC(OPNAME ## h264_qpel4_hv_lowpass)(uint8_t *p_dst, pixeltmp *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\

  916.     const int h=4;\

  917.     const int w=4;\

  918.     const int pad = (BIT_DEPTH == 10) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\

  919.     INIT_CLIP\

  920.     int i;\

  921.     pixel *dst = (pixel*)p_dst;\

  922.     pixel *src = (pixel*)p_src;\

  923.     dstStride >>= sizeof(pixel)-1;\

  924.     srcStride >>= sizeof(pixel)-1;\

  925.     src -= 2*srcStride;\

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

  927.     {\

  928.         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\

  929.         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\

  930.         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]) + pad;\

  931.         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]) + pad;\

  932.         tmp+=tmpStride;\

  933.         src+=srcStride;\

  934.     }\

  935.     tmp -= tmpStride*(h+5-2);\

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

  937.     {\

  938.         const int tmpB= tmp[-2*tmpStride] - pad;\

  939.         const int tmpA= tmp[-1*tmpStride] - pad;\

  940.         const int tmp0= tmp[0 *tmpStride] - pad;\

  941.         const int tmp1= tmp[1 *tmpStride] - pad;\

  942.         const int tmp2= tmp[2 *tmpStride] - pad;\

  943.         const int tmp3= tmp[3 *tmpStride] - pad;\

  944.         const int tmp4= tmp[4 *tmpStride] - pad;\

  945.         const int tmp5= tmp[5 *tmpStride] - pad;\

  946.         const int tmp6= tmp[6 *tmpStride] - pad;\

  947.         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\

  948.         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\

  949.         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\

  950.         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\

  951.         dst++;\

  952.         tmp++;\

  953.     }\

  954. }\

  955. \

  956. static void FUNC(OPNAME ## h264_qpel8_h_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\

  957.     const int h=8;\

  958.     INIT_CLIP\

  959.     int i;\

  960.     pixel *dst = (pixel*)p_dst;\

  961.     pixel *src = (pixel*)p_src;\

  962.     dstStride >>= sizeof(pixel)-1;\

  963.     srcStride >>= sizeof(pixel)-1;\

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

  965.     {\

  966.         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]));\

  967.         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]));\

  968.         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]));\

  969.         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]));\

  970.         OP(dst[4], (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]));\

  971.         OP(dst[5], (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]));\

  972.         OP(dst[6], (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]));\

  973.         OP(dst[7], (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]));\

  974.         dst+=dstStride;\

  975.         src+=srcStride;\

  976.     }\

  977. }\

  978. \

  979. static void FUNC(OPNAME ## h264_qpel8_v_lowpass)(uint8_t *p_dst, uint8_t *p_src, int dstStride, int srcStride){\

  980.     const int w=8;\

  981.     INIT_CLIP\

  982.     int i;\

  983.     pixel *dst = (pixel*)p_dst;\

  984.     pixel *src = (pixel*)p_src;\

  985.     dstStride >>= sizeof(pixel)-1;\

  986.     srcStride >>= sizeof(pixel)-1;\

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

  988.     {\

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

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

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

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

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

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

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

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

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

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

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

  1000.         const int src9= src[9 *srcStride];\

  1001.         const int src10=src[10*srcStride];\

  1002.         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\

  1003.         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\

  1004.         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\

  1005.         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\

  1006.         OP(dst[4*dstStride], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\

  1007.         OP(dst[5*dstStride], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\

  1008.         OP(dst[6*dstStride], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\

  1009.         OP(dst[7*dstStride], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\

  1010.         dst++;\

  1011.         src++;\

  1012.     }\

  1013. }\

  1014. \

  1015. static void FUNC(OPNAME ## h264_qpel8_hv_lowpass)(uint8_t *p_dst, pixeltmp *tmp, uint8_t *p_src, int dstStride, int tmpStride, int srcStride){\

  1016.     const int h=8;\

  1017.     const int w=8;\

  1018.     const int pad = (BIT_DEPTH == 10) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\

  1019.     INIT_CLIP\

  1020.     int i;\

  1021.     pixel *dst = (pixel*)p_dst;\

  1022.     pixel *src = (pixel*)p_src;\

  1023.     dstStride >>= sizeof(pixel)-1;\

  1024.     srcStride >>= sizeof(pixel)-1;\

  1025.     src -= 2*srcStride;\

  1026.     for(i=0; i<h+5; i++)\

  1027.     {\

  1028.         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]) + pad;\

  1029.         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]) + pad;\

  1030.         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]) + pad;\

  1031.         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]) + pad;\

  1032.         tmp[4]= (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]) + pad;\

  1033.         tmp[5]= (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]) + pad;\

  1034.         tmp[6]= (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]) + pad;\

  1035.         tmp[7]= (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]) + pad;\

  1036.         tmp+=tmpStride;\

  1037.         src+=srcStride;\

  1038.     }\

  1039.     tmp -= tmpStride*(h+5-2);\

  1040.     for(i=0; i<w; i++)\

  1041.     {\

  1042.         const int tmpB= tmp[-2*tmpStride] - pad;\

  1043.         const int tmpA= tmp[-1*tmpStride] - pad;\

  1044.         const int tmp0= tmp[0 *tmpStride] - pad;\

  1045.         const int tmp1= tmp[1 *tmpStride] - pad;\

  1046.         const int tmp2= tmp[2 *tmpStride] - pad;\

  1047.         const int tmp3= tmp[3 *tmpStride] - pad;\

  1048.         const int tmp4= tmp[4 *tmpStride] - pad;\

  1049.         const int tmp5= tmp[5 *tmpStride] - pad;\

  1050.         const int tmp6= tmp[6 *tmpStride] - pad;\

  1051.         const int tmp7= tmp[7 *tmpStride] - pad;\

  1052.         const int tmp8= tmp[8 *tmpStride] - pad;\

  1053.         const int tmp9= tmp[9 *tmpStride] - pad;\

  1054.         const int tmp10=tmp[10*tmpStride] - pad;\

  1055.         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\

  1056.         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\

  1057.         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\

  1058.         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\

  1059.         OP2(dst[4*dstStride], (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));\

  1060.         OP2(dst[5*dstStride], (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));\

  1061.         OP2(dst[6*dstStride], (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));\

  1062.         OP2(dst[7*dstStride], (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));\

  1063.         dst++;\

  1064.         tmp++;\

  1065.     }\

  1066. }\

  1067. \

  1068. static void FUNC(OPNAME ## h264_qpel16_v_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  1069.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\

  1070.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1071.     src += 8*srcStride;\

  1072.     dst += 8*dstStride;\

  1073.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\

  1074.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1075. }\

  1076. \

  1077. static void FUNC(OPNAME ## h264_qpel16_h_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  1078.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\

  1079.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1080.     src += 8*srcStride;\

  1081.     dst += 8*dstStride;\

  1082.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\

  1083.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1084. }\

  1085. \

  1086. static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, pixeltmp *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  1087.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\

  1088.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\

  1089.     src += 8*srcStride;\

  1090.     dst += 8*dstStride;\

  1091.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\

  1092.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\

  1093. }\

  1094. 

  1095. #define H264_MC(OPNAME, SIZE) \

  1096. static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\

  1097.     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\

  1098. }\

  1099. \

  1100. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\

  1101.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1102.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1103.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1104. }\

  1105. \

  1106. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\

  1107.     FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\

  1108. }\

  1109. \

  1110. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\

  1111.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1112.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1113.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1114. }\

  1115. \

  1116. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\

  1117.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1118.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1119.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1120.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1121.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1122.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1123. }\

  1124. \

  1125. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\

  1126.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1127.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1128.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1129.     FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\

  1130. }\

  1131. \

  1132. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\

  1133.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1134.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1135.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1136.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1137.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1138.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1139. }\

  1140. \

  1141. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(uint8_t *dst, uint8_t *src, int stride){\

  1142.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1143.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1144.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1145.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1146.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1147.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1148.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1149.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1150. }\

  1151. \

  1152. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\

  1153.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1154.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1155.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1156.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1157.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1158.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1159.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1160.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1161. }\

  1162. \

  1163. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\

  1164.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1165.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1166.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1167.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1168.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1169.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1170.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1171.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1172. }\

  1173. \

  1174. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\

  1175.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1176.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1177.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1178.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1179.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1180.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1181.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1182.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1183. }\

  1184. \

  1185. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\

  1186.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1187.     FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\

  1188. }\

  1189. \

  1190. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\

  1191.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1192.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1193.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1194.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1195.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1196.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1197. }\

  1198. \

  1199. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\

  1200.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1201.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1202.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1203.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1204.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1205.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1206. }\

  1207. \

  1208. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\

  1209.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1210.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1211.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1212.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1213.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1214.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1215.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1216.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1217.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1218. }\

  1219. \

  1220. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\

  1221.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1222.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1223.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1224.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1225.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1226.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1227.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1228.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1229.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1230. }\

  1231. 

  1232. #define op_avg(a, b)  a = (((a)+CLIP(((b) + 16)>>5)+1)>>1)

  1233. //#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)

  1234. #define op_put(a, b)  a = CLIP(((b) + 16)>>5)

  1235. #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)

  1236. #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)

  1237. 

  1238. H264_LOWPASS(put_       , op_put, op2_put)

  1239. H264_LOWPASS(avg_       , op_avg, op2_avg)

  1240. H264_MC(put_, 2)

  1241. H264_MC(put_, 4)

  1242. H264_MC(put_, 8)

  1243. H264_MC(put_, 16)

  1244. H264_MC(avg_, 4)

  1245. H264_MC(avg_, 8)

  1246. H264_MC(avg_, 16)

  1247. 

  1248. #undef op_avg

  1249. #undef op_put

  1250. #undef op2_avg

  1251. #undef op2_put

  1252. 

  1253. #if BIT_DEPTH == 8

  1254. #   define put_h264_qpel8_mc00_8_c  ff_put_pixels8x8_8_c

  1255. #   define avg_h264_qpel8_mc00_8_c  ff_avg_pixels8x8_8_c

  1256. #   define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c

  1257. #   define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c

  1258. #elif BIT_DEPTH == 9

  1259. #   define put_h264_qpel8_mc00_9_c  ff_put_pixels8x8_9_c

  1260. #   define avg_h264_qpel8_mc00_9_c  ff_avg_pixels8x8_9_c

  1261. #   define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c

  1262. #   define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c

  1263. #elif BIT_DEPTH == 10

  1264. #   define put_h264_qpel8_mc00_10_c  ff_put_pixels8x8_10_c

  1265. #   define avg_h264_qpel8_mc00_10_c  ff_avg_pixels8x8_10_c

  1266. #   define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c

  1267. #   define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c

  1268. #elif BIT_DEPTH == 12

  1269. #   define put_h264_qpel8_mc00_12_c  ff_put_pixels8x8_12_c

  1270. #   define avg_h264_qpel8_mc00_12_c  ff_avg_pixels8x8_12_c

  1271. #   define put_h264_qpel16_mc00_12_c ff_put_pixels16x16_12_c

  1272. #   define avg_h264_qpel16_mc00_12_c ff_avg_pixels16x16_12_c

  1273. #elif BIT_DEPTH == 14

  1274. #   define put_h264_qpel8_mc00_14_c  ff_put_pixels8x8_14_c

  1275. #   define avg_h264_qpel8_mc00_14_c  ff_avg_pixels8x8_14_c

  1276. #   define put_h264_qpel16_mc00_14_c ff_put_pixels16x16_14_c

  1277. #   define avg_h264_qpel16_mc00_14_c ff_avg_pixels16x16_14_c

  1278. #endif

  1279. 

  1280. void FUNCC(ff_put_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1281.     FUNCC(put_pixels8)(dst, src, stride, 8);

  1282. }

  1283. void FUNCC(ff_avg_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1284.     FUNCC(avg_pixels8)(dst, src, stride, 8);

  1285. }

  1286. void FUNCC(ff_put_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1287.     FUNCC(put_pixels16)(dst, src, stride, 16);

  1288. }

  1289. void FUNCC(ff_avg_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1290.     FUNCC(avg_pixels16)(dst, src, stride, 16);

  1291. }