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.
49232 Commits
32 Branches
1006MB
Tree: ca1bc188f4
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'ca1bc188f4'
${ noResults }
FFmpeg/libavcodec/dsputil_template.c

1212 lines
53KB
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. #define DCTELEM_FUNCS(dctcoef, suffix)                                  \

  117. static void FUNCC(get_pixels ## suffix)(int16_t *av_restrict _block,    \

  118.                                         const uint8_t *_pixels,         \

  119.                                         int line_size)                  \

  120. {                                                                       \

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

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

  123.     int i;                                                              \

  124.                                                                         \

  125.     /* read the pixels */                                               \

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

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

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

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

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

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

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

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

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

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

  136.         block += 8;                                                     \

  137.     }                                                                   \

  138. }                                                                       \

  139.                                                                         \

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

  141.                                          int16_t *_block,               \

  142.                                          int line_size)                 \

  143. {                                                                       \

  144.     int i;                                                              \

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

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

  147.     line_size /= sizeof(pixel);                                         \

  148.                                                                         \

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

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

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

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

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

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

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

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

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

  158.         pixels += line_size;                                            \

  159.         block += 8;                                                     \

  160.     }                                                                   \

  161. }                                                                       \

  162.                                                                         \

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

  164.                                          int16_t *_block,               \

  165.                                          int line_size)                 \

  166. {                                                                       \

  167.     int i;                                                              \

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

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

  170.     line_size /= sizeof(pixel);                                         \

  171.                                                                         \

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

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

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

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

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

  177.         pixels += line_size;                                            \

  178.         block += 4;                                                     \

  179.     }                                                                   \

  180. }                                                                       \

  181.                                                                         \

  182. static void FUNCC(clear_block ## suffix)(int16_t *block)                \

  183. {                                                                       \

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

  185. }                                                                       \

  186.                                                                         \

  187. /**                                                                     \

  188.  * memset(blocks, 0, sizeof(int16_t)*6*64)                              \

  189.  */                                                                     \

  190. static void FUNCC(clear_blocks ## suffix)(int16_t *blocks)              \

  191. {                                                                       \

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

  193. }

  194. 

  195. DCTELEM_FUNCS(int16_t, _16)

  196. #if BIT_DEPTH > 8

  197. DCTELEM_FUNCS(dctcoef, _32)

  198. #endif

  199. 

  200. #define PIXOP3(OPNAME, OP) \

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

  202.     int i;\

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

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

  205.         pixels+=line_size;\

  206.         block +=line_size;\

  207.     }\

  208. }\

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

  210.     int i;\

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

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

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

  214.         pixels+=line_size;\

  215.         block +=line_size;\

  216.     }\

  217. }\

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

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

  220. }\

  221. \

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

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

  224.     int i;\

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

  226.         pixel4 a,b;\

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

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

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

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

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

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

  233.     }\

  234. }\

  235. \

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

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

  238.     int i;\

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

  240.         pixel4 a,b;\

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

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

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

  244.     }\

  245. }\

  246. \

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

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

  249.     int i;\

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

  251.         pixel4 a,b;\

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

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

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

  255.     }\

  256. }\

  257. \

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

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

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

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

  262. }\

  263. \

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

  265. 

  266. #define PIXOP4(OPNAME, OP) \

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

  268.     int i;\

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

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

  271.         pixels+=line_size;\

  272.         block +=line_size;\

  273.     }\

  274. }\

  275. PIXOP3(OPNAME, OP)

  276. 

  277. #define PIXOP2(OPNAME, OP) \

  278. PIXOP4(OPNAME, OP)\

  279. \

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

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

  282.     int i;\

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

  284.         pixel4 a,b;\

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

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

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

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

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

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

  291.     }\

  292. }\

  293. \

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

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

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

  297.     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);\

  298. }\

  299. \

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

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

  302. }\

  303. \

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

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

  306. }\

  307. \

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

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

  310. }\

  311. \

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

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

  314. }\

  315. \

  316. 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,\

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

  318.     /* FIXME HIGH BIT DEPTH */\

  319.     int i;\

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

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

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

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

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

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

  326.         l0=  (a&0x03030303UL)\

  327.            + (b&0x03030303UL)\

  328.            + 0x02020202UL;\

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

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

  331.         l1=  (c&0x03030303UL)\

  332.            + (d&0x03030303UL);\

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

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

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

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

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

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

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

  340.         l0=  (a&0x03030303UL)\

  341.            + (b&0x03030303UL)\

  342.            + 0x02020202UL;\

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

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

  345.         l1=  (c&0x03030303UL)\

  346.            + (d&0x03030303UL);\

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

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

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

  350.     }\

  351. }\

  352. \

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

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

  355. }\

  356. \

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

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

  359. }\

  360. \

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

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

  363. }\

  364. \

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

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

  367. }\

  368. \

  369. 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,\

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

  371.     /* FIXME HIGH BIT DEPTH*/\

  372.     int i;\

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

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

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

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

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

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

  379.         l0=  (a&0x03030303UL)\

  380.            + (b&0x03030303UL)\

  381.            + 0x01010101UL;\

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

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

  384.         l1=  (c&0x03030303UL)\

  385.            + (d&0x03030303UL);\

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

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

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

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

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

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

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

  393.         l0=  (a&0x03030303UL)\

  394.            + (b&0x03030303UL)\

  395.            + 0x01010101UL;\

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

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

  398.         l1=  (c&0x03030303UL)\

  399.            + (d&0x03030303UL);\

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

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

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

  403.     }\

  404. }\

  405. 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,\

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

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

  408.     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);\

  409. }\

  410. 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,\

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

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

  413.     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);\

  414. }\

  415. \

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

  417. {\

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

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

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

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

  422.         a0= pixels[0];\

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

  424.         a0 += b0;\

  425.         b0 += pixels[2];\

  426. \

  427.         pixels+=line_size;\

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

  429.             a1= pixels[0];\

  430.             b1= pixels[1];\

  431.             a1 += b1;\

  432.             b1 += pixels[2];\

  433. \

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

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

  436. \

  437.             pixels+=line_size;\

  438.             block +=line_size;\

  439. \

  440.             a0= pixels[0];\

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

  442.             a0 += b0;\

  443.             b0 += pixels[2];\

  444. \

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

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

  447.             pixels+=line_size;\

  448.             block +=line_size;\

  449.         }\

  450. }\

  451. \

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

  453. {\

  454.         /* FIXME HIGH BIT DEPTH */\

  455.         int i;\

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

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

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

  459.                     + (b&0x03030303UL)\

  460.                     + 0x02020202UL;\

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

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

  463.         uint32_t l1,h1;\

  464. \

  465.         pixels+=line_size;\

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

  467.             uint32_t a= AV_RN32(pixels  );\

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

  469.             l1=  (a&0x03030303UL)\

  470.                + (b&0x03030303UL);\

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

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

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

  474.             pixels+=line_size;\

  475.             block +=line_size;\

  476.             a= AV_RN32(pixels  );\

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

  478.             l0=  (a&0x03030303UL)\

  479.                + (b&0x03030303UL)\

  480.                + 0x02020202UL;\

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

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

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

  484.             pixels+=line_size;\

  485.             block +=line_size;\

  486.         }\

  487. }\

  488. \

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

  490. {\

  491.     /* FIXME HIGH BIT DEPTH */\

  492.     int j;\

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

  494.         int i;\

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

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

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

  498.                     + (b&0x03030303UL)\

  499.                     + 0x02020202UL;\

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

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

  502.         uint32_t l1,h1;\

  503. \

  504.         pixels+=line_size;\

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

  506.             uint32_t a= AV_RN32(pixels  );\

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

  508.             l1=  (a&0x03030303UL)\

  509.                + (b&0x03030303UL);\

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

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

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

  513.             pixels+=line_size;\

  514.             block +=line_size;\

  515.             a= AV_RN32(pixels  );\

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

  517.             l0=  (a&0x03030303UL)\

  518.                + (b&0x03030303UL)\

  519.                + 0x02020202UL;\

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

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

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

  523.             pixels+=line_size;\

  524.             block +=line_size;\

  525.         }\

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

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

  528.     }\

  529. }\

  530. \

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

  532. {\

  533.     /* FIXME HIGH BIT DEPTH */\

  534.     int j;\

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

  536.         int i;\

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

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

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

  540.                     + (b&0x03030303UL)\

  541.                     + 0x01010101UL;\

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

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

  544.         uint32_t l1,h1;\

  545. \

  546.         pixels+=line_size;\

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

  548.             uint32_t a= AV_RN32(pixels  );\

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

  550.             l1=  (a&0x03030303UL)\

  551.                + (b&0x03030303UL);\

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

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

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

  555.             pixels+=line_size;\

  556.             block +=line_size;\

  557.             a= AV_RN32(pixels  );\

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

  559.             l0=  (a&0x03030303UL)\

  560.                + (b&0x03030303UL)\

  561.                + 0x01010101UL;\

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

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

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

  565.             pixels+=line_size;\

  566.             block +=line_size;\

  567.         }\

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

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

  570.     }\

  571. }\

  572. \

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

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

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

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

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

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

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

  580. 

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

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

  583. #if BIT_DEPTH == 8

  584. PIXOP2(avg, op_avg)

  585. PIXOP2(put, op_put)

  586. #else

  587. PIXOP3(avg, op_avg)

  588. PIXOP4(put, op_put)

  589. #endif

  590. #undef op_avg

  591. #undef op_put

  592. 

  593. #define H264_CHROMA_MC(OPNAME, OP)\

  594. 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){\

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

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

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

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

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

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

  601.     int i;\

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

  603.     \

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

  605. \

  606.     if(D){\

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

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

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

  610.             dst+= stride;\

  611.             src+= stride;\

  612.         }\

  613.     }else{\

  614.         const int E= B+C;\

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

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

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

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

  619.             dst+= stride;\

  620.             src+= stride;\

  621.         }\

  622.     }\

  623. }\

  624. \

  625. 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){\

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

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

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

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

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

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

  632.     int i;\

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

  634.     \

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

  636. \

  637.     if(D){\

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

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

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

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

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

  643.             dst+= stride;\

  644.             src+= stride;\

  645.         }\

  646.     }else{\

  647.         const int E= B+C;\

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

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

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

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

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

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

  654.             dst+= stride;\

  655.             src+= stride;\

  656.         }\

  657.     }\

  658. }\

  659. \

  660. 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){\

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

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

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

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

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

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

  667.     int i;\

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

  669.     \

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

  671. \

  672.     if(D){\

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

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

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

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

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

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

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

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

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

  682.             dst+= stride;\

  683.             src+= stride;\

  684.         }\

  685.     }else{\

  686.         const int E= B+C;\

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

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

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

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

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

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

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

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

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

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

  697.             dst+= stride;\

  698.             src+= stride;\

  699.         }\

  700.     }\

  701. }

  702. 

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

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

  705. 

  706. H264_CHROMA_MC(put_       , op_put)

  707. H264_CHROMA_MC(avg_       , op_avg)

  708. #undef op_avg

  709. #undef op_put

  710. 

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

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

  713.     const int h=2;\

  714.     INIT_CLIP\

  715.     int i;\

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

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

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

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

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

  721.     {\

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

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

  724.         dst+=dstStride;\

  725.         src+=srcStride;\

  726.     }\

  727. }\

  728. \

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

  730.     const int w=2;\

  731.     INIT_CLIP\

  732.     int i;\

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

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

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

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

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

  738.     {\

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

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

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

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

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

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

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

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

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

  748.         dst++;\

  749.         src++;\

  750.     }\

  751. }\

  752. \

  753. 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){\

  754.     const int h=2;\

  755.     const int w=2;\

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

  757.     INIT_CLIP\

  758.     int i;\

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

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

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

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

  763.     src -= 2*srcStride;\

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

  765.     {\

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

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

  768.         tmp+=tmpStride;\

  769.         src+=srcStride;\

  770.     }\

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

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

  773.     {\

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

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

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

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

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

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

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

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

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

  783.         dst++;\

  784.         tmp++;\

  785.     }\

  786. }\

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

  788.     const int h=4;\

  789.     INIT_CLIP\

  790.     int i;\

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

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

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

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

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

  796.     {\

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

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

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

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

  801.         dst+=dstStride;\

  802.         src+=srcStride;\

  803.     }\

  804. }\

  805. \

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

  807.     const int w=4;\

  808.     INIT_CLIP\

  809.     int i;\

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

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

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

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

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

  815.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

  829.         dst++;\

  830.         src++;\

  831.     }\

  832. }\

  833. \

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

  835.     const int h=4;\

  836.     const int w=4;\

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

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

  851.         tmp+=tmpStride;\

  852.         src+=srcStride;\

  853.     }\

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

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

  856.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

  870.         dst++;\

  871.         tmp++;\

  872.     }\

  873. }\

  874. \

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

  876.     const int h=8;\

  877.     INIT_CLIP\

  878.     int i;\

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

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

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

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

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

  884.     {\

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

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

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

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

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

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

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

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

  893.         dst+=dstStride;\

  894.         src+=srcStride;\

  895.     }\

  896. }\

  897. \

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

  899.     const int w=8;\

  900.     INIT_CLIP\

  901.     int i;\

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

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

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

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

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

  907.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

  929.         dst++;\

  930.         src++;\

  931.     }\

  932. }\

  933. \

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

  935.     const int h=8;\

  936.     const int w=8;\

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

  938.     INIT_CLIP\

  939.     int i;\

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

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

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

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

  944.     src -= 2*srcStride;\

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

  946.     {\

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

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

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

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

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

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

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

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

  955.         tmp+=tmpStride;\

  956.         src+=srcStride;\

  957.     }\

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

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

  960.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

  982.         dst++;\

  983.         tmp++;\

  984.     }\

  985. }\

  986. \

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

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

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

  990.     src += 8*srcStride;\

  991.     dst += 8*dstStride;\

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

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

  994. }\

  995. \

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

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

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

  999.     src += 8*srcStride;\

  1000.     dst += 8*dstStride;\

  1001.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\

  1002.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1003. }\

  1004. \

  1005. static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, pixeltmp *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  1006.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\

  1007.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\

  1008.     src += 8*srcStride;\

  1009.     dst += 8*dstStride;\

  1010.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\

  1011.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\

  1012. }\

  1013. 

  1014. #define H264_MC(OPNAME, SIZE) \

  1015. static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\

  1016.     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\

  1017. }\

  1018. \

  1019. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\

  1020.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1021.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1022.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1023. }\

  1024. \

  1025. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\

  1026.     FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\

  1027. }\

  1028. \

  1029. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\

  1030.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1031.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1032.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1033. }\

  1034. \

  1035. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\

  1036.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1037.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1038.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1039.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1040.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1041.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1042. }\

  1043. \

  1044. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\

  1045.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1046.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1047.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1048.     FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\

  1049. }\

  1050. \

  1051. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\

  1052.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1053.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1054.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1055.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1056.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1057.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1058. }\

  1059. \

  1060. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(uint8_t *dst, uint8_t *src, int stride){\

  1061.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1062.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1063.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1064.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1065.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1066.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1067.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1068.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1069. }\

  1070. \

  1071. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\

  1072.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1073.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1074.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1075.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1076.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1077.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1078.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1079.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1080. }\

  1081. \

  1082. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\

  1083.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1084.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1085.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1086.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1087.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1088.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1089.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1090.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1091. }\

  1092. \

  1093. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\

  1094.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1095.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1096.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1097.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1098.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1099.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1100.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1101.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1102. }\

  1103. \

  1104. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\

  1105.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1106.     FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\

  1107. }\

  1108. \

  1109. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\

  1110.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

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

  1112.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1113.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1114.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1115.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1116. }\

  1117. \

  1118. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\

  1119.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1120.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1121.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1122.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1123.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1124.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1125. }\

  1126. \

  1127. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\

  1128.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1129.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1130.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1131.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1132.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1133.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1134.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1135.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1136.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1137. }\

  1138. \

  1139. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\

  1140.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1141.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1142.     pixeltmp tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1143.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

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

  1145.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1146.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1147.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1148.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1149. }\

  1150. 

  1151. #define op_avg(a, b)  a = (((a)+CLIP(((b) + 16)>>5)+1)>>1)

  1152. //#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)

  1153. #define op_put(a, b)  a = CLIP(((b) + 16)>>5)

  1154. #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)

  1155. #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)

  1156. 

  1157. H264_LOWPASS(put_       , op_put, op2_put)

  1158. H264_LOWPASS(avg_       , op_avg, op2_avg)

  1159. H264_MC(put_, 2)

  1160. H264_MC(put_, 4)

  1161. H264_MC(put_, 8)

  1162. H264_MC(put_, 16)

  1163. H264_MC(avg_, 4)

  1164. H264_MC(avg_, 8)

  1165. H264_MC(avg_, 16)

  1166. 

  1167. #undef op_avg

  1168. #undef op_put

  1169. #undef op2_avg

  1170. #undef op2_put

  1171. 

  1172. #if BIT_DEPTH == 8

  1173. #   define put_h264_qpel8_mc00_8_c  ff_put_pixels8x8_8_c

  1174. #   define avg_h264_qpel8_mc00_8_c  ff_avg_pixels8x8_8_c

  1175. #   define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c

  1176. #   define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c

  1177. #elif BIT_DEPTH == 9

  1178. #   define put_h264_qpel8_mc00_9_c  ff_put_pixels8x8_9_c

  1179. #   define avg_h264_qpel8_mc00_9_c  ff_avg_pixels8x8_9_c

  1180. #   define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c

  1181. #   define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c

  1182. #elif BIT_DEPTH == 10

  1183. #   define put_h264_qpel8_mc00_10_c  ff_put_pixels8x8_10_c

  1184. #   define avg_h264_qpel8_mc00_10_c  ff_avg_pixels8x8_10_c

  1185. #   define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c

  1186. #   define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c

  1187. #elif BIT_DEPTH == 12

  1188. #   define put_h264_qpel8_mc00_12_c  ff_put_pixels8x8_12_c

  1189. #   define avg_h264_qpel8_mc00_12_c  ff_avg_pixels8x8_12_c

  1190. #   define put_h264_qpel16_mc00_12_c ff_put_pixels16x16_12_c

  1191. #   define avg_h264_qpel16_mc00_12_c ff_avg_pixels16x16_12_c

  1192. #elif BIT_DEPTH == 14

  1193. #   define put_h264_qpel8_mc00_14_c  ff_put_pixels8x8_14_c

  1194. #   define avg_h264_qpel8_mc00_14_c  ff_avg_pixels8x8_14_c

  1195. #   define put_h264_qpel16_mc00_14_c ff_put_pixels16x16_14_c

  1196. #   define avg_h264_qpel16_mc00_14_c ff_avg_pixels16x16_14_c

  1197. #endif

  1198. 

  1199. void FUNCC(ff_put_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1200.     FUNCC(put_pixels8)(dst, src, stride, 8);

  1201. }

  1202. void FUNCC(ff_avg_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1203.     FUNCC(avg_pixels8)(dst, src, stride, 8);

  1204. }

  1205. void FUNCC(ff_put_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1206.     FUNCC(put_pixels16)(dst, src, stride, 16);

  1207. }

  1208. void FUNCC(ff_avg_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1209.     FUNCC(avg_pixels16)(dst, src, stride, 16);

  1210. }