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.
36344 Commits
32 Branches
380MB
Tree: 04ea5491a8
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '04ea5491a8'
${ noResults }
FFmpeg/libavcodec/dsputil_template.c

1197 lines
52KB
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 Libav.

  9.  *

  10.  * Libav 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.  * Libav 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 Libav; 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 *_buf, int _wrap, int width, int height, int w, int h, int sides)

  83. {

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

  85.     int wrap = _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)(DCTELEM *restrict _block,       \

  118.                                         const uint8_t *_pixels,         \

  119.                                         int line_size)                  \

  120. {                                                                       \

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

  122.     dctcoef *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 *restrict _pixels,     \

  141.                                          DCTELEM *_block,               \

  142.                                          int line_size)                 \

  143. {                                                                       \

  144.     int i;                                                              \

  145.     pixel *restrict pixels = (pixel *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 *restrict _pixels,     \

  164.                                          DCTELEM *_block,               \

  165.                                          int line_size)                 \

  166. {                                                                       \

  167.     int i;                                                              \

  168.     pixel *restrict pixels = (pixel *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)(DCTELEM *block)                \

  183. {                                                                       \

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

  185. }                                                                       \

  186.                                                                         \

  187. /**                                                                     \

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

  189.  */                                                                     \

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

  191. {                                                                       \

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

  193. }

  194. 

  195. DCTELEM_FUNCS(DCTELEM, _16)

  196. #if BIT_DEPTH > 8

  197. DCTELEM_FUNCS(dctcoef, _32)

  198. #endif

  199. 

  200. #define PIXOP2(OPNAME, OP) \

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

  202.     int i;\

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

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

  205.         pixels+=line_size;\

  206.         block +=line_size;\

  207.     }\

  208. }\

  209. static void FUNCC(OPNAME ## _pixels4)(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.         pixels+=line_size;\

  214.         block +=line_size;\

  215.     }\

  216. }\

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

  218.     int i;\

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

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

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

  222.         pixels+=line_size;\

  223.         block +=line_size;\

  224.     }\

  225. }\

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

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

  228. }\

  229. \

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

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

  232.     int i;\

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

  234.         pixel4 a,b;\

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

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

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

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

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

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

  241.     }\

  242. }\

  243. \

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

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

  246.     int i;\

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

  248.         pixel4 a,b;\

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

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

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

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

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

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

  255.     }\

  256. }\

  257. \

  258. static inline void FUNC(OPNAME ## _pixels4_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.     int i;\

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

  262.         pixel4 a,b;\

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

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

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

  266.     }\

  267. }\

  268. \

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

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

  271.     int i;\

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

  273.         pixel4 a,b;\

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

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

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

  277.     }\

  278. }\

  279. \

  280. static inline void FUNC(OPNAME ## _pixels16_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.     FUNC(OPNAME ## _pixels8_l2)(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\

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

  284. }\

  285. \

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

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

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

  289.     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);\

  290. }\

  291. \

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

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

  294. }\

  295. \

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

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

  298. }\

  299. \

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

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

  302. }\

  303. \

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

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

  306. }\

  307. \

  308. 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,\

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

  310.     /* FIXME HIGH BIT DEPTH */\

  311.     int i;\

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

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

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

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

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

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

  318.         l0=  (a&0x03030303UL)\

  319.            + (b&0x03030303UL)\

  320.            + 0x02020202UL;\

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

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

  323.         l1=  (c&0x03030303UL)\

  324.            + (d&0x03030303UL);\

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

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

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

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

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

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

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

  332.         l0=  (a&0x03030303UL)\

  333.            + (b&0x03030303UL)\

  334.            + 0x02020202UL;\

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

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

  337.         l1=  (c&0x03030303UL)\

  338.            + (d&0x03030303UL);\

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

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

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

  342.     }\

  343. }\

  344. \

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

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

  347. }\

  348. \

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

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

  351. }\

  352. \

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

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

  355. }\

  356. \

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

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

  359. }\

  360. \

  361. 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,\

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

  363.     /* FIXME HIGH BIT DEPTH*/\

  364.     int i;\

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

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

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

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

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

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

  371.         l0=  (a&0x03030303UL)\

  372.            + (b&0x03030303UL)\

  373.            + 0x01010101UL;\

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

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

  376.         l1=  (c&0x03030303UL)\

  377.            + (d&0x03030303UL);\

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

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

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

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

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

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

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

  385.         l0=  (a&0x03030303UL)\

  386.            + (b&0x03030303UL)\

  387.            + 0x01010101UL;\

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

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

  390.         l1=  (c&0x03030303UL)\

  391.            + (d&0x03030303UL);\

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

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

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

  395.     }\

  396. }\

  397. 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,\

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

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

  400.     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);\

  401. }\

  402. 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,\

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

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

  405.     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);\

  406. }\

  407. \

  408. static inline void FUNCC(OPNAME ## _pixels2_xy2)(uint8_t *_block, const uint8_t *_pixels, int line_size, int h)\

  409. {\

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

  411.         pixel *block = (pixel*)_block;\

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

  413.         line_size /= sizeof(pixel);\

  414.         a0= pixels[0];\

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

  416.         a0 += b0;\

  417.         b0 += pixels[2];\

  418. \

  419.         pixels+=line_size;\

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

  421.             a1= pixels[0];\

  422.             b1= pixels[1];\

  423.             a1 += b1;\

  424.             b1 += pixels[2];\

  425. \

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

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

  428. \

  429.             pixels+=line_size;\

  430.             block +=line_size;\

  431. \

  432.             a0= pixels[0];\

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

  434.             a0 += b0;\

  435.             b0 += pixels[2];\

  436. \

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

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

  439.             pixels+=line_size;\

  440.             block +=line_size;\

  441.         }\

  442. }\

  443. \

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

  445. {\

  446.         /* FIXME HIGH BIT DEPTH */\

  447.         int i;\

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

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

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

  451.                     + (b&0x03030303UL)\

  452.                     + 0x02020202UL;\

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

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

  455.         uint32_t l1,h1;\

  456. \

  457.         pixels+=line_size;\

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

  459.             uint32_t a= AV_RN32(pixels  );\

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

  461.             l1=  (a&0x03030303UL)\

  462.                + (b&0x03030303UL);\

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

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

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

  466.             pixels+=line_size;\

  467.             block +=line_size;\

  468.             a= AV_RN32(pixels  );\

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

  470.             l0=  (a&0x03030303UL)\

  471.                + (b&0x03030303UL)\

  472.                + 0x02020202UL;\

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

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

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

  476.             pixels+=line_size;\

  477.             block +=line_size;\

  478.         }\

  479. }\

  480. \

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

  482. {\

  483.     /* FIXME HIGH BIT DEPTH */\

  484.     int j;\

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

  486.         int i;\

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

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

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

  490.                     + (b&0x03030303UL)\

  491.                     + 0x02020202UL;\

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

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

  494.         uint32_t l1,h1;\

  495. \

  496.         pixels+=line_size;\

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

  498.             uint32_t a= AV_RN32(pixels  );\

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

  500.             l1=  (a&0x03030303UL)\

  501.                + (b&0x03030303UL);\

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

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

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

  505.             pixels+=line_size;\

  506.             block +=line_size;\

  507.             a= AV_RN32(pixels  );\

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

  509.             l0=  (a&0x03030303UL)\

  510.                + (b&0x03030303UL)\

  511.                + 0x02020202UL;\

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

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

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

  515.             pixels+=line_size;\

  516.             block +=line_size;\

  517.         }\

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

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

  520.     }\

  521. }\

  522. \

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

  524. {\

  525.     /* FIXME HIGH BIT DEPTH */\

  526.     int j;\

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

  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.                     + 0x01010101UL;\

  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.                + 0x01010101UL;\

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

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

  562.     }\

  563. }\

  564. \

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

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

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

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

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

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

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

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

  573. 

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

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

  576. 

  577. PIXOP2(avg, op_avg)

  578. PIXOP2(put, op_put)

  579. #undef op_avg

  580. #undef op_put

  581. 

  582. #define put_no_rnd_pixels8_c  put_pixels8_c

  583. #define put_no_rnd_pixels16_c put_pixels16_c

  584. 

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

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

  587. }

  588. 

  589. #define H264_CHROMA_MC(OPNAME, OP)\

  590. static void FUNCC(OPNAME ## h264_chroma_mc2)(uint8_t *_dst/*align 8*/, uint8_t *_src/*align 1*/, int stride, int h, int x, int y){\

  591.     pixel *dst = (pixel*)_dst;\

  592.     pixel *src = (pixel*)_src;\

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

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

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

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

  597.     int i;\

  598.     stride /= sizeof(pixel);\

  599.     \

  600.     assert(x<8 && y<8 && x>=0 && y>=0);\

  601. \

  602.     if(D){\

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

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

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

  606.             dst+= stride;\

  607.             src+= stride;\

  608.         }\

  609.     }else{\

  610.         const int E= B+C;\

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

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

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

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

  615.             dst+= stride;\

  616.             src+= stride;\

  617.         }\

  618.     }\

  619. }\

  620. \

  621. static void FUNCC(OPNAME ## h264_chroma_mc4)(uint8_t *_dst/*align 8*/, uint8_t *_src/*align 1*/, int stride, int h, int x, int y){\

  622.     pixel *dst = (pixel*)_dst;\

  623.     pixel *src = (pixel*)_src;\

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

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

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

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

  628.     int i;\

  629.     stride /= sizeof(pixel);\

  630.     \

  631.     assert(x<8 && y<8 && x>=0 && y>=0);\

  632. \

  633.     if(D){\

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

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

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

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

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

  639.             dst+= stride;\

  640.             src+= stride;\

  641.         }\

  642.     }else{\

  643.         const int E= B+C;\

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

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

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

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

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

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

  650.             dst+= stride;\

  651.             src+= stride;\

  652.         }\

  653.     }\

  654. }\

  655. \

  656. static void FUNCC(OPNAME ## h264_chroma_mc8)(uint8_t *_dst/*align 8*/, uint8_t *_src/*align 1*/, int stride, int h, int x, int y){\

  657.     pixel *dst = (pixel*)_dst;\

  658.     pixel *src = (pixel*)_src;\

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

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

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

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

  663.     int i;\

  664.     stride /= sizeof(pixel);\

  665.     \

  666.     assert(x<8 && y<8 && x>=0 && y>=0);\

  667. \

  668.     if(D){\

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

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

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

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

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

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

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

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

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

  678.             dst+= stride;\

  679.             src+= stride;\

  680.         }\

  681.     }else{\

  682.         const int E= B+C;\

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

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

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

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

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

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

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

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

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

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

  693.             dst+= stride;\

  694.             src+= stride;\

  695.         }\

  696.     }\

  697. }

  698. 

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

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

  701. 

  702. H264_CHROMA_MC(put_       , op_put)

  703. H264_CHROMA_MC(avg_       , op_avg)

  704. #undef op_avg

  705. #undef op_put

  706. 

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

  708. static av_unused void FUNC(OPNAME ## h264_qpel2_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\

  709.     const int h=2;\

  710.     INIT_CLIP\

  711.     int i;\

  712.     pixel *dst = (pixel*)_dst;\

  713.     pixel *src = (pixel*)_src;\

  714.     dstStride /= sizeof(pixel);\

  715.     srcStride /= sizeof(pixel);\

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

  717.     {\

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

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

  720.         dst+=dstStride;\

  721.         src+=srcStride;\

  722.     }\

  723. }\

  724. \

  725. static av_unused void FUNC(OPNAME ## h264_qpel2_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\

  726.     const int w=2;\

  727.     INIT_CLIP\

  728.     int i;\

  729.     pixel *dst = (pixel*)_dst;\

  730.     pixel *src = (pixel*)_src;\

  731.     dstStride /= sizeof(pixel);\

  732.     srcStride /= sizeof(pixel);\

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

  734.     {\

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

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

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

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

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

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

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

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

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

  744.         dst++;\

  745.         src++;\

  746.     }\

  747. }\

  748. \

  749. static av_unused void FUNC(OPNAME ## h264_qpel2_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\

  750.     const int h=2;\

  751.     const int w=2;\

  752.     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\

  753.     INIT_CLIP\

  754.     int i;\

  755.     pixel *dst = (pixel*)_dst;\

  756.     pixel *src = (pixel*)_src;\

  757.     dstStride /= sizeof(pixel);\

  758.     srcStride /= sizeof(pixel);\

  759.     src -= 2*srcStride;\

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

  761.     {\

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

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

  764.         tmp+=tmpStride;\

  765.         src+=srcStride;\

  766.     }\

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

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

  769.     {\

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

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

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

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

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

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

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

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

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

  779.         dst++;\

  780.         tmp++;\

  781.     }\

  782. }\

  783. static void FUNC(OPNAME ## h264_qpel4_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\

  784.     const int h=4;\

  785.     INIT_CLIP\

  786.     int i;\

  787.     pixel *dst = (pixel*)_dst;\

  788.     pixel *src = (pixel*)_src;\

  789.     dstStride /= sizeof(pixel);\

  790.     srcStride /= sizeof(pixel);\

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

  792.     {\

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

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

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

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

  797.         dst+=dstStride;\

  798.         src+=srcStride;\

  799.     }\

  800. }\

  801. \

  802. static void FUNC(OPNAME ## h264_qpel4_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\

  803.     const int w=4;\

  804.     INIT_CLIP\

  805.     int i;\

  806.     pixel *dst = (pixel*)_dst;\

  807.     pixel *src = (pixel*)_src;\

  808.     dstStride /= sizeof(pixel);\

  809.     srcStride /= sizeof(pixel);\

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

  811.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

  825.         dst++;\

  826.         src++;\

  827.     }\

  828. }\

  829. \

  830. static void FUNC(OPNAME ## h264_qpel4_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\

  831.     const int h=4;\

  832.     const int w=4;\

  833.     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\

  834.     INIT_CLIP\

  835.     int i;\

  836.     pixel *dst = (pixel*)_dst;\

  837.     pixel *src = (pixel*)_src;\

  838.     dstStride /= sizeof(pixel);\

  839.     srcStride /= sizeof(pixel);\

  840.     src -= 2*srcStride;\

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

  842.     {\

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

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

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

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

  847.         tmp+=tmpStride;\

  848.         src+=srcStride;\

  849.     }\

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

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

  852.     {\

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

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

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

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

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

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

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

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

  861.         const int tmp6= tmp[6 *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.         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\

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

  866.         dst++;\

  867.         tmp++;\

  868.     }\

  869. }\

  870. \

  871. static void FUNC(OPNAME ## h264_qpel8_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\

  872.     const int h=8;\

  873.     INIT_CLIP\

  874.     int i;\

  875.     pixel *dst = (pixel*)_dst;\

  876.     pixel *src = (pixel*)_src;\

  877.     dstStride /= sizeof(pixel);\

  878.     srcStride /= sizeof(pixel);\

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

  880.     {\

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

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

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

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

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

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

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

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

  889.         dst+=dstStride;\

  890.         src+=srcStride;\

  891.     }\

  892. }\

  893. \

  894. static void FUNC(OPNAME ## h264_qpel8_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\

  895.     const int w=8;\

  896.     INIT_CLIP\

  897.     int i;\

  898.     pixel *dst = (pixel*)_dst;\

  899.     pixel *src = (pixel*)_src;\

  900.     dstStride /= sizeof(pixel);\

  901.     srcStride /= sizeof(pixel);\

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

  903.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

  925.         dst++;\

  926.         src++;\

  927.     }\

  928. }\

  929. \

  930. static void FUNC(OPNAME ## h264_qpel8_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\

  931.     const int h=8;\

  932.     const int w=8;\

  933.     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\

  934.     INIT_CLIP\

  935.     int i;\

  936.     pixel *dst = (pixel*)_dst;\

  937.     pixel *src = (pixel*)_src;\

  938.     dstStride /= sizeof(pixel);\

  939.     srcStride /= sizeof(pixel);\

  940.     src -= 2*srcStride;\

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

  942.     {\

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

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

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

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

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

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

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

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

  951.         tmp+=tmpStride;\

  952.         src+=srcStride;\

  953.     }\

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

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

  956.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

  978.         dst++;\

  979.         tmp++;\

  980.     }\

  981. }\

  982. \

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

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

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

  986.     src += 8*srcStride;\

  987.     dst += 8*dstStride;\

  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. }\

  991. \

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

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

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

  995.     src += 8*srcStride;\

  996.     dst += 8*dstStride;\

  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. }\

  1000. \

  1001. static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  1002.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\

  1003.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\

  1004.     src += 8*srcStride;\

  1005.     dst += 8*dstStride;\

  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. }\

  1009. 

  1010. #define H264_MC(OPNAME, SIZE) \

  1011. static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\

  1012.     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\

  1013. }\

  1014. \

  1015. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\

  1016.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1017.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1018.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1019. }\

  1020. \

  1021. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\

  1022.     FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\

  1023. }\

  1024. \

  1025. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\

  1026.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1027.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1028.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1029. }\

  1030. \

  1031. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\

  1032.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1033.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1034.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1035.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1036.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1037.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1038. }\

  1039. \

  1040. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\

  1041.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1042.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1043.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1044.     FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\

  1045. }\

  1046. \

  1047. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\

  1048.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1049.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1050.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1051.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1052.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1053.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1054. }\

  1055. \

  1056. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(uint8_t *dst, uint8_t *src, int stride){\

  1057.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1058.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1059.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1060.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1061.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1062.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1063.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1064.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1065. }\

  1066. \

  1067. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\

  1068.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1069.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1070.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1071.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1072.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1073.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1074.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1075.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1076. }\

  1077. \

  1078. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\

  1079.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1080.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1081.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1082.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1083.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1084.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1085.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1086.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1087. }\

  1088. \

  1089. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\

  1090.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1091.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1092.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1093.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1094.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1095.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1096.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1097.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1098. }\

  1099. \

  1100. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\

  1101.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1102.     FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\

  1103. }\

  1104. \

  1105. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\

  1106.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1107.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1108.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1109.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\

  1110.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1111.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1112. }\

  1113. \

  1114. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\

  1115.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1116.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1117.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1118.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1119.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1120.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1121. }\

  1122. \

  1123. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\

  1124.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1125.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1126.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1127.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1128.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1129.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1130.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1131.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1132.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1133. }\

  1134. \

  1135. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\

  1136.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1137.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1138.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1139.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1140.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1141.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1142.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1143.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1144.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1145. }\

  1146. 

  1147. #define op_avg(a, b)  a = (((a)+CLIP(((b) + 16)>>5)+1)>>1)

  1148. //#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)

  1149. #define op_put(a, b)  a = CLIP(((b) + 16)>>5)

  1150. #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)

  1151. #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)

  1152. 

  1153. H264_LOWPASS(put_       , op_put, op2_put)

  1154. H264_LOWPASS(avg_       , op_avg, op2_avg)

  1155. H264_MC(put_, 2)

  1156. H264_MC(put_, 4)

  1157. H264_MC(put_, 8)

  1158. H264_MC(put_, 16)

  1159. H264_MC(avg_, 4)

  1160. H264_MC(avg_, 8)

  1161. H264_MC(avg_, 16)

  1162. 

  1163. #undef op_avg

  1164. #undef op_put

  1165. #undef op2_avg

  1166. #undef op2_put

  1167. 

  1168. #if BIT_DEPTH == 8

  1169. #   define put_h264_qpel8_mc00_8_c  ff_put_pixels8x8_8_c

  1170. #   define avg_h264_qpel8_mc00_8_c  ff_avg_pixels8x8_8_c

  1171. #   define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c

  1172. #   define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c

  1173. #elif BIT_DEPTH == 9

  1174. #   define put_h264_qpel8_mc00_9_c  ff_put_pixels8x8_9_c

  1175. #   define avg_h264_qpel8_mc00_9_c  ff_avg_pixels8x8_9_c

  1176. #   define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c

  1177. #   define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c

  1178. #elif BIT_DEPTH == 10

  1179. #   define put_h264_qpel8_mc00_10_c  ff_put_pixels8x8_10_c

  1180. #   define avg_h264_qpel8_mc00_10_c  ff_avg_pixels8x8_10_c

  1181. #   define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c

  1182. #   define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c

  1183. #endif

  1184. 

  1185. void FUNCC(ff_put_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1186.     FUNCC(put_pixels8)(dst, src, stride, 8);

  1187. }

  1188. void FUNCC(ff_avg_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1189.     FUNCC(avg_pixels8)(dst, src, stride, 8);

  1190. }

  1191. void FUNCC(ff_put_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1192.     FUNCC(put_pixels16)(dst, src, stride, 16);

  1193. }

  1194. void FUNCC(ff_avg_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1195.     FUNCC(avg_pixels16)(dst, src, stride, 16);

  1196. }