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.
29565 Commits
32 Branches
353MB
Tree: e5985185d2
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'e5985185d2'
${ noResults }
FFmpeg/libavcodec/dsputil_template.c

1247 lines
50KB
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. /**

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

  118.  * @param buf destination buffer

  119.  * @param src source buffer

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

  121.  * @param block_w width of block

  122.  * @param block_h height of block

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

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

  125.  * @param w width of the source buffer

  126.  * @param h height of the source buffer

  127.  */

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

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

  130.     int x, y;

  131.     int start_y, start_x, end_y, end_x;

  132. 

  133.     if(src_y>= h){

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

  135.         src_y=h-1;

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

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

  138.         src_y=1-block_h;

  139.     }

  140.     if(src_x>= w){

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

  142.         src_x=w-1;

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

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

  145.         src_x=1-block_w;

  146.     }

  147. 

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

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

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

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

  152.     assert(start_y < end_y && block_h);

  153.     assert(start_x < end_x && block_w);

  154. 

  155.     w    = end_x - start_x;

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

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

  158. 

  159.     //top

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

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

  162.         buf += linesize;

  163.     }

  164. 

  165.     // copy existing part

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

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

  168.         src += linesize;

  169.         buf += linesize;

  170.     }

  171. 

  172.     //bottom

  173.     src -= linesize;

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

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

  176.         buf += linesize;

  177.     }

  178. 

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

  180.     while (block_h--){

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

  182.        //left

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

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

  185.         }

  186. 

  187.        //right

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

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

  190.         }

  191.         buf += linesize;

  192.     }

  193. }

  194. 

  195. static void FUNCC(add_pixels8)(uint8_t *restrict _pixels, DCTELEM *_block, int line_size)

  196. {

  197.     int i;

  198.     pixel *restrict pixels = (pixel *restrict)_pixels;

  199.     dctcoef *block = (dctcoef*)_block;

  200.     line_size /= sizeof(pixel);

  201. 

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

  203.         pixels[0] += block[0];

  204.         pixels[1] += block[1];

  205.         pixels[2] += block[2];

  206.         pixels[3] += block[3];

  207.         pixels[4] += block[4];

  208.         pixels[5] += block[5];

  209.         pixels[6] += block[6];

  210.         pixels[7] += block[7];

  211.         pixels += line_size;

  212.         block += 8;

  213.     }

  214. }

  215. 

  216. static void FUNCC(add_pixels4)(uint8_t *restrict _pixels, DCTELEM *_block, int line_size)

  217. {

  218.     int i;

  219.     pixel *restrict pixels = (pixel *restrict)_pixels;

  220.     dctcoef *block = (dctcoef*)_block;

  221.     line_size /= sizeof(pixel);

  222. 

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

  224.         pixels[0] += block[0];

  225.         pixels[1] += block[1];

  226.         pixels[2] += block[2];

  227.         pixels[3] += block[3];

  228.         pixels += line_size;

  229.         block += 4;

  230.     }

  231. }

  232. 

  233. #define PIXOP2(OPNAME, OP) \

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

  235.     int i;\

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

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

  238.         pixels+=line_size;\

  239.         block +=line_size;\

  240.     }\

  241. }\

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

  243.     int i;\

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

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

  246.         pixels+=line_size;\

  247.         block +=line_size;\

  248.     }\

  249. }\

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

  251.     int i;\

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

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

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

  255.         pixels+=line_size;\

  256.         block +=line_size;\

  257.     }\

  258. }\

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

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

  261. }\

  262. \

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

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

  265.     int i;\

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

  267.         pixel4 a,b;\

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

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

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

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

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

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

  274.     }\

  275. }\

  276. \

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

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

  279.     int i;\

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

  281.         pixel4 a,b;\

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

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

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

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

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

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

  288.     }\

  289. }\

  290. \

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

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

  293.     int i;\

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

  295.         pixel4 a,b;\

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

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

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

  299.     }\

  300. }\

  301. \

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

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

  304.     int i;\

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

  306.         pixel4 a,b;\

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

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

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

  310.     }\

  311. }\

  312. \

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

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

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

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

  317. }\

  318. \

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

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

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

  322.     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);\

  323. }\

  324. \

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

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

  327. }\

  328. \

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

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

  331. }\

  332. \

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

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

  335. }\

  336. \

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

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

  339. }\

  340. \

  341. 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,\

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

  343.     /* FIXME HIGH BIT DEPTH */\

  344.     int i;\

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

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

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

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

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

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

  351.         l0=  (a&0x03030303UL)\

  352.            + (b&0x03030303UL)\

  353.            + 0x02020202UL;\

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

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

  356.         l1=  (c&0x03030303UL)\

  357.            + (d&0x03030303UL);\

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

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

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

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

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

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

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

  365.         l0=  (a&0x03030303UL)\

  366.            + (b&0x03030303UL)\

  367.            + 0x02020202UL;\

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

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

  370.         l1=  (c&0x03030303UL)\

  371.            + (d&0x03030303UL);\

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

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

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

  375.     }\

  376. }\

  377. \

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

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

  380. }\

  381. \

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

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

  384. }\

  385. \

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

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

  388. }\

  389. \

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

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

  392. }\

  393. \

  394. 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,\

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

  396.     /* FIXME HIGH BIT DEPTH*/\

  397.     int i;\

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

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

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

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

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

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

  404.         l0=  (a&0x03030303UL)\

  405.            + (b&0x03030303UL)\

  406.            + 0x01010101UL;\

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

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

  409.         l1=  (c&0x03030303UL)\

  410.            + (d&0x03030303UL);\

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

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

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

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

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

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

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

  418.         l0=  (a&0x03030303UL)\

  419.            + (b&0x03030303UL)\

  420.            + 0x01010101UL;\

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

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

  423.         l1=  (c&0x03030303UL)\

  424.            + (d&0x03030303UL);\

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

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

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

  428.     }\

  429. }\

  430. 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,\

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

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

  433.     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);\

  434. }\

  435. 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,\

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

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

  438.     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);\

  439. }\

  440. \

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

  442. {\

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

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

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

  446.         line_size /= sizeof(pixel);\

  447.         a0= pixels[0];\

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

  449.         a0 += b0;\

  450.         b0 += pixels[2];\

  451. \

  452.         pixels+=line_size;\

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

  454.             a1= pixels[0];\

  455.             b1= pixels[1];\

  456.             a1 += b1;\

  457.             b1 += pixels[2];\

  458. \

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

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

  461. \

  462.             pixels+=line_size;\

  463.             block +=line_size;\

  464. \

  465.             a0= pixels[0];\

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

  467.             a0 += b0;\

  468.             b0 += pixels[2];\

  469. \

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

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

  472.             pixels+=line_size;\

  473.             block +=line_size;\

  474.         }\

  475. }\

  476. \

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

  478. {\

  479.         /* FIXME HIGH BIT DEPTH */\

  480.         int i;\

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

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

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

  484.                     + (b&0x03030303UL)\

  485.                     + 0x02020202UL;\

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

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

  488.         uint32_t l1,h1;\

  489. \

  490.         pixels+=line_size;\

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

  492.             uint32_t a= AV_RN32(pixels  );\

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

  494.             l1=  (a&0x03030303UL)\

  495.                + (b&0x03030303UL);\

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

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

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

  499.             pixels+=line_size;\

  500.             block +=line_size;\

  501.             a= AV_RN32(pixels  );\

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

  503.             l0=  (a&0x03030303UL)\

  504.                + (b&0x03030303UL)\

  505.                + 0x02020202UL;\

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

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

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

  509.             pixels+=line_size;\

  510.             block +=line_size;\

  511.         }\

  512. }\

  513. \

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

  515. {\

  516.     /* FIXME HIGH BIT DEPTH */\

  517.     int j;\

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

  519.         int i;\

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

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

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

  523.                     + (b&0x03030303UL)\

  524.                     + 0x02020202UL;\

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

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

  527.         uint32_t l1,h1;\

  528. \

  529.         pixels+=line_size;\

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

  531.             uint32_t a= AV_RN32(pixels  );\

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

  533.             l1=  (a&0x03030303UL)\

  534.                + (b&0x03030303UL);\

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

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

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

  538.             pixels+=line_size;\

  539.             block +=line_size;\

  540.             a= AV_RN32(pixels  );\

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

  542.             l0=  (a&0x03030303UL)\

  543.                + (b&0x03030303UL)\

  544.                + 0x02020202UL;\

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

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

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

  548.             pixels+=line_size;\

  549.             block +=line_size;\

  550.         }\

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

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

  553.     }\

  554. }\

  555. \

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

  557. {\

  558.     /* FIXME HIGH BIT DEPTH */\

  559.     int j;\

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

  561.         int i;\

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

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

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

  565.                     + (b&0x03030303UL)\

  566.                     + 0x01010101UL;\

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

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

  569.         uint32_t l1,h1;\

  570. \

  571.         pixels+=line_size;\

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

  573.             uint32_t a= AV_RN32(pixels  );\

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

  575.             l1=  (a&0x03030303UL)\

  576.                + (b&0x03030303UL);\

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

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

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

  580.             pixels+=line_size;\

  581.             block +=line_size;\

  582.             a= AV_RN32(pixels  );\

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

  584.             l0=  (a&0x03030303UL)\

  585.                + (b&0x03030303UL)\

  586.                + 0x01010101UL;\

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

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

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

  590.             pixels+=line_size;\

  591.             block +=line_size;\

  592.         }\

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

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

  595.     }\

  596. }\

  597. \

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

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

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

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

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

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

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

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

  606. 

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

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

  609. 

  610. PIXOP2(avg, op_avg)

  611. PIXOP2(put, op_put)

  612. #undef op_avg

  613. #undef op_put

  614. 

  615. #define put_no_rnd_pixels8_c  put_pixels8_c

  616. #define put_no_rnd_pixels16_c put_pixels16_c

  617. 

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

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

  620. }

  621. 

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

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

  624. }

  625. 

  626. #define H264_CHROMA_MC(OPNAME, OP)\

  627. 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){\

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

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

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

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

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

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

  634.     int i;\

  635.     stride /= sizeof(pixel);\

  636.     \

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

  638. \

  639.     if(D){\

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

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

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

  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.             dst+= stride;\

  653.             src+= stride;\

  654.         }\

  655.     }\

  656. }\

  657. \

  658. 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){\

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

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

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

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

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

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

  665.     int i;\

  666.     stride /= sizeof(pixel);\

  667.     \

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

  669. \

  670.     if(D){\

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

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

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

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

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

  676.             dst+= stride;\

  677.             src+= stride;\

  678.         }\

  679.     }else{\

  680.         const int E= B+C;\

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

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

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

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

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

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

  687.             dst+= stride;\

  688.             src+= stride;\

  689.         }\

  690.     }\

  691. }\

  692. \

  693. 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){\

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

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

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

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

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

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

  700.     int i;\

  701.     stride /= sizeof(pixel);\

  702.     \

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

  704. \

  705.     if(D){\

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

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

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

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

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

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

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

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

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

  715.             dst+= stride;\

  716.             src+= stride;\

  717.         }\

  718.     }else{\

  719.         const int E= B+C;\

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

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

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

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

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

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

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

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

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

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

  730.             dst+= stride;\

  731.             src+= stride;\

  732.         }\

  733.     }\

  734. }

  735. 

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

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

  738. 

  739. H264_CHROMA_MC(put_       , op_put)

  740. H264_CHROMA_MC(avg_       , op_avg)

  741. #undef op_avg

  742. #undef op_put

  743. 

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

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

  746.     const int h=2;\

  747.     INIT_CLIP\

  748.     int i;\

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

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

  751.     dstStride /= sizeof(pixel);\

  752.     srcStride /= sizeof(pixel);\

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

  754.     {\

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

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

  757.         dst+=dstStride;\

  758.         src+=srcStride;\

  759.     }\

  760. }\

  761. \

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

  763.     const int w=2;\

  764.     INIT_CLIP\

  765.     int i;\

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

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

  768.     dstStride /= sizeof(pixel);\

  769.     srcStride /= sizeof(pixel);\

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

  771.     {\

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

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

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

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

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

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

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

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

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

  781.         dst++;\

  782.         src++;\

  783.     }\

  784. }\

  785. \

  786. 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){\

  787.     const int h=2;\

  788.     const int w=2;\

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

  790.     INIT_CLIP\

  791.     int i;\

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

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

  794.     dstStride /= sizeof(pixel);\

  795.     srcStride /= sizeof(pixel);\

  796.     src -= 2*srcStride;\

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

  798.     {\

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

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

  801.         tmp+=tmpStride;\

  802.         src+=srcStride;\

  803.     }\

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

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

  806.     {\

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

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

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

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

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

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

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

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

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

  816.         dst++;\

  817.         tmp++;\

  818.     }\

  819. }\

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

  821.     const int h=4;\

  822.     INIT_CLIP\

  823.     int i;\

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

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

  826.     dstStride /= sizeof(pixel);\

  827.     srcStride /= sizeof(pixel);\

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

  829.     {\

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

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

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

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

  834.         dst+=dstStride;\

  835.         src+=srcStride;\

  836.     }\

  837. }\

  838. \

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

  840.     const int w=4;\

  841.     INIT_CLIP\

  842.     int i;\

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

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

  845.     dstStride /= sizeof(pixel);\

  846.     srcStride /= sizeof(pixel);\

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

  848.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

  862.         dst++;\

  863.         src++;\

  864.     }\

  865. }\

  866. \

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

  868.     const int h=4;\

  869.     const int w=4;\

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

  871.     INIT_CLIP\

  872.     int i;\

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

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

  875.     dstStride /= sizeof(pixel);\

  876.     srcStride /= sizeof(pixel);\

  877.     src -= 2*srcStride;\

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

  879.     {\

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

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

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

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

  884.         tmp+=tmpStride;\

  885.         src+=srcStride;\

  886.     }\

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

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

  889.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

  903.         dst++;\

  904.         tmp++;\

  905.     }\

  906. }\

  907. \

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

  909.     const int h=8;\

  910.     INIT_CLIP\

  911.     int i;\

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

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

  914.     dstStride /= sizeof(pixel);\

  915.     srcStride /= sizeof(pixel);\

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

  917.     {\

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

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

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

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

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

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

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

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

  926.         dst+=dstStride;\

  927.         src+=srcStride;\

  928.     }\

  929. }\

  930. \

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

  932.     const int w=8;\

  933.     INIT_CLIP\

  934.     int i;\

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

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

  937.     dstStride /= sizeof(pixel);\

  938.     srcStride /= sizeof(pixel);\

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

  940.     {\

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

  962.         dst++;\

  963.         src++;\

  964.     }\

  965. }\

  966. \

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

  968.     const int h=8;\

  969.     const int w=8;\

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

  971.     INIT_CLIP\

  972.     int i;\

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

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

  975.     dstStride /= sizeof(pixel);\

  976.     srcStride /= sizeof(pixel);\

  977.     src -= 2*srcStride;\

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

  979.     {\

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

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

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

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

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

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

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

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

  988.         tmp+=tmpStride;\

  989.         src+=srcStride;\

  990.     }\

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

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

  993.     {\

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

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

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

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

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

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

  1000.         const int tmp4= tmp[4 *tmpStride] - pad;\

  1001.         const int tmp5= tmp[5 *tmpStride] - pad;\

  1002.         const int tmp6= tmp[6 *tmpStride] - pad;\

  1003.         const int tmp7= tmp[7 *tmpStride] - pad;\

  1004.         const int tmp8= tmp[8 *tmpStride] - pad;\

  1005.         const int tmp9= tmp[9 *tmpStride] - pad;\

  1006.         const int tmp10=tmp[10*tmpStride] - pad;\

  1007.         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\

  1008.         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\

  1009.         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\

  1010.         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\

  1011.         OP2(dst[4*dstStride], (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));\

  1012.         OP2(dst[5*dstStride], (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));\

  1013.         OP2(dst[6*dstStride], (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));\

  1014.         OP2(dst[7*dstStride], (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));\

  1015.         dst++;\

  1016.         tmp++;\

  1017.     }\

  1018. }\

  1019. \

  1020. static void FUNC(OPNAME ## h264_qpel16_v_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  1021.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\

  1022.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1023.     src += 8*srcStride;\

  1024.     dst += 8*dstStride;\

  1025.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\

  1026.     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1027. }\

  1028. \

  1029. static void FUNC(OPNAME ## h264_qpel16_h_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  1030.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\

  1031.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1032.     src += 8*srcStride;\

  1033.     dst += 8*dstStride;\

  1034.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\

  1035.     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\

  1036. }\

  1037. \

  1038. static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  1039.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\

  1040.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\

  1041.     src += 8*srcStride;\

  1042.     dst += 8*dstStride;\

  1043.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\

  1044.     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\

  1045. }\

  1046. 

  1047. #define H264_MC(OPNAME, SIZE) \

  1048. static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\

  1049.     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\

  1050. }\

  1051. \

  1052. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\

  1053.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

  1054.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1055.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1056. }\

  1057. \

  1058. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\

  1059.     FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\

  1060. }\

  1061. \

  1062. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\

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

  1064.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\

  1065.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\

  1066. }\

  1067. \

  1068. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\

  1069.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1070.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

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

  1072.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1073.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1074.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1075. }\

  1076. \

  1077. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\

  1078.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1079.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1080.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1081.     FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\

  1082. }\

  1083. \

  1084. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\

  1085.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1086.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1087.     uint8_t half[SIZE*SIZE*sizeof(pixel)];\

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

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

  1090.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1091. }\

  1092. \

  1093. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(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, SIZE*sizeof(pixel), stride);\

  1099.     FUNC(copy_block ## SIZE )(full, src - stride*2, 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 ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\

  1105.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1106.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

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

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

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

  1110.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1111.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1112.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1113. }\

  1114. \

  1115. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\

  1116.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1117.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1118.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

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

  1120.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1121.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1122.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1123.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1124. }\

  1125. \

  1126. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\

  1127.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1128.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1129.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1130.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1131.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1132.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1133.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1134.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1135. }\

  1136. \

  1137. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\

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

  1139.     FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\

  1140. }\

  1141. \

  1142. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\

  1143.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

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

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

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

  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, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1149. }\

  1150. \

  1151. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\

  1152.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1153.     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\

  1154.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1155.     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\

  1156.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1157.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1158. }\

  1159. \

  1160. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\

  1161.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1162.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1163.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1164.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

  1165.     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\

  1166.     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1167.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1168.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1169.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1170. }\

  1171. \

  1172. static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\

  1173.     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\

  1174.     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\

  1175.     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\

  1176.     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\

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

  1178.     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\

  1179.     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\

  1180.     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\

  1181.     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\

  1182. }\

  1183. 

  1184. #define op_avg(a, b)  a = (((a)+CLIP(((b) + 16)>>5)+1)>>1)

  1185. //#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)

  1186. #define op_put(a, b)  a = CLIP(((b) + 16)>>5)

  1187. #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)

  1188. #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)

  1189. 

  1190. H264_LOWPASS(put_       , op_put, op2_put)

  1191. H264_LOWPASS(avg_       , op_avg, op2_avg)

  1192. H264_MC(put_, 2)

  1193. H264_MC(put_, 4)

  1194. H264_MC(put_, 8)

  1195. H264_MC(put_, 16)

  1196. H264_MC(avg_, 4)

  1197. H264_MC(avg_, 8)

  1198. H264_MC(avg_, 16)

  1199. 

  1200. #undef op_avg

  1201. #undef op_put

  1202. #undef op2_avg

  1203. #undef op2_put

  1204. 

  1205. #if BIT_DEPTH == 8

  1206. #   define put_h264_qpel8_mc00_8_c  ff_put_pixels8x8_8_c

  1207. #   define avg_h264_qpel8_mc00_8_c  ff_avg_pixels8x8_8_c

  1208. #   define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c

  1209. #   define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c

  1210. #elif BIT_DEPTH == 9

  1211. #   define put_h264_qpel8_mc00_9_c  ff_put_pixels8x8_9_c

  1212. #   define avg_h264_qpel8_mc00_9_c  ff_avg_pixels8x8_9_c

  1213. #   define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c

  1214. #   define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c

  1215. #elif BIT_DEPTH == 10

  1216. #   define put_h264_qpel8_mc00_10_c  ff_put_pixels8x8_10_c

  1217. #   define avg_h264_qpel8_mc00_10_c  ff_avg_pixels8x8_10_c

  1218. #   define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c

  1219. #   define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c

  1220. #endif

  1221. 

  1222. void FUNCC(ff_put_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1223.     FUNCC(put_pixels8)(dst, src, stride, 8);

  1224. }

  1225. void FUNCC(ff_avg_pixels8x8)(uint8_t *dst, uint8_t *src, int stride) {

  1226.     FUNCC(avg_pixels8)(dst, src, stride, 8);

  1227. }

  1228. void FUNCC(ff_put_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1229.     FUNCC(put_pixels16)(dst, src, stride, 16);

  1230. }

  1231. void FUNCC(ff_avg_pixels16x16)(uint8_t *dst, uint8_t *src, int stride) {

  1232.     FUNCC(avg_pixels16)(dst, src, stride, 16);

  1233. }

  1234. 

  1235. static void FUNCC(clear_block)(DCTELEM *block)

  1236. {

  1237.     memset(block, 0, sizeof(dctcoef)*64);

  1238. }

  1239. 

  1240. /**

  1241.  * memset(blocks, 0, sizeof(DCTELEM)*6*64)

  1242.  */

  1243. static void FUNCC(clear_blocks)(DCTELEM *blocks)

  1244. {

  1245.     memset(blocks, 0, sizeof(dctcoef)*6*64);

  1246. }