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.
49674 Commits
32 Branches
649MB
Tree: b5884db437
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'b5884db437'
${ noResults }
FFmpeg/libavcodec/x86/h264_qpel.c

627 lines
30KB
Raw Blame History 

  1. /*

  2.  * Copyright (c) 2004-2005 Michael Niedermayer, Loren Merritt

  3.  * Copyright (c) 2011 Daniel Kang

  4.  *

  5.  * This file is part of FFmpeg.

  6.  *

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

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

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

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

  11.  *

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

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

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

  15.  * Lesser General Public License for more details.

  16.  *

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

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

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

  20.  */

  21. 

  22. #include "libavutil/attributes.h"

  23. #include "libavutil/cpu.h"

  24. #include "libavutil/x86/asm.h"

  25. #include "libavutil/x86/cpu.h"

  26. #include "libavcodec/dsputil.h"

  27. #include "libavcodec/h264qpel.h"

  28. #include "libavcodec/mpegvideo.h"

  29. #include "dsputil_mmx.h"

  30. 

  31. #if HAVE_YASM

  32. void ff_put_pixels4_mmxext(uint8_t *block, const uint8_t *pixels, int line_size, int h);

  33. void ff_avg_pixels4_mmxext(uint8_t *block, const uint8_t *pixels, int line_size, int h);

  34. void ff_put_pixels8_mmxext(uint8_t *block, const uint8_t *pixels, int line_size, int h);

  35. void ff_avg_pixels8_mmxext(uint8_t *block, const uint8_t *pixels, int line_size, int h);

  36. static void ff_put_pixels16_mmxext(uint8_t *block, const uint8_t *pixels,

  37.                                    int line_size, int h)

  38. {

  39.     ff_put_pixels8_mmxext(block,     pixels,     line_size, h);

  40.     ff_put_pixels8_mmxext(block + 8, pixels + 8, line_size, h);

  41. }

  42. static void ff_avg_pixels16_mmxext(uint8_t *block, const uint8_t *pixels,

  43.                                    int line_size, int h)

  44. {

  45.     ff_avg_pixels8_mmxext(block,     pixels,     line_size, h);

  46.     ff_avg_pixels8_mmxext(block + 8, pixels + 8, line_size, h);

  47. }

  48. void ff_put_pixels4_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,

  49.                               int dstStride, int src1Stride, int h);

  50. void ff_avg_pixels4_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,

  51.                               int dstStride, int src1Stride, int h);

  52. void ff_put_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,

  53.                               int dstStride, int src1Stride, int h);

  54. void ff_avg_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,

  55.                               int dstStride, int src1Stride, int h);

  56. void ff_put_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,

  57.                                int dstStride, int src1Stride, int h);

  58. void ff_avg_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,

  59.                                int dstStride, int src1Stride, int h);

  60. void ff_put_pixels16_sse2(uint8_t *block, const uint8_t *pixels,

  61.                           int line_size, int h);

  62. void ff_avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels,

  63.                           int line_size, int h);

  64. #define ff_put_pixels8_l2_sse2  ff_put_pixels8_l2_mmxext

  65. #define ff_avg_pixels8_l2_sse2  ff_avg_pixels8_l2_mmxext

  66. #define ff_put_pixels16_l2_sse2 ff_put_pixels16_l2_mmxext

  67. #define ff_avg_pixels16_l2_sse2 ff_avg_pixels16_l2_mmxext

  68. 

  69. #define DEF_QPEL(OPNAME)\

  70. void ff_ ## OPNAME ## _h264_qpel4_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride);\

  71. void ff_ ## OPNAME ## _h264_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride);\

  72. void ff_ ## OPNAME ## _h264_qpel8_h_lowpass_ssse3(uint8_t *dst, uint8_t *src, int dstStride, int srcStride);\

  73. void ff_ ## OPNAME ## _h264_qpel4_h_lowpass_l2_mmxext(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride);\

  74. void ff_ ## OPNAME ## _h264_qpel8_h_lowpass_l2_mmxext(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride);\

  75. void ff_ ## OPNAME ## _h264_qpel8_h_lowpass_l2_ssse3(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride);\

  76. void ff_ ## OPNAME ## _h264_qpel4_v_lowpass_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride);\

  77. void ff_ ## OPNAME ## _h264_qpel8or16_v_lowpass_op_mmxext(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h);\

  78. void ff_ ## OPNAME ## _h264_qpel8or16_v_lowpass_sse2(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h);\

  79. void ff_ ## OPNAME ## _h264_qpel4_hv_lowpass_v_mmxext(uint8_t *src, int16_t *tmp, int srcStride);\

  80. void ff_ ## OPNAME ## _h264_qpel4_hv_lowpass_h_mmxext(int16_t *tmp, uint8_t *dst, int dstStride);\

  81. void ff_ ## OPNAME ## _h264_qpel8or16_hv1_lowpass_op_mmxext(uint8_t *src, int16_t *tmp, int srcStride, int size);\

  82. void ff_ ## OPNAME ## _h264_qpel8or16_hv1_lowpass_op_sse2(uint8_t *src, int16_t *tmp, int srcStride, int size);\

  83. void ff_ ## OPNAME ## _h264_qpel8or16_hv2_lowpass_op_mmxext(uint8_t *dst, int16_t *tmp, int dstStride, int unused, int h);\

  84. void ff_ ## OPNAME ## _h264_qpel8or16_hv2_lowpass_ssse3(uint8_t *dst, int16_t *tmp, int dstStride, int tmpStride, int size);\

  85. void ff_ ## OPNAME ## _pixels4_l2_shift5_mmxext(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h);\

  86. void ff_ ## OPNAME ## _pixels8_l2_shift5_mmxext(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h);

  87. 

  88. DEF_QPEL(avg)

  89. DEF_QPEL(put)

  90. 

  91. #define QPEL_H264(OPNAME, OP, MMX)\

  92. static av_always_inline void ff_ ## OPNAME ## h264_qpel4_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  93.     int w=3;\

  94.     src -= 2*srcStride+2;\

  95.     while(w--){\

  96.         ff_ ## OPNAME ## h264_qpel4_hv_lowpass_v_mmxext(src, tmp, srcStride);\

  97.         tmp += 4;\

  98.         src += 4;\

  99.     }\

  100.     tmp -= 3*4;\

  101.     ff_ ## OPNAME ## h264_qpel4_hv_lowpass_h_mmxext(tmp, dst, dstStride);\

  102. }\

  103. \

  104. static av_always_inline void ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\

  105.     src -= 2*srcStride;\

  106.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_op_mmxext(dst, src, dstStride, srcStride, h);\

  107.     src += 4;\

  108.     dst += 4;\

  109.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_op_mmxext(dst, src, dstStride, srcStride, h);\

  110. }\

  111. static av_always_inline void ff_ ## OPNAME ## h264_qpel8or16_hv1_lowpass_ ## MMX(int16_t *tmp, uint8_t *src, int tmpStride, int srcStride, int size){\

  112.     int w = (size+8)>>2;\

  113.     src -= 2*srcStride+2;\

  114.     while(w--){\

  115.         ff_ ## OPNAME ## h264_qpel8or16_hv1_lowpass_op_mmxext(src, tmp, srcStride, size);\

  116.         tmp += 4;\

  117.         src += 4;\

  118.     }\

  119. }\

  120. static av_always_inline void ff_ ## OPNAME ## h264_qpel8or16_hv2_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, int dstStride, int tmpStride, int size){\

  121.     int w = size>>4;\

  122.     do{\

  123.     ff_ ## OPNAME ## h264_qpel8or16_hv2_lowpass_op_mmxext(dst, tmp, dstStride, 0, size);\

  124.     tmp += 8;\

  125.     dst += 8;\

  126.     }while(w--);\

  127. }\

  128. \

  129. static av_always_inline void ff_ ## OPNAME ## h264_qpel8_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  130.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst  , src  , dstStride, srcStride, 8);\

  131. }\

  132. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  133.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst  , src  , dstStride, srcStride, 16);\

  134.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\

  135. }\

  136. \

  137. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  138.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst  , src  , dstStride, srcStride);\

  139.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\

  140.     src += 8*srcStride;\

  141.     dst += 8*dstStride;\

  142.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst  , src  , dstStride, srcStride);\

  143.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\

  144. }\

  145. \

  146. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\

  147.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst  , src  , src2  , dstStride, src2Stride);\

  148.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst+8, src+8, src2+8, dstStride, src2Stride);\

  149.     src += 8*dstStride;\

  150.     dst += 8*dstStride;\

  151.     src2 += 8*src2Stride;\

  152.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst  , src  , src2  , dstStride, src2Stride);\

  153.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst+8, src+8, src2+8, dstStride, src2Stride);\

  154. }\

  155. \

  156. static av_always_inline void ff_ ## OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride, int size){\

  157.     ff_put_h264_qpel8or16_hv1_lowpass_ ## MMX(tmp, src, tmpStride, srcStride, size);\

  158.     ff_ ## OPNAME ## h264_qpel8or16_hv2_lowpass_ ## MMX(dst, tmp, dstStride, tmpStride, size);\

  159. }\

  160. static av_always_inline void ff_ ## OPNAME ## h264_qpel8_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  161.     ff_ ## OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(dst  , tmp  , src  , dstStride, tmpStride, srcStride, 8);\

  162. }\

  163. \

  164. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  165.     ff_ ## OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(dst  , tmp  , src  , dstStride, tmpStride, srcStride, 16);\

  166. }\

  167. \

  168. static av_always_inline void ff_ ## OPNAME ## pixels16_l2_shift5_ ## MMX(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h)\

  169. {\

  170.     ff_ ## OPNAME ## pixels8_l2_shift5_ ## MMX(dst  , src16  , src8  , dstStride, src8Stride, h);\

  171.     ff_ ## OPNAME ## pixels8_l2_shift5_ ## MMX(dst+8, src16+8, src8+8, dstStride, src8Stride, h);\

  172. }\

  173. 

  174. 

  175. #if ARCH_X86_64

  176. #define QPEL_H264_H16_XMM(OPNAME, OP, MMX)\

  177. 

  178. void ff_avg_h264_qpel16_h_lowpass_l2_ssse3(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride);

  179. void ff_put_h264_qpel16_h_lowpass_l2_ssse3(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride);

  180. 

  181. #else // ARCH_X86_64

  182. #define QPEL_H264_H16_XMM(OPNAME, OP, MMX)\

  183. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\

  184.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst  , src  , src2  , dstStride, src2Stride);\

  185.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst+8, src+8, src2+8, dstStride, src2Stride);\

  186.     src += 8*dstStride;\

  187.     dst += 8*dstStride;\

  188.     src2 += 8*src2Stride;\

  189.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst  , src  , src2  , dstStride, src2Stride);\

  190.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst+8, src+8, src2+8, dstStride, src2Stride);\

  191. }

  192. #endif // ARCH_X86_64

  193. 

  194. #define QPEL_H264_H_XMM(OPNAME, OP, MMX)\

  195. QPEL_H264_H16_XMM(OPNAME, OP, MMX)\

  196. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  197.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst  , src  , dstStride, srcStride);\

  198.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\

  199.     src += 8*srcStride;\

  200.     dst += 8*dstStride;\

  201.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst  , src  , dstStride, srcStride);\

  202.     ff_ ## OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\

  203. }\

  204. 

  205. #define QPEL_H264_V_XMM(OPNAME, OP, MMX)\

  206. static av_always_inline void ff_ ## OPNAME ## h264_qpel8_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  207.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst  , src  , dstStride, srcStride, 8);\

  208. }\

  209. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\

  210.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst  , src  , dstStride, srcStride, 16);\

  211.     ff_ ## OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\

  212. }

  213. 

  214. static av_always_inline void ff_put_h264_qpel8or16_hv1_lowpass_sse2(int16_t *tmp, uint8_t *src, int tmpStride, int srcStride, int size){

  215.     int w = (size+8)>>3;

  216.     src -= 2*srcStride+2;

  217.     while(w--){

  218.         ff_put_h264_qpel8or16_hv1_lowpass_op_sse2(src, tmp, srcStride, size);

  219.         tmp += 8;

  220.         src += 8;

  221.     }

  222. }

  223. 

  224. #define QPEL_H264_HV_XMM(OPNAME, OP, MMX)\

  225. static av_always_inline void ff_ ## OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride, int size){\

  226.     ff_put_h264_qpel8or16_hv1_lowpass_sse2(tmp, src, tmpStride, srcStride, size);\

  227.     ff_ ## OPNAME ## h264_qpel8or16_hv2_lowpass_ ## MMX(dst, tmp, dstStride, tmpStride, size);\

  228. }\

  229. static av_always_inline void ff_ ## OPNAME ## h264_qpel8_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  230.     ff_ ## OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(dst, tmp, src, dstStride, tmpStride, srcStride, 8);\

  231. }\

  232. static av_always_inline void ff_ ## OPNAME ## h264_qpel16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\

  233.     ff_ ## OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(dst, tmp, src, dstStride, tmpStride, srcStride, 16);\

  234. }\

  235. 

  236. #define ff_put_h264_qpel8_h_lowpass_l2_sse2  ff_put_h264_qpel8_h_lowpass_l2_mmxext

  237. #define ff_avg_h264_qpel8_h_lowpass_l2_sse2  ff_avg_h264_qpel8_h_lowpass_l2_mmxext

  238. #define ff_put_h264_qpel16_h_lowpass_l2_sse2 ff_put_h264_qpel16_h_lowpass_l2_mmxext

  239. #define ff_avg_h264_qpel16_h_lowpass_l2_sse2 ff_avg_h264_qpel16_h_lowpass_l2_mmxext

  240. 

  241. #define ff_put_h264_qpel8_v_lowpass_ssse3  ff_put_h264_qpel8_v_lowpass_sse2

  242. #define ff_avg_h264_qpel8_v_lowpass_ssse3  ff_avg_h264_qpel8_v_lowpass_sse2

  243. #define ff_put_h264_qpel16_v_lowpass_ssse3 ff_put_h264_qpel16_v_lowpass_sse2

  244. #define ff_avg_h264_qpel16_v_lowpass_ssse3 ff_avg_h264_qpel16_v_lowpass_sse2

  245. 

  246. #define ff_put_h264_qpel8or16_hv2_lowpass_sse2 ff_put_h264_qpel8or16_hv2_lowpass_mmxext

  247. #define ff_avg_h264_qpel8or16_hv2_lowpass_sse2 ff_avg_h264_qpel8or16_hv2_lowpass_mmxext

  248. 

  249. #define H264_MC(OPNAME, SIZE, MMX, ALIGN) \

  250. H264_MC_C(OPNAME, SIZE, MMX, ALIGN)\

  251. H264_MC_V(OPNAME, SIZE, MMX, ALIGN)\

  252. H264_MC_H(OPNAME, SIZE, MMX, ALIGN)\

  253. H264_MC_HV(OPNAME, SIZE, MMX, ALIGN)\

  254. 

  255. static void put_h264_qpel16_mc00_sse2 (uint8_t *dst, uint8_t *src, int stride){

  256.     ff_put_pixels16_sse2(dst, src, stride, 16);

  257. }

  258. static void avg_h264_qpel16_mc00_sse2 (uint8_t *dst, uint8_t *src, int stride){

  259.     ff_avg_pixels16_sse2(dst, src, stride, 16);

  260. }

  261. #define put_h264_qpel8_mc00_sse2 put_h264_qpel8_mc00_mmxext

  262. #define avg_h264_qpel8_mc00_sse2 avg_h264_qpel8_mc00_mmxext

  263. 

  264. #define H264_MC_C(OPNAME, SIZE, MMX, ALIGN) \

  265. static void OPNAME ## h264_qpel ## SIZE ## _mc00_ ## MMX (uint8_t *dst, uint8_t *src, int stride){\

  266.     ff_ ## OPNAME ## pixels ## SIZE ## _ ## MMX(dst, src, stride, SIZE);\

  267. }\

  268. 

  269. #define H264_MC_H(OPNAME, SIZE, MMX, ALIGN) \

  270. static void OPNAME ## h264_qpel ## SIZE ## _mc10_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  271.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, src, stride, stride);\

  272. }\

  273. \

  274. static void OPNAME ## h264_qpel ## SIZE ## _mc20_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  275.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_ ## MMX(dst, src, stride, stride);\

  276. }\

  277. \

  278. static void OPNAME ## h264_qpel ## SIZE ## _mc30_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  279.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, src+1, stride, stride);\

  280. }\

  281. 

  282. #define H264_MC_V(OPNAME, SIZE, MMX, ALIGN) \

  283. static void OPNAME ## h264_qpel ## SIZE ## _mc01_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  284.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*SIZE];\

  285.     ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\

  286.     ff_ ## OPNAME ## pixels ## SIZE ## _l2_ ## MMX(dst, src, temp, stride, stride, SIZE);\

  287. }\

  288. \

  289. static void OPNAME ## h264_qpel ## SIZE ## _mc02_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  290.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _v_lowpass_ ## MMX(dst, src, stride, stride);\

  291. }\

  292. \

  293. static void OPNAME ## h264_qpel ## SIZE ## _mc03_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  294.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*SIZE];\

  295.     ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\

  296.     ff_ ## OPNAME ## pixels ## SIZE ## _l2_ ## MMX(dst, src+stride, temp, stride, stride, SIZE);\

  297. }\

  298. 

  299. #define H264_MC_HV(OPNAME, SIZE, MMX, ALIGN) \

  300. static void OPNAME ## h264_qpel ## SIZE ## _mc11_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  301.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*SIZE];\

  302.     ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\

  303.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, temp, stride, SIZE);\

  304. }\

  305. \

  306. static void OPNAME ## h264_qpel ## SIZE ## _mc31_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  307.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*SIZE];\

  308.     ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src+1, SIZE, stride);\

  309.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, temp, stride, SIZE);\

  310. }\

  311. \

  312. static void OPNAME ## h264_qpel ## SIZE ## _mc13_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  313.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*SIZE];\

  314.     ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src, SIZE, stride);\

  315.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src+stride, temp, stride, SIZE);\

  316. }\

  317. \

  318. static void OPNAME ## h264_qpel ## SIZE ## _mc33_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  319.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*SIZE];\

  320.     ff_put_h264_qpel ## SIZE ## _v_lowpass_ ## MMX(temp, src+1, SIZE, stride);\

  321.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src+stride, temp, stride, SIZE);\

  322. }\

  323. \

  324. static void OPNAME ## h264_qpel ## SIZE ## _mc22_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  325.     DECLARE_ALIGNED(ALIGN, uint16_t, temp)[SIZE*(SIZE<8?12:24)];\

  326.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(dst, temp, src, stride, SIZE, stride);\

  327. }\

  328. \

  329. static void OPNAME ## h264_qpel ## SIZE ## _mc21_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  330.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE];\

  331.     uint8_t * const halfHV= temp;\

  332.     int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\

  333.     assert(((int)temp & 7) == 0);\

  334.     ff_put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, SIZE, stride);\

  335.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, halfHV, stride, SIZE);\

  336. }\

  337. \

  338. static void OPNAME ## h264_qpel ## SIZE ## _mc23_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  339.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE];\

  340.     uint8_t * const halfHV= temp;\

  341.     int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\

  342.     assert(((int)temp & 7) == 0);\

  343.     ff_put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, SIZE, stride);\

  344.     ff_ ## OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src+stride, halfHV, stride, SIZE);\

  345. }\

  346. \

  347. static void OPNAME ## h264_qpel ## SIZE ## _mc12_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  348.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE];\

  349.     uint8_t * const halfHV= temp;\

  350.     int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\

  351.     assert(((int)temp & 7) == 0);\

  352.     ff_put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, SIZE, stride);\

  353.     ff_ ## OPNAME ## pixels ## SIZE ## _l2_shift5_mmxext(dst, halfV+2, halfHV, stride, SIZE, SIZE);\

  354. }\

  355. \

  356. static void OPNAME ## h264_qpel ## SIZE ## _mc32_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  357.     DECLARE_ALIGNED(ALIGN, uint8_t, temp)[SIZE*(SIZE<8?12:24)*2 + SIZE*SIZE];\

  358.     uint8_t * const halfHV= temp;\

  359.     int16_t * const halfV= (int16_t*)(temp + SIZE*SIZE);\

  360.     assert(((int)temp & 7) == 0);\

  361.     ff_put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, SIZE, stride);\

  362.     ff_ ## OPNAME ## pixels ## SIZE ## _l2_shift5_mmxext(dst, halfV+3, halfHV, stride, SIZE, SIZE);\

  363. }\

  364. 

  365. #define H264_MC_4816(MMX)\

  366. H264_MC(put_, 4, MMX, 8)\

  367. H264_MC(put_, 8, MMX, 8)\

  368. H264_MC(put_, 16,MMX, 8)\

  369. H264_MC(avg_, 4, MMX, 8)\

  370. H264_MC(avg_, 8, MMX, 8)\

  371. H264_MC(avg_, 16,MMX, 8)\

  372. 

  373. #define H264_MC_816(QPEL, XMM)\

  374. QPEL(put_, 8, XMM, 16)\

  375. QPEL(put_, 16,XMM, 16)\

  376. QPEL(avg_, 8, XMM, 16)\

  377. QPEL(avg_, 16,XMM, 16)\

  378. 

  379. #undef PAVGB

  380. #define PAVGB "pavgb"

  381. QPEL_H264(put_,        PUT_OP, mmxext)

  382. QPEL_H264(avg_, AVG_MMXEXT_OP, mmxext)

  383. QPEL_H264_V_XMM(put_,       PUT_OP, sse2)

  384. QPEL_H264_V_XMM(avg_,AVG_MMXEXT_OP, sse2)

  385. QPEL_H264_HV_XMM(put_,       PUT_OP, sse2)

  386. QPEL_H264_HV_XMM(avg_,AVG_MMXEXT_OP, sse2)

  387. QPEL_H264_H_XMM(put_,       PUT_OP, ssse3)

  388. QPEL_H264_H_XMM(avg_,AVG_MMXEXT_OP, ssse3)

  389. QPEL_H264_HV_XMM(put_,       PUT_OP, ssse3)

  390. QPEL_H264_HV_XMM(avg_,AVG_MMXEXT_OP, ssse3)

  391. #undef PAVGB

  392. 

  393. H264_MC_4816(mmxext)

  394. H264_MC_816(H264_MC_V, sse2)

  395. H264_MC_816(H264_MC_HV, sse2)

  396. H264_MC_816(H264_MC_H, ssse3)

  397. H264_MC_816(H264_MC_HV, ssse3)

  398. 

  399. 

  400. //10bit

  401. #define LUMA_MC_OP(OP, NUM, DEPTH, TYPE, OPT) \

  402. void ff_ ## OP ## _h264_qpel ## NUM ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT \

  403.     (uint8_t *dst, uint8_t *src, int stride);

  404. 

  405. #define LUMA_MC_ALL(DEPTH, TYPE, OPT) \

  406.     LUMA_MC_OP(put,  4, DEPTH, TYPE, OPT) \

  407.     LUMA_MC_OP(avg,  4, DEPTH, TYPE, OPT) \

  408.     LUMA_MC_OP(put,  8, DEPTH, TYPE, OPT) \

  409.     LUMA_MC_OP(avg,  8, DEPTH, TYPE, OPT) \

  410.     LUMA_MC_OP(put, 16, DEPTH, TYPE, OPT) \

  411.     LUMA_MC_OP(avg, 16, DEPTH, TYPE, OPT)

  412. 

  413. #define LUMA_MC_816(DEPTH, TYPE, OPT) \

  414.     LUMA_MC_OP(put,  8, DEPTH, TYPE, OPT) \

  415.     LUMA_MC_OP(avg,  8, DEPTH, TYPE, OPT) \

  416.     LUMA_MC_OP(put, 16, DEPTH, TYPE, OPT) \

  417.     LUMA_MC_OP(avg, 16, DEPTH, TYPE, OPT)

  418. 

  419. LUMA_MC_ALL(10, mc00, mmxext)

  420. LUMA_MC_ALL(10, mc10, mmxext)

  421. LUMA_MC_ALL(10, mc20, mmxext)

  422. LUMA_MC_ALL(10, mc30, mmxext)

  423. LUMA_MC_ALL(10, mc01, mmxext)

  424. LUMA_MC_ALL(10, mc11, mmxext)

  425. LUMA_MC_ALL(10, mc21, mmxext)

  426. LUMA_MC_ALL(10, mc31, mmxext)

  427. LUMA_MC_ALL(10, mc02, mmxext)

  428. LUMA_MC_ALL(10, mc12, mmxext)

  429. LUMA_MC_ALL(10, mc22, mmxext)

  430. LUMA_MC_ALL(10, mc32, mmxext)

  431. LUMA_MC_ALL(10, mc03, mmxext)

  432. LUMA_MC_ALL(10, mc13, mmxext)

  433. LUMA_MC_ALL(10, mc23, mmxext)

  434. LUMA_MC_ALL(10, mc33, mmxext)

  435. 

  436. LUMA_MC_816(10, mc00, sse2)

  437. LUMA_MC_816(10, mc10, sse2)

  438. LUMA_MC_816(10, mc10, sse2_cache64)

  439. LUMA_MC_816(10, mc10, ssse3_cache64)

  440. LUMA_MC_816(10, mc20, sse2)

  441. LUMA_MC_816(10, mc20, sse2_cache64)

  442. LUMA_MC_816(10, mc20, ssse3_cache64)

  443. LUMA_MC_816(10, mc30, sse2)

  444. LUMA_MC_816(10, mc30, sse2_cache64)

  445. LUMA_MC_816(10, mc30, ssse3_cache64)

  446. LUMA_MC_816(10, mc01, sse2)

  447. LUMA_MC_816(10, mc11, sse2)

  448. LUMA_MC_816(10, mc21, sse2)

  449. LUMA_MC_816(10, mc31, sse2)

  450. LUMA_MC_816(10, mc02, sse2)

  451. LUMA_MC_816(10, mc12, sse2)

  452. LUMA_MC_816(10, mc22, sse2)

  453. LUMA_MC_816(10, mc32, sse2)

  454. LUMA_MC_816(10, mc03, sse2)

  455. LUMA_MC_816(10, mc13, sse2)

  456. LUMA_MC_816(10, mc23, sse2)

  457. LUMA_MC_816(10, mc33, sse2)

  458. 

  459. #define QPEL16_OPMC(OP, MC, MMX)\

  460. void ff_ ## OP ## _h264_qpel16_ ## MC ## _10_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\

  461.     ff_ ## OP ## _h264_qpel8_ ## MC ## _10_ ## MMX(dst   , src   , stride);\

  462.     ff_ ## OP ## _h264_qpel8_ ## MC ## _10_ ## MMX(dst+16, src+16, stride);\

  463.     src += 8*stride;\

  464.     dst += 8*stride;\

  465.     ff_ ## OP ## _h264_qpel8_ ## MC ## _10_ ## MMX(dst   , src   , stride);\

  466.     ff_ ## OP ## _h264_qpel8_ ## MC ## _10_ ## MMX(dst+16, src+16, stride);\

  467. }

  468. 

  469. #define QPEL16_OP(MC, MMX)\

  470. QPEL16_OPMC(put, MC, MMX)\

  471. QPEL16_OPMC(avg, MC, MMX)

  472. 

  473. #define QPEL16(MMX)\

  474. QPEL16_OP(mc00, MMX)\

  475. QPEL16_OP(mc01, MMX)\

  476. QPEL16_OP(mc02, MMX)\

  477. QPEL16_OP(mc03, MMX)\

  478. QPEL16_OP(mc10, MMX)\

  479. QPEL16_OP(mc11, MMX)\

  480. QPEL16_OP(mc12, MMX)\

  481. QPEL16_OP(mc13, MMX)\

  482. QPEL16_OP(mc20, MMX)\

  483. QPEL16_OP(mc21, MMX)\

  484. QPEL16_OP(mc22, MMX)\

  485. QPEL16_OP(mc23, MMX)\

  486. QPEL16_OP(mc30, MMX)\

  487. QPEL16_OP(mc31, MMX)\

  488. QPEL16_OP(mc32, MMX)\

  489. QPEL16_OP(mc33, MMX)

  490. 

  491. #if ARCH_X86_32 && HAVE_YASM && CONFIG_H264QPEL // ARCH_X86_64 implies SSE2+

  492. QPEL16(mmxext)

  493. #endif

  494. 

  495. #endif /* HAVE_YASM */

  496. 

  497. #define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX)                          \

  498.     do {                                                                     \

  499.     c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \

  500.     c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \

  501.     c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \

  502.     c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \

  503.     c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \

  504.     c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \

  505.     c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \

  506.     c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \

  507.     c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \

  508.     c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \

  509.     c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \

  510.     c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \

  511.     c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \

  512.     c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \

  513.     c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \

  514.     c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \

  515.     } while (0)

  516. 

  517. #define H264_QPEL_FUNCS(x, y, CPU)                                                            \

  518.     do {                                                                                      \

  519.         c->put_h264_qpel_pixels_tab[0][x + y * 4] = put_h264_qpel16_mc ## x ## y ## _ ## CPU; \

  520.         c->put_h264_qpel_pixels_tab[1][x + y * 4] = put_h264_qpel8_mc  ## x ## y ## _ ## CPU; \

  521.         c->avg_h264_qpel_pixels_tab[0][x + y * 4] = avg_h264_qpel16_mc ## x ## y ## _ ## CPU; \

  522.         c->avg_h264_qpel_pixels_tab[1][x + y * 4] = avg_h264_qpel8_mc  ## x ## y ## _ ## CPU; \

  523.     } while (0)

  524. 

  525. #define H264_QPEL_FUNCS_10(x, y, CPU)                                                               \

  526.     do {                                                                                            \

  527.         c->put_h264_qpel_pixels_tab[0][x + y * 4] = ff_put_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \

  528.         c->put_h264_qpel_pixels_tab[1][x + y * 4] = ff_put_h264_qpel8_mc  ## x ## y ## _10_ ## CPU; \

  529.         c->avg_h264_qpel_pixels_tab[0][x + y * 4] = ff_avg_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \

  530.         c->avg_h264_qpel_pixels_tab[1][x + y * 4] = ff_avg_h264_qpel8_mc  ## x ## y ## _10_ ## CPU; \

  531.     } while (0)

  532. 

  533. av_cold void ff_h264qpel_init_x86(H264QpelContext *c, int bit_depth)

  534. {

  535. #if HAVE_YASM

  536.     int high_bit_depth = bit_depth > 8;

  537.     int mm_flags = av_get_cpu_flags();

  538. 

  539.     if (EXTERNAL_MMXEXT(mm_flags)) {

  540.         if (!high_bit_depth) {

  541.             SET_QPEL_FUNCS(put_h264_qpel, 0, 16, mmxext, );

  542.             SET_QPEL_FUNCS(put_h264_qpel, 1,  8, mmxext, );

  543.             SET_QPEL_FUNCS(put_h264_qpel, 2,  4, mmxext, );

  544.             SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, mmxext, );

  545.             SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, mmxext, );

  546.             SET_QPEL_FUNCS(avg_h264_qpel, 2,  4, mmxext, );

  547.         } else if (bit_depth == 10) {

  548. #if ARCH_X86_32

  549.             SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_mmxext, ff_);

  550.             SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_mmxext, ff_);

  551.             SET_QPEL_FUNCS(put_h264_qpel, 1,  8, 10_mmxext, ff_);

  552.             SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, 10_mmxext, ff_);

  553. #endif

  554.             SET_QPEL_FUNCS(put_h264_qpel, 2, 4,  10_mmxext, ff_);

  555.             SET_QPEL_FUNCS(avg_h264_qpel, 2, 4,  10_mmxext, ff_);

  556.         }

  557.     }

  558. 

  559.     if (EXTERNAL_SSE2(mm_flags)) {

  560.         if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW) && !high_bit_depth) {

  561.             // these functions are slower than mmx on AMD, but faster on Intel

  562.             H264_QPEL_FUNCS(0, 0, sse2);

  563.         }

  564. 

  565.         if (!high_bit_depth) {

  566.             H264_QPEL_FUNCS(0, 1, sse2);

  567.             H264_QPEL_FUNCS(0, 2, sse2);

  568.             H264_QPEL_FUNCS(0, 3, sse2);

  569.             H264_QPEL_FUNCS(1, 1, sse2);

  570.             H264_QPEL_FUNCS(1, 2, sse2);

  571.             H264_QPEL_FUNCS(1, 3, sse2);

  572.             H264_QPEL_FUNCS(2, 1, sse2);

  573.             H264_QPEL_FUNCS(2, 2, sse2);

  574.             H264_QPEL_FUNCS(2, 3, sse2);

  575.             H264_QPEL_FUNCS(3, 1, sse2);

  576.             H264_QPEL_FUNCS(3, 2, sse2);

  577.             H264_QPEL_FUNCS(3, 3, sse2);

  578.         }

  579. 

  580.         if (bit_depth == 10) {

  581.             SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_sse2, ff_);

  582.             SET_QPEL_FUNCS(put_h264_qpel, 1,  8, 10_sse2, ff_);

  583.             SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_sse2, ff_);

  584.             SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, 10_sse2, ff_);

  585.             H264_QPEL_FUNCS_10(1, 0, sse2_cache64);

  586.             H264_QPEL_FUNCS_10(2, 0, sse2_cache64);

  587.             H264_QPEL_FUNCS_10(3, 0, sse2_cache64);

  588.         }

  589.     }

  590. 

  591.     if (EXTERNAL_SSSE3(mm_flags)) {

  592.         if (!high_bit_depth) {

  593.             H264_QPEL_FUNCS(1, 0, ssse3);

  594.             H264_QPEL_FUNCS(1, 1, ssse3);

  595.             H264_QPEL_FUNCS(1, 2, ssse3);

  596.             H264_QPEL_FUNCS(1, 3, ssse3);

  597.             H264_QPEL_FUNCS(2, 0, ssse3);

  598.             H264_QPEL_FUNCS(2, 1, ssse3);

  599.             H264_QPEL_FUNCS(2, 2, ssse3);

  600.             H264_QPEL_FUNCS(2, 3, ssse3);

  601.             H264_QPEL_FUNCS(3, 0, ssse3);

  602.             H264_QPEL_FUNCS(3, 1, ssse3);

  603.             H264_QPEL_FUNCS(3, 2, ssse3);

  604.             H264_QPEL_FUNCS(3, 3, ssse3);

  605.         }

  606. 

  607.         if (bit_depth == 10) {

  608.             H264_QPEL_FUNCS_10(1, 0, ssse3_cache64);

  609.             H264_QPEL_FUNCS_10(2, 0, ssse3_cache64);

  610.             H264_QPEL_FUNCS_10(3, 0, ssse3_cache64);

  611.         }

  612.     }

  613. 

  614.     if (EXTERNAL_AVX(mm_flags)) {

  615.         /* AVX implies 64 byte cache lines without the need to avoid unaligned

  616.          * memory accesses that cross the boundary between two cache lines.

  617.          * TODO: Port X264_CPU_CACHELINE_32/64 detection from x264 to avoid

  618.          * having to treat SSE2 functions with such properties as AVX. */

  619.         if (bit_depth == 10) {

  620.             H264_QPEL_FUNCS_10(1, 0, sse2);

  621.             H264_QPEL_FUNCS_10(2, 0, sse2);

  622.             H264_QPEL_FUNCS_10(3, 0, sse2);

  623.         }

  624.     }

  625. #endif

  626. }