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.
83678 Commits
32 Branches
303MB
Tree: 88307b3eec
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '88307b3eec'
${ noResults }
FFmpeg/libavcodec/x86/h264dsp_init.c

426 lines
19KB
Raw Blame History 

  1. /*

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

  3.  *

  4.  * This file is part of FFmpeg.

  5.  *

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

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

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

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

  10.  *

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

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

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

  14.  * Lesser General Public License for more details.

  15.  *

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

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

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

  19.  */

  20. 

  21. #include "libavutil/attributes.h"

  22. #include "libavutil/cpu.h"

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

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

  25. #include "libavcodec/h264dsp.h"

  26. 

  27. /***********************************/

  28. /* IDCT */

  29. #define IDCT_ADD_FUNC(NUM, DEPTH, OPT)                                  \

  30. void ff_h264_idct ## NUM ## _add_ ## DEPTH ## _ ## OPT(uint8_t *dst,    \

  31.                                                        int16_t *block,  \

  32.                                                        int stride);

  33. 

  34. IDCT_ADD_FUNC(, 8, mmx)

  35. IDCT_ADD_FUNC(, 10, sse2)

  36. IDCT_ADD_FUNC(_dc, 8, mmxext)

  37. IDCT_ADD_FUNC(_dc, 10, mmxext)

  38. IDCT_ADD_FUNC(8_dc, 8, mmxext)

  39. IDCT_ADD_FUNC(8_dc, 10, sse2)

  40. IDCT_ADD_FUNC(8, 8, mmx)

  41. IDCT_ADD_FUNC(8, 8, sse2)

  42. IDCT_ADD_FUNC(8, 10, sse2)

  43. IDCT_ADD_FUNC(, 10, avx)

  44. IDCT_ADD_FUNC(8_dc, 10, avx)

  45. IDCT_ADD_FUNC(8, 10, avx)

  46. 

  47. 

  48. #define IDCT_ADD_REP_FUNC(NUM, REP, DEPTH, OPT)                         \

  49. void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT       \

  50.     (uint8_t *dst, const int *block_offset,                             \

  51.      int16_t *block, int stride, const uint8_t nnzc[6 * 8]);

  52. 

  53. IDCT_ADD_REP_FUNC(8, 4, 8, mmx)

  54. IDCT_ADD_REP_FUNC(8, 4, 8, mmxext)

  55. IDCT_ADD_REP_FUNC(8, 4, 8, sse2)

  56. IDCT_ADD_REP_FUNC(8, 4, 10, sse2)

  57. IDCT_ADD_REP_FUNC(8, 4, 10, avx)

  58. IDCT_ADD_REP_FUNC(, 16, 8, mmx)

  59. IDCT_ADD_REP_FUNC(, 16, 8, mmxext)

  60. IDCT_ADD_REP_FUNC(, 16, 8, sse2)

  61. IDCT_ADD_REP_FUNC(, 16, 10, sse2)

  62. IDCT_ADD_REP_FUNC(, 16intra, 8, mmx)

  63. IDCT_ADD_REP_FUNC(, 16intra, 8, mmxext)

  64. IDCT_ADD_REP_FUNC(, 16intra, 8, sse2)

  65. IDCT_ADD_REP_FUNC(, 16intra, 10, sse2)

  66. IDCT_ADD_REP_FUNC(, 16, 10, avx)

  67. IDCT_ADD_REP_FUNC(, 16intra, 10, avx)

  68. 

  69. 

  70. #define IDCT_ADD_REP_FUNC2(NUM, REP, DEPTH, OPT)                      \

  71. void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT     \

  72.     (uint8_t **dst, const int *block_offset,                          \

  73.      int16_t *block, int stride, const uint8_t nnzc[6 * 8]);

  74. 

  75. IDCT_ADD_REP_FUNC2(, 8, 8, mmx)

  76. IDCT_ADD_REP_FUNC2(, 8, 8, mmxext)

  77. IDCT_ADD_REP_FUNC2(, 8, 8, sse2)

  78. IDCT_ADD_REP_FUNC2(, 8, 10, sse2)

  79. IDCT_ADD_REP_FUNC2(, 8, 10, avx)

  80. 

  81. IDCT_ADD_REP_FUNC2(, 8_422, 8, mmx)

  82. 

  83. IDCT_ADD_REP_FUNC2(, 8_422, 10, sse2)

  84. IDCT_ADD_REP_FUNC2(, 8_422, 10, avx)

  85. 

  86. void ff_h264_luma_dc_dequant_idct_mmx(int16_t *output, int16_t *input, int qmul);

  87. void ff_h264_luma_dc_dequant_idct_sse2(int16_t *output, int16_t *input, int qmul);

  88. 

  89. /***********************************/

  90. /* deblocking */

  91. 

  92. void ff_h264_loop_filter_strength_mmxext(int16_t bS[2][4][4], uint8_t nnz[40],

  93.                                          int8_t ref[2][40],

  94.                                          int16_t mv[2][40][2],

  95.                                          int bidir, int edges, int step,

  96.                                          int mask_mv0, int mask_mv1, int field);

  97. 

  98. #define LF_FUNC(DIR, TYPE, DEPTH, OPT)                                        \

  99. void ff_deblock_ ## DIR ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT(uint8_t *pix,  \

  100.                                                                int stride,    \

  101.                                                                int alpha,     \

  102.                                                                int beta,      \

  103.                                                                int8_t *tc0);

  104. #define LF_IFUNC(DIR, TYPE, DEPTH, OPT) \

  105. void ff_deblock_ ## DIR ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT(uint8_t *pix,  \

  106.                                                                int stride,    \

  107.                                                                int alpha,     \

  108.                                                                int beta);

  109. 

  110. #define LF_FUNCS(type, depth)                   \

  111. LF_FUNC(h,  chroma,          depth, mmxext)     \

  112. LF_IFUNC(h, chroma_intra,    depth, mmxext)     \

  113. LF_FUNC(h,  chroma422,       depth, mmxext)     \

  114. LF_IFUNC(h, chroma422_intra, depth, mmxext)     \

  115. LF_FUNC(v,  chroma,          depth, mmxext)     \

  116. LF_IFUNC(v, chroma_intra,    depth, mmxext)     \

  117. LF_FUNC(h,  luma,            depth, mmxext)     \

  118. LF_IFUNC(h, luma_intra,      depth, mmxext)     \

  119. LF_FUNC(h,  luma,            depth, sse2)       \

  120. LF_IFUNC(h, luma_intra,      depth, sse2)       \

  121. LF_FUNC(v,  luma,            depth, sse2)       \

  122. LF_IFUNC(v, luma_intra,      depth, sse2)       \

  123. LF_FUNC(h,  chroma,          depth, sse2)       \

  124. LF_IFUNC(h, chroma_intra,    depth, sse2)       \

  125. LF_FUNC(h,  chroma422,       depth, sse2)       \

  126. LF_IFUNC(h, chroma422_intra, depth, sse2)       \

  127. LF_FUNC(v,  chroma,          depth, sse2)       \

  128. LF_IFUNC(v, chroma_intra,    depth, sse2)       \

  129. LF_FUNC(h,  luma,            depth, avx)        \

  130. LF_IFUNC(h, luma_intra,      depth, avx)        \

  131. LF_FUNC(v,  luma,            depth, avx)        \

  132. LF_IFUNC(v, luma_intra,      depth, avx)        \

  133. LF_FUNC(h,  chroma,          depth, avx)        \

  134. LF_IFUNC(h, chroma_intra,    depth, avx)        \

  135. LF_FUNC(h,  chroma422,       depth, avx)        \

  136. LF_IFUNC(h, chroma422_intra, depth, avx)        \

  137. LF_FUNC(v,  chroma,          depth, avx)        \

  138. LF_IFUNC(v, chroma_intra,    depth, avx)

  139. 

  140. LF_FUNC(h, luma_mbaff, 8, sse2)

  141. LF_FUNC(h, luma_mbaff, 8, avx)

  142. 

  143. LF_FUNCS(uint8_t,   8)

  144. LF_FUNCS(uint16_t, 10)

  145. 

  146. #if ARCH_X86_32 && HAVE_MMXEXT_EXTERNAL

  147. LF_FUNC(v8, luma, 8, mmxext)

  148. static void deblock_v_luma_8_mmxext(uint8_t *pix, int stride, int alpha,

  149.                                     int beta, int8_t *tc0)

  150. {

  151.     if ((tc0[0] & tc0[1]) >= 0)

  152.         ff_deblock_v8_luma_8_mmxext(pix + 0, stride, alpha, beta, tc0);

  153.     if ((tc0[2] & tc0[3]) >= 0)

  154.         ff_deblock_v8_luma_8_mmxext(pix + 8, stride, alpha, beta, tc0 + 2);

  155. }

  156. LF_IFUNC(v8, luma_intra, 8, mmxext)

  157. static void deblock_v_luma_intra_8_mmxext(uint8_t *pix, int stride,

  158.                                           int alpha, int beta)

  159. {

  160.     ff_deblock_v8_luma_intra_8_mmxext(pix + 0, stride, alpha, beta);

  161.     ff_deblock_v8_luma_intra_8_mmxext(pix + 8, stride, alpha, beta);

  162. }

  163. #endif /* ARCH_X86_32 && HAVE_MMXEXT_EXTERNAL */

  164. 

  165. LF_FUNC(v,  luma,       10, mmxext)

  166. LF_IFUNC(v, luma_intra, 10, mmxext)

  167. 

  168. /***********************************/

  169. /* weighted prediction */

  170. 

  171. #define H264_WEIGHT(W, OPT)                                             \

  172. void ff_h264_weight_ ## W ## _ ## OPT(uint8_t *dst, ptrdiff_t stride,   \

  173.                                       int height, int log2_denom,       \

  174.                                       int weight, int offset);

  175. 

  176. #define H264_BIWEIGHT(W, OPT)                                           \

  177. void ff_h264_biweight_ ## W ## _ ## OPT(uint8_t *dst, uint8_t *src,     \

  178.                                         ptrdiff_t stride, int height,   \

  179.                                         int log2_denom, int weightd,    \

  180.                                         int weights, int offset);

  181. 

  182. #define H264_BIWEIGHT_MMX(W)                    \

  183.     H264_WEIGHT(W, mmxext)                      \

  184.     H264_BIWEIGHT(W, mmxext)

  185. 

  186. #define H264_BIWEIGHT_MMX_SSE(W)                \

  187.     H264_BIWEIGHT_MMX(W)                        \

  188.     H264_WEIGHT(W, sse2)                        \

  189.     H264_BIWEIGHT(W, sse2)                      \

  190.     H264_BIWEIGHT(W, ssse3)

  191. 

  192. H264_BIWEIGHT_MMX_SSE(16)

  193. H264_BIWEIGHT_MMX_SSE(8)

  194. H264_BIWEIGHT_MMX(4)

  195. 

  196. #define H264_WEIGHT_10(W, DEPTH, OPT)                                   \

  197. void ff_h264_weight_ ## W ## _ ## DEPTH ## _ ## OPT(uint8_t *dst,       \

  198.                                                     ptrdiff_t stride,   \

  199.                                                     int height,         \

  200.                                                     int log2_denom,     \

  201.                                                     int weight,         \

  202.                                                     int offset);

  203. 

  204. #define H264_BIWEIGHT_10(W, DEPTH, OPT)                                 \

  205. void ff_h264_biweight_ ## W ## _ ## DEPTH ## _ ## OPT(uint8_t *dst,     \

  206.                                                       uint8_t *src,     \

  207.                                                       ptrdiff_t stride, \

  208.                                                       int height,       \

  209.                                                       int log2_denom,   \

  210.                                                       int weightd,      \

  211.                                                       int weights,      \

  212.                                                       int offset);

  213. 

  214. #define H264_BIWEIGHT_10_SSE(W, DEPTH)          \

  215.     H264_WEIGHT_10(W, DEPTH, sse2)              \

  216.     H264_WEIGHT_10(W, DEPTH, sse4)              \

  217.     H264_BIWEIGHT_10(W, DEPTH, sse2)            \

  218.     H264_BIWEIGHT_10(W, DEPTH, sse4)

  219. 

  220. H264_BIWEIGHT_10_SSE(16, 10)

  221. H264_BIWEIGHT_10_SSE(8,  10)

  222. H264_BIWEIGHT_10_SSE(4,  10)

  223. 

  224. av_cold void ff_h264dsp_init_x86(H264DSPContext *c, const int bit_depth,

  225.                                  const int chroma_format_idc)

  226. {

  227. #if HAVE_YASM

  228.     int cpu_flags = av_get_cpu_flags();

  229. 

  230.     if (EXTERNAL_MMXEXT(cpu_flags) && chroma_format_idc <= 1)

  231.         c->h264_loop_filter_strength = ff_h264_loop_filter_strength_mmxext;

  232. 

  233.     if (bit_depth == 8) {

  234.         if (EXTERNAL_MMX(cpu_flags)) {

  235.             c->h264_idct_dc_add   =

  236.             c->h264_idct_add      = ff_h264_idct_add_8_mmx;

  237.             c->h264_idct8_dc_add  =

  238.             c->h264_idct8_add     = ff_h264_idct8_add_8_mmx;

  239. 

  240.             c->h264_idct_add16 = ff_h264_idct_add16_8_mmx;

  241.             c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmx;

  242.             if (chroma_format_idc <= 1) {

  243.                 c->h264_idct_add8 = ff_h264_idct_add8_8_mmx;

  244.             } else {

  245.                 c->h264_idct_add8 = ff_h264_idct_add8_422_8_mmx;

  246.             }

  247.             c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmx;

  248.             if (cpu_flags & AV_CPU_FLAG_CMOV)

  249.                 c->h264_luma_dc_dequant_idct = ff_h264_luma_dc_dequant_idct_mmx;

  250.         }

  251.         if (EXTERNAL_MMXEXT(cpu_flags)) {

  252.             c->h264_idct_dc_add  = ff_h264_idct_dc_add_8_mmxext;

  253.             c->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_mmxext;

  254.             c->h264_idct_add16   = ff_h264_idct_add16_8_mmxext;

  255.             c->h264_idct8_add4   = ff_h264_idct8_add4_8_mmxext;

  256.             if (chroma_format_idc <= 1)

  257.                 c->h264_idct_add8 = ff_h264_idct_add8_8_mmxext;

  258.             c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmxext;

  259. 

  260.             c->h264_v_loop_filter_chroma       = ff_deblock_v_chroma_8_mmxext;

  261.             c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_8_mmxext;

  262.             if (chroma_format_idc <= 1) {

  263.                 c->h264_h_loop_filter_chroma       = ff_deblock_h_chroma_8_mmxext;

  264.                 c->h264_h_loop_filter_chroma_intra = ff_deblock_h_chroma_intra_8_mmxext;

  265.             } else {

  266.                 c->h264_h_loop_filter_chroma = ff_deblock_h_chroma422_8_mmxext;

  267.                 c->h264_h_loop_filter_chroma_intra = ff_deblock_h_chroma422_intra_8_mmxext;

  268.             }

  269. #if ARCH_X86_32 && HAVE_MMXEXT_EXTERNAL

  270.             c->h264_v_loop_filter_luma       = deblock_v_luma_8_mmxext;

  271.             c->h264_h_loop_filter_luma       = ff_deblock_h_luma_8_mmxext;

  272.             c->h264_v_loop_filter_luma_intra = deblock_v_luma_intra_8_mmxext;

  273.             c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_mmxext;

  274. #endif /* ARCH_X86_32 && HAVE_MMXEXT_EXTERNAL */

  275.             c->weight_h264_pixels_tab[0] = ff_h264_weight_16_mmxext;

  276.             c->weight_h264_pixels_tab[1] = ff_h264_weight_8_mmxext;

  277.             c->weight_h264_pixels_tab[2] = ff_h264_weight_4_mmxext;

  278. 

  279.             c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_mmxext;

  280.             c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_mmxext;

  281.             c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_mmxext;

  282.         }

  283.         if (EXTERNAL_SSE2(cpu_flags)) {

  284.             c->h264_idct8_add  = ff_h264_idct8_add_8_sse2;

  285. 

  286.             c->h264_idct_add16 = ff_h264_idct_add16_8_sse2;

  287.             c->h264_idct8_add4 = ff_h264_idct8_add4_8_sse2;

  288.             if (chroma_format_idc <= 1)

  289.                 c->h264_idct_add8 = ff_h264_idct_add8_8_sse2;

  290.             c->h264_idct_add16intra      = ff_h264_idct_add16intra_8_sse2;

  291.             c->h264_luma_dc_dequant_idct = ff_h264_luma_dc_dequant_idct_sse2;

  292. 

  293.             c->weight_h264_pixels_tab[0] = ff_h264_weight_16_sse2;

  294.             c->weight_h264_pixels_tab[1] = ff_h264_weight_8_sse2;

  295. 

  296.             c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_sse2;

  297.             c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_sse2;

  298. 

  299.             c->h264_v_loop_filter_luma       = ff_deblock_v_luma_8_sse2;

  300.             c->h264_h_loop_filter_luma       = ff_deblock_h_luma_8_sse2;

  301.             c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_sse2;

  302.             c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_sse2;

  303. 

  304. #if ARCH_X86_64

  305.             c->h264_h_loop_filter_luma_mbaff = ff_deblock_h_luma_mbaff_8_sse2;

  306. #endif

  307.         }

  308.         if (EXTERNAL_SSSE3(cpu_flags)) {

  309.             c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_ssse3;

  310.             c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_ssse3;

  311.         }

  312.         if (EXTERNAL_AVX(cpu_flags)) {

  313.             c->h264_v_loop_filter_luma       = ff_deblock_v_luma_8_avx;

  314.             c->h264_h_loop_filter_luma       = ff_deblock_h_luma_8_avx;

  315.             c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_avx;

  316.             c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_avx;

  317. #if ARCH_X86_64

  318.             c->h264_h_loop_filter_luma_mbaff = ff_deblock_h_luma_mbaff_8_avx;

  319. #endif

  320. 

  321.             c->h264_v_loop_filter_chroma       = ff_deblock_v_chroma_8_avx;

  322.             if (chroma_format_idc <= 1) {

  323.                 c->h264_h_loop_filter_chroma       = ff_deblock_h_chroma_8_avx;

  324.             } else {

  325.                 c->h264_h_loop_filter_chroma       = ff_deblock_h_chroma422_8_avx;

  326.             }

  327.         }

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

  329.         if (EXTERNAL_MMXEXT(cpu_flags)) {

  330. #if ARCH_X86_32

  331.             c->h264_v_loop_filter_chroma       = ff_deblock_v_chroma_10_mmxext;

  332.             c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_mmxext;

  333.             if (chroma_format_idc <= 1) {

  334.                 c->h264_h_loop_filter_chroma = ff_deblock_h_chroma_10_mmxext;

  335.             } else {

  336.                 c->h264_h_loop_filter_chroma = ff_deblock_h_chroma422_10_mmxext;

  337.             }

  338.             c->h264_v_loop_filter_luma         = ff_deblock_v_luma_10_mmxext;

  339.             c->h264_h_loop_filter_luma         = ff_deblock_h_luma_10_mmxext;

  340.             c->h264_v_loop_filter_luma_intra   = ff_deblock_v_luma_intra_10_mmxext;

  341.             c->h264_h_loop_filter_luma_intra   = ff_deblock_h_luma_intra_10_mmxext;

  342. #endif /* ARCH_X86_32 */

  343.             c->h264_idct_dc_add = ff_h264_idct_dc_add_10_mmxext;

  344.         }

  345.         if (EXTERNAL_SSE2(cpu_flags)) {

  346.             c->h264_idct_add     = ff_h264_idct_add_10_sse2;

  347.             c->h264_idct8_dc_add = ff_h264_idct8_dc_add_10_sse2;

  348. 

  349.             c->h264_idct_add16 = ff_h264_idct_add16_10_sse2;

  350.             if (chroma_format_idc <= 1) {

  351.                 c->h264_idct_add8 = ff_h264_idct_add8_10_sse2;

  352.             } else {

  353.                 c->h264_idct_add8 = ff_h264_idct_add8_422_10_sse2;

  354.             }

  355.             c->h264_idct_add16intra = ff_h264_idct_add16intra_10_sse2;

  356. #if HAVE_ALIGNED_STACK

  357.             c->h264_idct8_add  = ff_h264_idct8_add_10_sse2;

  358.             c->h264_idct8_add4 = ff_h264_idct8_add4_10_sse2;

  359. #endif /* HAVE_ALIGNED_STACK */

  360. 

  361.             c->weight_h264_pixels_tab[0] = ff_h264_weight_16_10_sse2;

  362.             c->weight_h264_pixels_tab[1] = ff_h264_weight_8_10_sse2;

  363.             c->weight_h264_pixels_tab[2] = ff_h264_weight_4_10_sse2;

  364. 

  365.             c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_10_sse2;

  366.             c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_10_sse2;

  367.             c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_10_sse2;

  368. 

  369.             c->h264_v_loop_filter_chroma       = ff_deblock_v_chroma_10_sse2;

  370.             c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_sse2;

  371.             if (chroma_format_idc <= 1) {

  372.                 c->h264_h_loop_filter_chroma = ff_deblock_h_chroma_10_sse2;

  373.             } else {

  374.                 c->h264_h_loop_filter_chroma = ff_deblock_h_chroma422_10_sse2;

  375.             }

  376. #if HAVE_ALIGNED_STACK

  377.             c->h264_v_loop_filter_luma       = ff_deblock_v_luma_10_sse2;

  378.             c->h264_h_loop_filter_luma       = ff_deblock_h_luma_10_sse2;

  379.             c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_sse2;

  380.             c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_sse2;

  381. #endif /* HAVE_ALIGNED_STACK */

  382.         }

  383.         if (EXTERNAL_SSE4(cpu_flags)) {

  384.             c->weight_h264_pixels_tab[0] = ff_h264_weight_16_10_sse4;

  385.             c->weight_h264_pixels_tab[1] = ff_h264_weight_8_10_sse4;

  386.             c->weight_h264_pixels_tab[2] = ff_h264_weight_4_10_sse4;

  387. 

  388.             c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_10_sse4;

  389.             c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_10_sse4;

  390.             c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_10_sse4;

  391.         }

  392.         if (EXTERNAL_AVX(cpu_flags)) {

  393.             c->h264_idct_dc_add  =

  394.             c->h264_idct_add     = ff_h264_idct_add_10_avx;

  395.             c->h264_idct8_dc_add = ff_h264_idct8_dc_add_10_avx;

  396. 

  397.             c->h264_idct_add16 = ff_h264_idct_add16_10_avx;

  398.             if (chroma_format_idc <= 1) {

  399.                 c->h264_idct_add8 = ff_h264_idct_add8_10_avx;

  400.             } else {

  401.                 c->h264_idct_add8 = ff_h264_idct_add8_422_10_avx;

  402.             }

  403.             c->h264_idct_add16intra = ff_h264_idct_add16intra_10_avx;

  404. #if HAVE_ALIGNED_STACK

  405.             c->h264_idct8_add  = ff_h264_idct8_add_10_avx;

  406.             c->h264_idct8_add4 = ff_h264_idct8_add4_10_avx;

  407. #endif /* HAVE_ALIGNED_STACK */

  408. 

  409.             c->h264_v_loop_filter_chroma       = ff_deblock_v_chroma_10_avx;

  410.             c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_avx;

  411.             if (chroma_format_idc <= 1) {

  412.                 c->h264_h_loop_filter_chroma = ff_deblock_h_chroma_10_avx;

  413.             } else {

  414.                 c->h264_h_loop_filter_chroma = ff_deblock_h_chroma422_10_avx;

  415.             }

  416. #if HAVE_ALIGNED_STACK

  417.             c->h264_v_loop_filter_luma         = ff_deblock_v_luma_10_avx;

  418.             c->h264_h_loop_filter_luma         = ff_deblock_h_luma_10_avx;

  419.             c->h264_v_loop_filter_luma_intra   = ff_deblock_v_luma_intra_10_avx;

  420.             c->h264_h_loop_filter_luma_intra   = ff_deblock_h_luma_intra_10_avx;

  421. #endif /* HAVE_ALIGNED_STACK */

  422.         }

  423.     }

  424. #endif

  425. }