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.
32661 Commits
32 Branches
535MB
Tree: bd66f073fe
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'bd66f073fe'
${ noResults }
FFmpeg/libavcodec/x86/vp8dsp-init.c

426 lines
20KB
Raw Blame History 

  1. /*

  2.  * VP8 DSP functions x86-optimized

  3.  * Copyright (c) 2010 Ronald S. Bultje <rsbultje@gmail.com>

  4.  * Copyright (c) 2010 Jason Garrett-Glaser <darkshikari@gmail.com>

  5.  *

  6.  * This file is part of Libav.

  7.  *

  8.  * Libav is free software; you can redistribute it and/or

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

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

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

  12.  *

  13.  * Libav is distributed in the hope that it will be useful,

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

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

  16.  * Lesser General Public License for more details.

  17.  *

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

  19.  * License along with Libav; if not, write to the Free Software

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

  21.  */

  22. 

  23. #include "libavutil/cpu.h"

  24. #include "libavutil/x86_cpu.h"

  25. #include "libavcodec/vp8dsp.h"

  26. 

  27. #if HAVE_YASM

  28. 

  29. /*

  30.  * MC functions

  31.  */

  32. extern void ff_put_vp8_epel4_h4_mmxext(uint8_t *dst, ptrdiff_t dststride,

  33.                                        uint8_t *src, ptrdiff_t srcstride,

  34.                                        int height, int mx, int my);

  35. extern void ff_put_vp8_epel4_h6_mmxext(uint8_t *dst, ptrdiff_t dststride,

  36.                                        uint8_t *src, ptrdiff_t srcstride,

  37.                                        int height, int mx, int my);

  38. extern void ff_put_vp8_epel4_v4_mmxext(uint8_t *dst, ptrdiff_t dststride,

  39.                                        uint8_t *src, ptrdiff_t srcstride,

  40.                                        int height, int mx, int my);

  41. extern void ff_put_vp8_epel4_v6_mmxext(uint8_t *dst, ptrdiff_t dststride,

  42.                                        uint8_t *src, ptrdiff_t srcstride,

  43.                                        int height, int mx, int my);

  44. 

  45. extern void ff_put_vp8_epel8_h4_sse2  (uint8_t *dst, ptrdiff_t dststride,

  46.                                        uint8_t *src, ptrdiff_t srcstride,

  47.                                        int height, int mx, int my);

  48. extern void ff_put_vp8_epel8_h6_sse2  (uint8_t *dst, ptrdiff_t dststride,

  49.                                        uint8_t *src, ptrdiff_t srcstride,

  50.                                        int height, int mx, int my);

  51. extern void ff_put_vp8_epel8_v4_sse2  (uint8_t *dst, ptrdiff_t dststride,

  52.                                        uint8_t *src, ptrdiff_t srcstride,

  53.                                        int height, int mx, int my);

  54. extern void ff_put_vp8_epel8_v6_sse2  (uint8_t *dst, ptrdiff_t dststride,

  55.                                        uint8_t *src, ptrdiff_t srcstride,

  56.                                        int height, int mx, int my);

  57. 

  58. extern void ff_put_vp8_epel4_h4_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  59.                                        uint8_t *src, ptrdiff_t srcstride,

  60.                                        int height, int mx, int my);

  61. extern void ff_put_vp8_epel4_h6_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  62.                                        uint8_t *src, ptrdiff_t srcstride,

  63.                                        int height, int mx, int my);

  64. extern void ff_put_vp8_epel4_v4_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  65.                                        uint8_t *src, ptrdiff_t srcstride,

  66.                                        int height, int mx, int my);

  67. extern void ff_put_vp8_epel4_v6_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  68.                                        uint8_t *src, ptrdiff_t srcstride,

  69.                                        int height, int mx, int my);

  70. extern void ff_put_vp8_epel8_h4_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  71.                                        uint8_t *src, ptrdiff_t srcstride,

  72.                                        int height, int mx, int my);

  73. extern void ff_put_vp8_epel8_h6_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  74.                                        uint8_t *src, ptrdiff_t srcstride,

  75.                                        int height, int mx, int my);

  76. extern void ff_put_vp8_epel8_v4_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  77.                                        uint8_t *src, ptrdiff_t srcstride,

  78.                                        int height, int mx, int my);

  79. extern void ff_put_vp8_epel8_v6_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  80.                                        uint8_t *src, ptrdiff_t srcstride,

  81.                                        int height, int mx, int my);

  82. 

  83. extern void ff_put_vp8_bilinear4_h_mmxext(uint8_t *dst, ptrdiff_t dststride,

  84.                                           uint8_t *src, ptrdiff_t srcstride,

  85.                                           int height, int mx, int my);

  86. extern void ff_put_vp8_bilinear8_h_sse2  (uint8_t *dst, ptrdiff_t dststride,

  87.                                           uint8_t *src, ptrdiff_t srcstride,

  88.                                           int height, int mx, int my);

  89. extern void ff_put_vp8_bilinear4_h_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  90.                                           uint8_t *src, ptrdiff_t srcstride,

  91.                                           int height, int mx, int my);

  92. extern void ff_put_vp8_bilinear8_h_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  93.                                           uint8_t *src, ptrdiff_t srcstride,

  94.                                           int height, int mx, int my);

  95. 

  96. extern void ff_put_vp8_bilinear4_v_mmxext(uint8_t *dst, ptrdiff_t dststride,

  97.                                           uint8_t *src, ptrdiff_t srcstride,

  98.                                           int height, int mx, int my);

  99. extern void ff_put_vp8_bilinear8_v_sse2  (uint8_t *dst, ptrdiff_t dststride,

  100.                                           uint8_t *src, ptrdiff_t srcstride,

  101.                                           int height, int mx, int my);

  102. extern void ff_put_vp8_bilinear4_v_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  103.                                           uint8_t *src, ptrdiff_t srcstride,

  104.                                           int height, int mx, int my);

  105. extern void ff_put_vp8_bilinear8_v_ssse3 (uint8_t *dst, ptrdiff_t dststride,

  106.                                           uint8_t *src, ptrdiff_t srcstride,

  107.                                           int height, int mx, int my);

  108. 

  109. 

  110. extern void ff_put_vp8_pixels8_mmx (uint8_t *dst, ptrdiff_t dststride,

  111.                                     uint8_t *src, ptrdiff_t srcstride,

  112.                                     int height, int mx, int my);

  113. extern void ff_put_vp8_pixels16_mmx(uint8_t *dst, ptrdiff_t dststride,

  114.                                     uint8_t *src, ptrdiff_t srcstride,

  115.                                     int height, int mx, int my);

  116. extern void ff_put_vp8_pixels16_sse(uint8_t *dst, ptrdiff_t dststride,

  117.                                     uint8_t *src, ptrdiff_t srcstride,

  118.                                     int height, int mx, int my);

  119. 

  120. #define TAP_W16(OPT, FILTERTYPE, TAPTYPE) \

  121. static void ff_put_vp8_ ## FILTERTYPE ## 16_ ## TAPTYPE ## _ ## OPT( \

  122.     uint8_t *dst,  ptrdiff_t dststride, uint8_t *src, \

  123.     ptrdiff_t srcstride, int height, int mx, int my) \

  124. { \

  125.     ff_put_vp8_ ## FILTERTYPE ## 8_ ## TAPTYPE ## _ ## OPT( \

  126.         dst,     dststride, src,     srcstride, height, mx, my); \

  127.     ff_put_vp8_ ## FILTERTYPE ## 8_ ## TAPTYPE ## _ ## OPT( \

  128.         dst + 8, dststride, src + 8, srcstride, height, mx, my); \

  129. }

  130. #define TAP_W8(OPT, FILTERTYPE, TAPTYPE) \

  131. static void ff_put_vp8_ ## FILTERTYPE ## 8_ ## TAPTYPE ## _ ## OPT( \

  132.     uint8_t *dst,  ptrdiff_t dststride, uint8_t *src, \

  133.     ptrdiff_t srcstride, int height, int mx, int my) \

  134. { \

  135.     ff_put_vp8_ ## FILTERTYPE ## 4_ ## TAPTYPE ## _ ## OPT( \

  136.         dst,     dststride, src,     srcstride, height, mx, my); \

  137.     ff_put_vp8_ ## FILTERTYPE ## 4_ ## TAPTYPE ## _ ## OPT( \

  138.         dst + 4, dststride, src + 4, srcstride, height, mx, my); \

  139. }

  140. 

  141. TAP_W8 (mmxext, epel, h4)

  142. TAP_W8 (mmxext, epel, h6)

  143. TAP_W16(mmxext, epel, h6)

  144. TAP_W8 (mmxext, epel, v4)

  145. TAP_W8 (mmxext, epel, v6)

  146. TAP_W16(mmxext, epel, v6)

  147. TAP_W8 (mmxext, bilinear, h)

  148. TAP_W16(mmxext, bilinear, h)

  149. TAP_W8 (mmxext, bilinear, v)

  150. TAP_W16(mmxext, bilinear, v)

  151. 

  152. TAP_W16(sse2,   epel, h6)

  153. TAP_W16(sse2,   epel, v6)

  154. TAP_W16(sse2,   bilinear, h)

  155. TAP_W16(sse2,   bilinear, v)

  156. 

  157. TAP_W16(ssse3,  epel, h6)

  158. TAP_W16(ssse3,  epel, v6)

  159. TAP_W16(ssse3,  bilinear, h)

  160. TAP_W16(ssse3,  bilinear, v)

  161. 

  162. #define HVTAP(OPT, ALIGN, TAPNUMX, TAPNUMY, SIZE, MAXHEIGHT) \

  163. static void ff_put_vp8_epel ## SIZE ## _h ## TAPNUMX ## v ## TAPNUMY ## _ ## OPT( \

  164.     uint8_t *dst, ptrdiff_t dststride, uint8_t *src, \

  165.     ptrdiff_t srcstride, int height, int mx, int my) \

  166. { \

  167.     DECLARE_ALIGNED(ALIGN, uint8_t, tmp)[SIZE * (MAXHEIGHT + TAPNUMY - 1)]; \

  168.     uint8_t *tmpptr = tmp + SIZE * (TAPNUMY / 2 - 1); \

  169.     src -= srcstride * (TAPNUMY / 2 - 1); \

  170.     ff_put_vp8_epel ## SIZE ## _h ## TAPNUMX ## _ ## OPT( \

  171.         tmp, SIZE,      src,    srcstride, height + TAPNUMY - 1, mx, my); \

  172.     ff_put_vp8_epel ## SIZE ## _v ## TAPNUMY ## _ ## OPT( \

  173.         dst, dststride, tmpptr, SIZE,      height,               mx, my); \

  174. }

  175. 

  176. #define HVTAPMMX(x, y) \

  177. HVTAP(mmxext, 8, x, y,  4,  8) \

  178. HVTAP(mmxext, 8, x, y,  8, 16)

  179. 

  180. HVTAPMMX(4, 4)

  181. HVTAPMMX(4, 6)

  182. HVTAPMMX(6, 4)

  183. HVTAPMMX(6, 6)

  184. HVTAP(mmxext, 8, 6, 6, 16, 16)

  185. 

  186. #define HVTAPSSE2(x, y, w) \

  187. HVTAP(sse2,  16, x, y, w, 16) \

  188. HVTAP(ssse3, 16, x, y, w, 16)

  189. 

  190. HVTAPSSE2(4, 4, 8)

  191. HVTAPSSE2(4, 6, 8)

  192. HVTAPSSE2(6, 4, 8)

  193. HVTAPSSE2(6, 6, 8)

  194. HVTAPSSE2(6, 6, 16)

  195. 

  196. HVTAP(ssse3, 16, 4, 4, 4, 8)

  197. HVTAP(ssse3, 16, 4, 6, 4, 8)

  198. HVTAP(ssse3, 16, 6, 4, 4, 8)

  199. HVTAP(ssse3, 16, 6, 6, 4, 8)

  200. 

  201. #define HVBILIN(OPT, ALIGN, SIZE, MAXHEIGHT) \

  202. static void ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT( \

  203.     uint8_t *dst, ptrdiff_t dststride, uint8_t *src, \

  204.     ptrdiff_t srcstride, int height, int mx, int my) \

  205. { \

  206.     DECLARE_ALIGNED(ALIGN, uint8_t, tmp)[SIZE * (MAXHEIGHT + 2)]; \

  207.     ff_put_vp8_bilinear ## SIZE ## _h_ ## OPT( \

  208.         tmp, SIZE,      src, srcstride, height + 1, mx, my); \

  209.     ff_put_vp8_bilinear ## SIZE ## _v_ ## OPT( \

  210.         dst, dststride, tmp, SIZE,      height,     mx, my); \

  211. }

  212. 

  213. HVBILIN(mmxext, 8,  4,  8)

  214. HVBILIN(mmxext, 8,  8, 16)

  215. HVBILIN(mmxext, 8, 16, 16)

  216. HVBILIN(sse2,   8,  8, 16)

  217. HVBILIN(sse2,   8, 16, 16)

  218. HVBILIN(ssse3,  8,  4,  8)

  219. HVBILIN(ssse3,  8,  8, 16)

  220. HVBILIN(ssse3,  8, 16, 16)

  221. 

  222. extern void ff_vp8_idct_dc_add_mmx(uint8_t *dst, DCTELEM block[16],

  223.                                    ptrdiff_t stride);

  224. extern void ff_vp8_idct_dc_add_sse4(uint8_t *dst, DCTELEM block[16],

  225.                                     ptrdiff_t stride);

  226. extern void ff_vp8_idct_dc_add4y_mmx(uint8_t *dst, DCTELEM block[4][16],

  227.                                       ptrdiff_t stride);

  228. extern void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, DCTELEM block[4][16],

  229.                                       ptrdiff_t stride);

  230. extern void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, DCTELEM block[2][16],

  231.                                       ptrdiff_t stride);

  232. extern void ff_vp8_luma_dc_wht_mmx(DCTELEM block[4][4][16], DCTELEM dc[16]);

  233. extern void ff_vp8_luma_dc_wht_sse(DCTELEM block[4][4][16], DCTELEM dc[16]);

  234. extern void ff_vp8_idct_add_mmx(uint8_t *dst, DCTELEM block[16],

  235.                                 ptrdiff_t stride);

  236. extern void ff_vp8_idct_add_sse(uint8_t *dst, DCTELEM block[16],

  237.                                 ptrdiff_t stride);

  238. 

  239. #define DECLARE_LOOP_FILTER(NAME)\

  240. extern void ff_vp8_v_loop_filter_simple_ ## NAME(uint8_t *dst, \

  241.                                                  ptrdiff_t stride, \

  242.                                                  int flim);\

  243. extern void ff_vp8_h_loop_filter_simple_ ## NAME(uint8_t *dst, \

  244.                                                  ptrdiff_t stride, \

  245.                                                  int flim);\

  246. extern void ff_vp8_v_loop_filter16y_inner_ ## NAME (uint8_t *dst, \

  247.                                                      ptrdiff_t stride,\

  248.                                                     int e, int i, int hvt);\

  249. extern void ff_vp8_h_loop_filter16y_inner_ ## NAME (uint8_t *dst, \

  250.                                                     ptrdiff_t stride,\

  251.                                                     int e, int i, int hvt);\

  252. extern void ff_vp8_v_loop_filter8uv_inner_ ## NAME (uint8_t *dstU, \

  253.                                                     uint8_t *dstV,\

  254.                                                     ptrdiff_t s, \

  255.                                                     int e, int i, int hvt);\

  256. extern void ff_vp8_h_loop_filter8uv_inner_ ## NAME (uint8_t *dstU, \

  257.                                                     uint8_t *dstV,\

  258.                                                     ptrdiff_t s, \

  259.                                                     int e, int i, int hvt);\

  260. extern void ff_vp8_v_loop_filter16y_mbedge_ ## NAME(uint8_t *dst, \

  261.                                                     ptrdiff_t stride,\

  262.                                                     int e, int i, int hvt);\

  263. extern void ff_vp8_h_loop_filter16y_mbedge_ ## NAME(uint8_t *dst, \

  264.                                                     ptrdiff_t stride,\

  265.                                                     int e, int i, int hvt);\

  266. extern void ff_vp8_v_loop_filter8uv_mbedge_ ## NAME(uint8_t *dstU, \

  267.                                                     uint8_t *dstV,\

  268.                                                     ptrdiff_t s, \

  269.                                                     int e, int i, int hvt);\

  270. extern void ff_vp8_h_loop_filter8uv_mbedge_ ## NAME(uint8_t *dstU, \

  271.                                                     uint8_t *dstV,\

  272.                                                     ptrdiff_t s, \

  273.                                                     int e, int i, int hvt);

  274. 

  275. DECLARE_LOOP_FILTER(mmx)

  276. DECLARE_LOOP_FILTER(mmxext)

  277. DECLARE_LOOP_FILTER(sse2)

  278. DECLARE_LOOP_FILTER(ssse3)

  279. DECLARE_LOOP_FILTER(sse4)

  280. 

  281. #endif

  282. 

  283. #define VP8_LUMA_MC_FUNC(IDX, SIZE, OPT) \

  284.     c->put_vp8_epel_pixels_tab[IDX][0][2] = ff_put_vp8_epel ## SIZE ## _h6_ ## OPT; \

  285.     c->put_vp8_epel_pixels_tab[IDX][2][0] = ff_put_vp8_epel ## SIZE ## _v6_ ## OPT; \

  286.     c->put_vp8_epel_pixels_tab[IDX][2][2] = ff_put_vp8_epel ## SIZE ## _h6v6_ ## OPT

  287. 

  288. #define VP8_MC_FUNC(IDX, SIZE, OPT) \

  289.     c->put_vp8_epel_pixels_tab[IDX][0][1] = ff_put_vp8_epel ## SIZE ## _h4_ ## OPT; \

  290.     c->put_vp8_epel_pixels_tab[IDX][1][0] = ff_put_vp8_epel ## SIZE ## _v4_ ## OPT; \

  291.     c->put_vp8_epel_pixels_tab[IDX][1][1] = ff_put_vp8_epel ## SIZE ## _h4v4_ ## OPT; \

  292.     c->put_vp8_epel_pixels_tab[IDX][1][2] = ff_put_vp8_epel ## SIZE ## _h6v4_ ## OPT; \

  293.     c->put_vp8_epel_pixels_tab[IDX][2][1] = ff_put_vp8_epel ## SIZE ## _h4v6_ ## OPT; \

  294.     VP8_LUMA_MC_FUNC(IDX, SIZE, OPT)

  295. 

  296. #define VP8_BILINEAR_MC_FUNC(IDX, SIZE, OPT) \

  297.     c->put_vp8_bilinear_pixels_tab[IDX][0][1] = ff_put_vp8_bilinear ## SIZE ## _h_ ## OPT; \

  298.     c->put_vp8_bilinear_pixels_tab[IDX][0][2] = ff_put_vp8_bilinear ## SIZE ## _h_ ## OPT; \

  299.     c->put_vp8_bilinear_pixels_tab[IDX][1][0] = ff_put_vp8_bilinear ## SIZE ## _v_ ## OPT; \

  300.     c->put_vp8_bilinear_pixels_tab[IDX][1][1] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT; \

  301.     c->put_vp8_bilinear_pixels_tab[IDX][1][2] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT; \

  302.     c->put_vp8_bilinear_pixels_tab[IDX][2][0] = ff_put_vp8_bilinear ## SIZE ## _v_ ## OPT; \

  303.     c->put_vp8_bilinear_pixels_tab[IDX][2][1] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT; \

  304.     c->put_vp8_bilinear_pixels_tab[IDX][2][2] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT

  305. 

  306. 

  307. av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)

  308. {

  309. #if HAVE_YASM

  310.     int mm_flags = av_get_cpu_flags();

  311. 

  312.     if (mm_flags & AV_CPU_FLAG_MMX) {

  313.         c->vp8_idct_dc_add    = ff_vp8_idct_dc_add_mmx;

  314.         c->vp8_idct_dc_add4y  = ff_vp8_idct_dc_add4y_mmx;

  315.         c->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_mmx;

  316.         c->vp8_idct_add       = ff_vp8_idct_add_mmx;

  317.         c->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_mmx;

  318.         c->put_vp8_epel_pixels_tab[0][0][0]     =

  319.         c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_mmx;

  320.         c->put_vp8_epel_pixels_tab[1][0][0]     =

  321.         c->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_mmx;

  322. 

  323.         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmx;

  324.         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmx;

  325. 

  326.         c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmx;

  327.         c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmx;

  328.         c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_mmx;

  329.         c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_mmx;

  330. 

  331.         c->vp8_v_loop_filter16y       = ff_vp8_v_loop_filter16y_mbedge_mmx;

  332.         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_mmx;

  333.         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_mmx;

  334.         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_mmx;

  335.     }

  336. 

  337.     /* note that 4-tap width=16 functions are missing because w=16

  338.      * is only used for luma, and luma is always a copy or sixtap. */

  339.     if (mm_flags & AV_CPU_FLAG_MMX2) {

  340.         VP8_LUMA_MC_FUNC(0, 16, mmxext);

  341.         VP8_MC_FUNC(1, 8, mmxext);

  342.         VP8_MC_FUNC(2, 4, mmxext);

  343.         VP8_BILINEAR_MC_FUNC(0, 16, mmxext);

  344.         VP8_BILINEAR_MC_FUNC(1, 8, mmxext);

  345.         VP8_BILINEAR_MC_FUNC(2, 4, mmxext);

  346. 

  347.         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmxext;

  348.         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmxext;

  349. 

  350.         c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmxext;

  351.         c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmxext;

  352.         c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_mmxext;

  353.         c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_mmxext;

  354. 

  355.         c->vp8_v_loop_filter16y       = ff_vp8_v_loop_filter16y_mbedge_mmxext;

  356.         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_mmxext;

  357.         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_mmxext;

  358.         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_mmxext;

  359.     }

  360. 

  361.     if (mm_flags & AV_CPU_FLAG_SSE) {

  362.         c->vp8_idct_add                         = ff_vp8_idct_add_sse;

  363.         c->vp8_luma_dc_wht                      = ff_vp8_luma_dc_wht_sse;

  364.         c->put_vp8_epel_pixels_tab[0][0][0]     =

  365.         c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse;

  366.     }

  367. 

  368.     if (mm_flags & (AV_CPU_FLAG_SSE2|AV_CPU_FLAG_SSE2SLOW)) {

  369.         VP8_LUMA_MC_FUNC(0, 16, sse2);

  370.         VP8_MC_FUNC(1, 8, sse2);

  371.         VP8_BILINEAR_MC_FUNC(0, 16, sse2);

  372.         VP8_BILINEAR_MC_FUNC(1, 8, sse2);

  373. 

  374.         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2;

  375. 

  376.         c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2;

  377.         c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_sse2;

  378. 

  379.         c->vp8_v_loop_filter16y       = ff_vp8_v_loop_filter16y_mbedge_sse2;

  380.         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_sse2;

  381.     }

  382. 

  383.     if (mm_flags & AV_CPU_FLAG_SSE2) {

  384.         c->vp8_idct_dc_add4y          = ff_vp8_idct_dc_add4y_sse2;

  385. 

  386.         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2;

  387. 

  388.         c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2;

  389.         c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2;

  390. 

  391.         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_sse2;

  392.         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_sse2;

  393.     }

  394. 

  395.     if (mm_flags & AV_CPU_FLAG_SSSE3) {

  396.         VP8_LUMA_MC_FUNC(0, 16, ssse3);

  397.         VP8_MC_FUNC(1, 8, ssse3);

  398.         VP8_MC_FUNC(2, 4, ssse3);

  399.         VP8_BILINEAR_MC_FUNC(0, 16, ssse3);

  400.         VP8_BILINEAR_MC_FUNC(1, 8, ssse3);

  401.         VP8_BILINEAR_MC_FUNC(2, 4, ssse3);

  402. 

  403.         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_ssse3;

  404.         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_ssse3;

  405. 

  406.         c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_ssse3;

  407.         c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_ssse3;

  408.         c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_ssse3;

  409.         c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_ssse3;

  410. 

  411.         c->vp8_v_loop_filter16y       = ff_vp8_v_loop_filter16y_mbedge_ssse3;

  412.         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_ssse3;

  413.         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_ssse3;

  414.         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_ssse3;

  415.     }

  416. 

  417.     if (mm_flags & AV_CPU_FLAG_SSE4) {

  418.         c->vp8_idct_dc_add                  = ff_vp8_idct_dc_add_sse4;

  419. 

  420.         c->vp8_h_loop_filter_simple   = ff_vp8_h_loop_filter_simple_sse4;

  421.         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_sse4;

  422.         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_sse4;

  423.     }

  424. #endif

  425. }