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.
37297 Commits
32 Branches
552MB
Tree: 1b343cedd7
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '1b343cedd7'
${ noResults }
FFmpeg/libavcodec/x86/hpeldsp_init.c

264 lines
10KB
Raw Blame History 

  1. /*

  2.  * MMX optimized DSP utils

  3.  * Copyright (c) 2000, 2001 Fabrice Bellard

  4.  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>

  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.  * MMX optimization by Nick Kurshev <nickols_k@mail.ru>

  23.  */

  24. 

  25. #include "libavutil/cpu.h"

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

  27. #include "libavcodec/hpeldsp.h"

  28. #include "dsputil_mmx.h"

  29. 

  30. void ff_put_pixels8_x2_mmxext(uint8_t *block, const uint8_t *pixels,

  31.                               ptrdiff_t line_size, int h);

  32. void ff_put_pixels8_x2_3dnow(uint8_t *block, const uint8_t *pixels,

  33.                              ptrdiff_t line_size, int h);

  34. void ff_put_pixels16_x2_mmxext(uint8_t *block, const uint8_t *pixels,

  35.                                ptrdiff_t line_size, int h);

  36. void ff_put_pixels16_x2_3dnow(uint8_t *block, const uint8_t *pixels,

  37.                               ptrdiff_t line_size, int h);

  38. void ff_put_no_rnd_pixels8_x2_mmxext(uint8_t *block, const uint8_t *pixels,

  39.                                      ptrdiff_t line_size, int h);

  40. void ff_put_no_rnd_pixels8_x2_3dnow(uint8_t *block, const uint8_t *pixels,

  41.                                     ptrdiff_t line_size, int h);

  42. void ff_put_no_rnd_pixels8_x2_exact_mmxext(uint8_t *block,

  43.                                            const uint8_t *pixels,

  44.                                            ptrdiff_t line_size, int h);

  45. void ff_put_no_rnd_pixels8_x2_exact_3dnow(uint8_t *block,

  46.                                           const uint8_t *pixels,

  47.                                           ptrdiff_t line_size, int h);

  48. void ff_put_pixels8_y2_mmxext(uint8_t *block, const uint8_t *pixels,

  49.                               ptrdiff_t line_size, int h);

  50. void ff_put_pixels8_y2_3dnow(uint8_t *block, const uint8_t *pixels,

  51.                              ptrdiff_t line_size, int h);

  52. void ff_put_no_rnd_pixels8_y2_mmxext(uint8_t *block, const uint8_t *pixels,

  53.                                      ptrdiff_t line_size, int h);

  54. void ff_put_no_rnd_pixels8_y2_3dnow(uint8_t *block, const uint8_t *pixels,

  55.                                     ptrdiff_t line_size, int h);

  56. void ff_put_no_rnd_pixels8_y2_exact_mmxext(uint8_t *block,

  57.                                            const uint8_t *pixels,

  58.                                            ptrdiff_t line_size, int h);

  59. void ff_put_no_rnd_pixels8_y2_exact_3dnow(uint8_t *block,

  60.                                           const uint8_t *pixels,

  61.                                           ptrdiff_t line_size, int h);

  62. void ff_avg_pixels8_3dnow(uint8_t *block, const uint8_t *pixels,

  63.                           ptrdiff_t line_size, int h);

  64. void ff_avg_pixels8_x2_mmxext(uint8_t *block, const uint8_t *pixels,

  65.                               ptrdiff_t line_size, int h);

  66. void ff_avg_pixels8_x2_3dnow(uint8_t *block, const uint8_t *pixels,

  67.                              ptrdiff_t line_size, int h);

  68. void ff_avg_pixels8_y2_mmxext(uint8_t *block, const uint8_t *pixels,

  69.                               ptrdiff_t line_size, int h);

  70. void ff_avg_pixels8_y2_3dnow(uint8_t *block, const uint8_t *pixels,

  71.                              ptrdiff_t line_size, int h);

  72. void ff_avg_pixels8_xy2_mmxext(uint8_t *block, const uint8_t *pixels,

  73.                                ptrdiff_t line_size, int h);

  74. void ff_avg_pixels8_xy2_3dnow(uint8_t *block, const uint8_t *pixels,

  75.                               ptrdiff_t line_size, int h);

  76. 

  77. #define avg_pixels8_mmx         ff_avg_pixels8_mmx

  78. #define avg_pixels8_x2_mmx      ff_avg_pixels8_x2_mmx

  79. #define avg_pixels16_mmx        ff_avg_pixels16_mmx

  80. #define avg_pixels8_xy2_mmx     ff_avg_pixels8_xy2_mmx

  81. #define avg_pixels16_xy2_mmx    ff_avg_pixels16_xy2_mmx

  82. #define put_pixels8_mmx         ff_put_pixels8_mmx

  83. #define put_pixels16_mmx        ff_put_pixels16_mmx

  84. #define put_pixels8_xy2_mmx     ff_put_pixels8_xy2_mmx

  85. #define put_pixels16_xy2_mmx    ff_put_pixels16_xy2_mmx

  86. #define avg_no_rnd_pixels16_mmx ff_avg_pixels16_mmx

  87. #define put_no_rnd_pixels8_mmx  ff_put_pixels8_mmx

  88. #define put_no_rnd_pixels16_mmx ff_put_pixels16_mmx

  89. 

  90. #if HAVE_INLINE_ASM

  91. 

  92. /***********************************/

  93. /* MMX no rounding */

  94. #define DEF(x, y) x ## _no_rnd_ ## y ## _mmx

  95. #define SET_RND  MOVQ_WONE

  96. #define PAVGBP(a, b, c, d, e, f)        PAVGBP_MMX_NO_RND(a, b, c, d, e, f)

  97. #define PAVGB(a, b, c, e)               PAVGB_MMX_NO_RND(a, b, c, e)

  98. #define STATIC static

  99. 

  100. #include "rnd_template.c"

  101. #include "hpeldsp_rnd_template.c"

  102. 

  103. #undef DEF

  104. #undef SET_RND

  105. #undef PAVGBP

  106. #undef PAVGB

  107. #undef STATIC

  108. /***********************************/

  109. /* MMX rounding */

  110. 

  111. #define DEF(x, y) x ## _ ## y ## _mmx

  112. #define SET_RND  MOVQ_WTWO

  113. #define PAVGBP(a, b, c, d, e, f)        PAVGBP_MMX(a, b, c, d, e, f)

  114. #define PAVGB(a, b, c, e)               PAVGB_MMX(a, b, c, e)

  115. 

  116. #include "hpeldsp_rnd_template.c"

  117. 

  118. #undef DEF

  119. #undef SET_RND

  120. #undef PAVGBP

  121. #undef PAVGB

  122. 

  123. #endif /* HAVE_INLINE_ASM */

  124. 

  125. 

  126. #if HAVE_YASM

  127. /***********************************/

  128. /* 3Dnow specific */

  129. 

  130. #define DEF(x) x ## _3dnow

  131. 

  132. #include "hpeldsp_avg_template.c"

  133. 

  134. #undef DEF

  135. 

  136. /***********************************/

  137. /* MMXEXT specific */

  138. 

  139. #define DEF(x) x ## _mmxext

  140. 

  141. #include "hpeldsp_avg_template.c"

  142. 

  143. #undef DEF

  144. 

  145. #endif /* HAVE_YASM */

  146. 

  147. #define SET_HPEL_FUNCS(PFX, IDX, SIZE, CPU)                                     \

  148.     do {                                                                        \

  149.         c->PFX ## _pixels_tab IDX [0] = PFX ## _pixels ## SIZE ## _     ## CPU; \

  150.         c->PFX ## _pixels_tab IDX [1] = PFX ## _pixels ## SIZE ## _x2_  ## CPU; \

  151.         c->PFX ## _pixels_tab IDX [2] = PFX ## _pixels ## SIZE ## _y2_  ## CPU; \

  152.         c->PFX ## _pixels_tab IDX [3] = PFX ## _pixels ## SIZE ## _xy2_ ## CPU; \

  153.     } while (0)

  154. 

  155. static void hpeldsp_init_mmx(HpelDSPContext *c, int flags, int mm_flags)

  156. {

  157. #if HAVE_MMX_INLINE

  158.     SET_HPEL_FUNCS(put,        [0], 16, mmx);

  159.     SET_HPEL_FUNCS(put_no_rnd, [0], 16, mmx);

  160.     SET_HPEL_FUNCS(avg,        [0], 16, mmx);

  161.     SET_HPEL_FUNCS(avg_no_rnd,    , 16, mmx);

  162.     SET_HPEL_FUNCS(put,        [1],  8, mmx);

  163.     SET_HPEL_FUNCS(put_no_rnd, [1],  8, mmx);

  164.     SET_HPEL_FUNCS(avg,        [1],  8, mmx);

  165. #endif /* HAVE_MMX_INLINE */

  166. }

  167. 

  168. static void hpeldsp_init_mmxext(HpelDSPContext *c, int flags, int mm_flags)

  169. {

  170. #if HAVE_MMXEXT_EXTERNAL

  171.     c->put_pixels_tab[0][1] = ff_put_pixels16_x2_mmxext;

  172.     c->put_pixels_tab[0][2] = put_pixels16_y2_mmxext;

  173. 

  174.     c->avg_pixels_tab[0][0] = avg_pixels16_mmxext;

  175.     c->avg_pixels_tab[0][1] = avg_pixels16_x2_mmxext;

  176.     c->avg_pixels_tab[0][2] = avg_pixels16_y2_mmxext;

  177. 

  178.     c->put_pixels_tab[1][1] = ff_put_pixels8_x2_mmxext;

  179.     c->put_pixels_tab[1][2] = ff_put_pixels8_y2_mmxext;

  180. 

  181.     c->avg_pixels_tab[1][0] = ff_avg_pixels8_mmxext;

  182.     c->avg_pixels_tab[1][1] = ff_avg_pixels8_x2_mmxext;

  183.     c->avg_pixels_tab[1][2] = ff_avg_pixels8_y2_mmxext;

  184. 

  185.     if (!(flags & CODEC_FLAG_BITEXACT)) {

  186.         c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_mmxext;

  187.         c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_mmxext;

  188.         c->put_no_rnd_pixels_tab[1][1] = ff_put_no_rnd_pixels8_x2_mmxext;

  189.         c->put_no_rnd_pixels_tab[1][2] = ff_put_no_rnd_pixels8_y2_mmxext;

  190. 

  191.         c->avg_pixels_tab[0][3] = avg_pixels16_xy2_mmxext;

  192.         c->avg_pixels_tab[1][3] = ff_avg_pixels8_xy2_mmxext;

  193.     }

  194. 

  195.     if (flags & CODEC_FLAG_BITEXACT && CONFIG_VP3_DECODER) {

  196.         c->put_no_rnd_pixels_tab[1][1] = ff_put_no_rnd_pixels8_x2_exact_mmxext;

  197.         c->put_no_rnd_pixels_tab[1][2] = ff_put_no_rnd_pixels8_y2_exact_mmxext;

  198.     }

  199. #endif /* HAVE_MMXEXT_EXTERNAL */

  200. }

  201. 

  202. static void hpeldsp_init_3dnow(HpelDSPContext *c, int flags, int mm_flags)

  203. {

  204. #if HAVE_AMD3DNOW_EXTERNAL

  205.     c->put_pixels_tab[0][1] = ff_put_pixels16_x2_3dnow;

  206.     c->put_pixels_tab[0][2] = put_pixels16_y2_3dnow;

  207. 

  208.     c->avg_pixels_tab[0][0] = avg_pixels16_3dnow;

  209.     c->avg_pixels_tab[0][1] = avg_pixels16_x2_3dnow;

  210.     c->avg_pixels_tab[0][2] = avg_pixels16_y2_3dnow;

  211. 

  212.     c->put_pixels_tab[1][1] = ff_put_pixels8_x2_3dnow;

  213.     c->put_pixels_tab[1][2] = ff_put_pixels8_y2_3dnow;

  214. 

  215.     c->avg_pixels_tab[1][0] = ff_avg_pixels8_3dnow;

  216.     c->avg_pixels_tab[1][1] = ff_avg_pixels8_x2_3dnow;

  217.     c->avg_pixels_tab[1][2] = ff_avg_pixels8_y2_3dnow;

  218. 

  219.     if (!(flags & CODEC_FLAG_BITEXACT)){

  220.         c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_3dnow;

  221.         c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_3dnow;

  222.         c->put_no_rnd_pixels_tab[1][1] = ff_put_no_rnd_pixels8_x2_3dnow;

  223.         c->put_no_rnd_pixels_tab[1][2] = ff_put_no_rnd_pixels8_y2_3dnow;

  224. 

  225.         c->avg_pixels_tab[0][3] = avg_pixels16_xy2_3dnow;

  226.         c->avg_pixels_tab[1][3] = ff_avg_pixels8_xy2_3dnow;

  227.     }

  228. 

  229.     if (flags & CODEC_FLAG_BITEXACT && CONFIG_VP3_DECODER) {

  230.         c->put_no_rnd_pixels_tab[1][1] = ff_put_no_rnd_pixels8_x2_exact_3dnow;

  231.         c->put_no_rnd_pixels_tab[1][2] = ff_put_no_rnd_pixels8_y2_exact_3dnow;

  232.     }

  233. #endif /* HAVE_AMD3DNOW_EXTERNAL */

  234. }

  235. 

  236. static void hpeldsp_init_sse2(HpelDSPContext *c, int flags, int mm_flags)

  237. {

  238. #if HAVE_SSE2_EXTERNAL

  239.     if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {

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

  241.         c->put_pixels_tab[0][0]        = ff_put_pixels16_sse2;

  242.         c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_sse2;

  243.         c->avg_pixels_tab[0][0]        = ff_avg_pixels16_sse2;

  244.     }

  245. #endif /* HAVE_SSE2_EXTERNAL */

  246. }

  247. 

  248. void ff_hpeldsp_init_x86(HpelDSPContext *c, int flags)

  249. {

  250.     int mm_flags = av_get_cpu_flags();

  251. 

  252.     if (mm_flags & AV_CPU_FLAG_MMX)

  253.         hpeldsp_init_mmx(c, flags, mm_flags);

  254. 

  255.     if (mm_flags & AV_CPU_FLAG_MMXEXT)

  256.         hpeldsp_init_mmxext(c, flags, mm_flags);

  257. 

  258.     if (mm_flags & AV_CPU_FLAG_3DNOW)

  259.         hpeldsp_init_3dnow(c, flags, mm_flags);

  260. 

  261.     if (mm_flags & AV_CPU_FLAG_SSE2)

  262.         hpeldsp_init_sse2(c, flags, mm_flags);

  263. }