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.
39979 Commits
32 Branches
821MB
Tree: 593886b5dd
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '593886b5dd'
${ noResults }
FFmpeg/libavcodec/x86/hpeldsp_init.c

283 lines
11KB
Raw Blame History 

  1. /*

  2.  * SIMD-optimized halfpel functions

  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/attributes.h"

  26. #include "libavutil/cpu.h"

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

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

  29. #include "libavcodec/avcodec.h"

  30. #include "libavcodec/hpeldsp.h"

  31. #include "libavcodec/pixels.h"

  32. #include "fpel.h"

  33. #include "hpeldsp.h"

  34. 

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

  36.                               ptrdiff_t line_size, int h);

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

  38.                              ptrdiff_t line_size, int h);

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

  40.                                ptrdiff_t line_size, int h);

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

  42.                               ptrdiff_t line_size, int h);

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

  44.                                      ptrdiff_t line_size, int h);

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

  46.                                     ptrdiff_t line_size, int h);

  47. void ff_put_no_rnd_pixels8_x2_exact_mmxext(uint8_t *block,

  48.                                            const uint8_t *pixels,

  49.                                            ptrdiff_t line_size, int h);

  50. void ff_put_no_rnd_pixels8_x2_exact_3dnow(uint8_t *block,

  51.                                           const uint8_t *pixels,

  52.                                           ptrdiff_t line_size, int h);

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

  54.                               ptrdiff_t line_size, int h);

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

  56.                              ptrdiff_t line_size, int h);

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

  58.                                      ptrdiff_t line_size, int h);

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

  60.                                     ptrdiff_t line_size, int h);

  61. void ff_put_no_rnd_pixels8_y2_exact_mmxext(uint8_t *block,

  62.                                            const uint8_t *pixels,

  63.                                            ptrdiff_t line_size, int h);

  64. void ff_put_no_rnd_pixels8_y2_exact_3dnow(uint8_t *block,

  65.                                           const uint8_t *pixels,

  66.                                           ptrdiff_t line_size, int h);

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

  68.                           ptrdiff_t line_size, int h);

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

  70.                               ptrdiff_t line_size, int h);

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

  72.                              ptrdiff_t line_size, int h);

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

  74.                               ptrdiff_t line_size, int h);

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

  76.                              ptrdiff_t line_size, int h);

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

  78.                                ptrdiff_t line_size, int h);

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

  80.                               ptrdiff_t line_size, int h);

  81. 

  82. #define avg_pixels8_mmx         ff_avg_pixels8_mmx

  83. #define avg_pixels8_x2_mmx      ff_avg_pixels8_x2_mmx

  84. #define avg_pixels16_mmx        ff_avg_pixels16_mmx

  85. #define avg_pixels8_xy2_mmx     ff_avg_pixels8_xy2_mmx

  86. #define avg_pixels16_xy2_mmx    ff_avg_pixels16_xy2_mmx

  87. #define put_pixels8_mmx         ff_put_pixels8_mmx

  88. #define put_pixels16_mmx        ff_put_pixels16_mmx

  89. #define put_pixels8_xy2_mmx     ff_put_pixels8_xy2_mmx

  90. #define put_pixels16_xy2_mmx    ff_put_pixels16_xy2_mmx

  91. #define avg_no_rnd_pixels16_mmx ff_avg_pixels16_mmx

  92. #define put_no_rnd_pixels8_mmx  ff_put_pixels8_mmx

  93. #define put_no_rnd_pixels16_mmx ff_put_pixels16_mmx

  94. 

  95. #if HAVE_INLINE_ASM

  96. 

  97. /***********************************/

  98. /* MMX no rounding */

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

  100. #define SET_RND  MOVQ_WONE

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

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

  103. #define STATIC static

  104. 

  105. #include "rnd_template.c"

  106. #include "hpeldsp_rnd_template.c"

  107. 

  108. #undef DEF

  109. #undef SET_RND

  110. #undef PAVGBP

  111. #undef PAVGB

  112. #undef STATIC

  113. 

  114. CALL_2X_PIXELS(avg_no_rnd_pixels16_y2_mmx, avg_no_rnd_pixels8_y2_mmx, 8)

  115. CALL_2X_PIXELS(put_no_rnd_pixels16_y2_mmx, put_no_rnd_pixels8_y2_mmx, 8)

  116. 

  117. CALL_2X_PIXELS(avg_no_rnd_pixels16_xy2_mmx, avg_no_rnd_pixels8_xy2_mmx, 8)

  118. CALL_2X_PIXELS(put_no_rnd_pixels16_xy2_mmx, put_no_rnd_pixels8_xy2_mmx, 8)

  119. 

  120. /***********************************/

  121. /* MMX rounding */

  122. 

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

  124. #define SET_RND  MOVQ_WTWO

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

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

  127. 

  128. #include "hpeldsp_rnd_template.c"

  129. 

  130. #undef DEF

  131. #define DEF(x, y) ff_ ## x ## _ ## y ## _mmx

  132. #define STATIC

  133. 

  134. #include "rnd_template.c"

  135. 

  136. #undef DEF

  137. #undef SET_RND

  138. #undef PAVGBP

  139. #undef PAVGB

  140. 

  141. CALL_2X_PIXELS(avg_pixels16_y2_mmx, avg_pixels8_y2_mmx, 8)

  142. CALL_2X_PIXELS(put_pixels16_y2_mmx, put_pixels8_y2_mmx, 8)

  143. 

  144. CALL_2X_PIXELS_EXPORT(ff_avg_pixels16_xy2_mmx, ff_avg_pixels8_xy2_mmx, 8)

  145. CALL_2X_PIXELS_EXPORT(ff_put_pixels16_xy2_mmx, ff_put_pixels8_xy2_mmx, 8)

  146. 

  147. #endif /* HAVE_INLINE_ASM */

  148. 

  149. 

  150. #if HAVE_YASM

  151. 

  152. #define HPELDSP_AVG_PIXELS16(CPUEXT)                      \

  153.     CALL_2X_PIXELS(put_no_rnd_pixels16_x2 ## CPUEXT, ff_put_no_rnd_pixels8_x2 ## CPUEXT, 8) \

  154.     CALL_2X_PIXELS(put_pixels16_y2        ## CPUEXT, ff_put_pixels8_y2        ## CPUEXT, 8) \

  155.     CALL_2X_PIXELS(put_no_rnd_pixels16_y2 ## CPUEXT, ff_put_no_rnd_pixels8_y2 ## CPUEXT, 8) \

  156.     CALL_2X_PIXELS(avg_pixels16           ## CPUEXT, ff_avg_pixels8           ## CPUEXT, 8) \

  157.     CALL_2X_PIXELS(avg_pixels16_x2        ## CPUEXT, ff_avg_pixels8_x2        ## CPUEXT, 8) \

  158.     CALL_2X_PIXELS(avg_pixels16_y2        ## CPUEXT, ff_avg_pixels8_y2        ## CPUEXT, 8) \

  159.     CALL_2X_PIXELS(avg_pixels16_xy2       ## CPUEXT, ff_avg_pixels8_xy2       ## CPUEXT, 8)

  160. 

  161. HPELDSP_AVG_PIXELS16(_3dnow)

  162. HPELDSP_AVG_PIXELS16(_mmxext)

  163. 

  164. #endif /* HAVE_YASM */

  165. 

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

  167.     do {                                                                        \

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

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

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

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

  172.     } while (0)

  173. 

  174. static void hpeldsp_init_mmx(HpelDSPContext *c, int flags, int cpu_flags)

  175. {

  176. #if HAVE_MMX_INLINE

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

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

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

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

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

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

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

  184. #endif /* HAVE_MMX_INLINE */

  185. }

  186. 

  187. static void hpeldsp_init_mmxext(HpelDSPContext *c, int flags, int cpu_flags)

  188. {

  189. #if HAVE_MMXEXT_EXTERNAL

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

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

  192. 

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

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

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

  196. 

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

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

  199. 

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

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

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

  203. 

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

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

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

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

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

  209. 

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

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

  212.     }

  213. 

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

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

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

  217.     }

  218. #endif /* HAVE_MMXEXT_EXTERNAL */

  219. }

  220. 

  221. static void hpeldsp_init_3dnow(HpelDSPContext *c, int flags, int cpu_flags)

  222. {

  223. #if HAVE_AMD3DNOW_EXTERNAL

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

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

  226. 

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

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

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

  230. 

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

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

  233. 

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

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

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

  237. 

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

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

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

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

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

  243. 

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

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

  246.     }

  247. 

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

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

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

  251.     }

  252. #endif /* HAVE_AMD3DNOW_EXTERNAL */

  253. }

  254. 

  255. static void hpeldsp_init_sse2(HpelDSPContext *c, int flags, int cpu_flags)

  256. {

  257. #if HAVE_SSE2_EXTERNAL

  258.     if (!(cpu_flags & AV_CPU_FLAG_SSE2SLOW)) {

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

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

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

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

  263.     }

  264. #endif /* HAVE_SSE2_EXTERNAL */

  265. }

  266. 

  267. av_cold void ff_hpeldsp_init_x86(HpelDSPContext *c, int flags)

  268. {

  269.     int cpu_flags = av_get_cpu_flags();

  270. 

  271.     if (INLINE_MMX(cpu_flags))

  272.         hpeldsp_init_mmx(c, flags, cpu_flags);

  273. 

  274.     if (EXTERNAL_AMD3DNOW(cpu_flags))

  275.         hpeldsp_init_3dnow(c, flags, cpu_flags);

  276. 

  277.     if (EXTERNAL_MMXEXT(cpu_flags))

  278.         hpeldsp_init_mmxext(c, flags, cpu_flags);

  279. 

  280.     if (EXTERNAL_SSE2(cpu_flags))

  281.         hpeldsp_init_sse2(c, flags, cpu_flags);

  282. }