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.
37285 Commits
32 Branches
313MB
Tree: 481575d255
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '481575d255'
${ noResults }
FFmpeg/libavcodec/x86/rnd_template.c

180 lines
7.5KB
Raw Blame History 

  1. /*

  2.  * DSP utils mmx functions are compiled twice for rnd/no_rnd

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

  4.  * Copyright (c) 2003-2004 Michael Niedermayer <michaelni@gmx.at>

  5.  *

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

  7.  * mostly rewritten by Michael Niedermayer <michaelni@gmx.at>

  8.  * and improved by Zdenek Kabelac <kabi@users.sf.net>

  9.  *

  10.  * This file is part of Libav.

  11.  *

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

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

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

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

  16.  *

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

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

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

  20.  * Lesser General Public License for more details.

  21.  *

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

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

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

  25.  */

  26. 

  27. // put_pixels

  28. static void DEF(put, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)

  29. {

  30.     MOVQ_ZERO(mm7);

  31.     SET_RND(mm6); // =2 for rnd  and  =1 for no_rnd version

  32.     __asm__ volatile(

  33.         "movq   (%1), %%mm0             \n\t"

  34.         "movq   1(%1), %%mm4            \n\t"

  35.         "movq   %%mm0, %%mm1            \n\t"

  36.         "movq   %%mm4, %%mm5            \n\t"

  37.         "punpcklbw %%mm7, %%mm0         \n\t"

  38.         "punpcklbw %%mm7, %%mm4         \n\t"

  39.         "punpckhbw %%mm7, %%mm1         \n\t"

  40.         "punpckhbw %%mm7, %%mm5         \n\t"

  41.         "paddusw %%mm0, %%mm4           \n\t"

  42.         "paddusw %%mm1, %%mm5           \n\t"

  43.         "xor    %%"REG_a", %%"REG_a"    \n\t"

  44.         "add    %3, %1                  \n\t"

  45.         ".p2align 3                     \n\t"

  46.         "1:                             \n\t"

  47.         "movq   (%1, %%"REG_a"), %%mm0  \n\t"

  48.         "movq   1(%1, %%"REG_a"), %%mm2 \n\t"

  49.         "movq   %%mm0, %%mm1            \n\t"

  50.         "movq   %%mm2, %%mm3            \n\t"

  51.         "punpcklbw %%mm7, %%mm0         \n\t"

  52.         "punpcklbw %%mm7, %%mm2         \n\t"

  53.         "punpckhbw %%mm7, %%mm1         \n\t"

  54.         "punpckhbw %%mm7, %%mm3         \n\t"

  55.         "paddusw %%mm2, %%mm0           \n\t"

  56.         "paddusw %%mm3, %%mm1           \n\t"

  57.         "paddusw %%mm6, %%mm4           \n\t"

  58.         "paddusw %%mm6, %%mm5           \n\t"

  59.         "paddusw %%mm0, %%mm4           \n\t"

  60.         "paddusw %%mm1, %%mm5           \n\t"

  61.         "psrlw  $2, %%mm4               \n\t"

  62.         "psrlw  $2, %%mm5               \n\t"

  63.         "packuswb  %%mm5, %%mm4         \n\t"

  64.         "movq   %%mm4, (%2, %%"REG_a")  \n\t"

  65.         "add    %3, %%"REG_a"           \n\t"

  66. 

  67.         "movq   (%1, %%"REG_a"), %%mm2  \n\t" // 0 <-> 2   1 <-> 3

  68.         "movq   1(%1, %%"REG_a"), %%mm4 \n\t"

  69.         "movq   %%mm2, %%mm3            \n\t"

  70.         "movq   %%mm4, %%mm5            \n\t"

  71.         "punpcklbw %%mm7, %%mm2         \n\t"

  72.         "punpcklbw %%mm7, %%mm4         \n\t"

  73.         "punpckhbw %%mm7, %%mm3         \n\t"

  74.         "punpckhbw %%mm7, %%mm5         \n\t"

  75.         "paddusw %%mm2, %%mm4           \n\t"

  76.         "paddusw %%mm3, %%mm5           \n\t"

  77.         "paddusw %%mm6, %%mm0           \n\t"

  78.         "paddusw %%mm6, %%mm1           \n\t"

  79.         "paddusw %%mm4, %%mm0           \n\t"

  80.         "paddusw %%mm5, %%mm1           \n\t"

  81.         "psrlw  $2, %%mm0               \n\t"

  82.         "psrlw  $2, %%mm1               \n\t"

  83.         "packuswb  %%mm1, %%mm0         \n\t"

  84.         "movq   %%mm0, (%2, %%"REG_a")  \n\t"

  85.         "add    %3, %%"REG_a"           \n\t"

  86. 

  87.         "subl   $2, %0                  \n\t"

  88.         "jnz    1b                      \n\t"

  89.         :"+g"(h), "+S"(pixels)

  90.         :"D"(block), "r"((x86_reg)line_size)

  91.         :REG_a, "memory");

  92. }

  93. 

  94. // avg_pixels

  95. // this routine is 'slightly' suboptimal but mostly unused

  96. static void DEF(avg, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)

  97. {

  98.     MOVQ_ZERO(mm7);

  99.     SET_RND(mm6); // =2 for rnd  and  =1 for no_rnd version

  100.     __asm__ volatile(

  101.         "movq   (%1), %%mm0             \n\t"

  102.         "movq   1(%1), %%mm4            \n\t"

  103.         "movq   %%mm0, %%mm1            \n\t"

  104.         "movq   %%mm4, %%mm5            \n\t"

  105.         "punpcklbw %%mm7, %%mm0         \n\t"

  106.         "punpcklbw %%mm7, %%mm4         \n\t"

  107.         "punpckhbw %%mm7, %%mm1         \n\t"

  108.         "punpckhbw %%mm7, %%mm5         \n\t"

  109.         "paddusw %%mm0, %%mm4           \n\t"

  110.         "paddusw %%mm1, %%mm5           \n\t"

  111.         "xor    %%"REG_a", %%"REG_a"    \n\t"

  112.         "add    %3, %1                  \n\t"

  113.         ".p2align 3                     \n\t"

  114.         "1:                             \n\t"

  115.         "movq   (%1, %%"REG_a"), %%mm0  \n\t"

  116.         "movq   1(%1, %%"REG_a"), %%mm2 \n\t"

  117.         "movq   %%mm0, %%mm1            \n\t"

  118.         "movq   %%mm2, %%mm3            \n\t"

  119.         "punpcklbw %%mm7, %%mm0         \n\t"

  120.         "punpcklbw %%mm7, %%mm2         \n\t"

  121.         "punpckhbw %%mm7, %%mm1         \n\t"

  122.         "punpckhbw %%mm7, %%mm3         \n\t"

  123.         "paddusw %%mm2, %%mm0           \n\t"

  124.         "paddusw %%mm3, %%mm1           \n\t"

  125.         "paddusw %%mm6, %%mm4           \n\t"

  126.         "paddusw %%mm6, %%mm5           \n\t"

  127.         "paddusw %%mm0, %%mm4           \n\t"

  128.         "paddusw %%mm1, %%mm5           \n\t"

  129.         "psrlw  $2, %%mm4               \n\t"

  130.         "psrlw  $2, %%mm5               \n\t"

  131.                 "movq   (%2, %%"REG_a"), %%mm3  \n\t"

  132.         "packuswb  %%mm5, %%mm4         \n\t"

  133.                 "pcmpeqd %%mm2, %%mm2   \n\t"

  134.                 "paddb %%mm2, %%mm2     \n\t"

  135.                 OP_AVG(%%mm3, %%mm4, %%mm5, %%mm2)

  136.                 "movq   %%mm5, (%2, %%"REG_a")  \n\t"

  137.         "add    %3, %%"REG_a"                \n\t"

  138. 

  139.         "movq   (%1, %%"REG_a"), %%mm2  \n\t" // 0 <-> 2   1 <-> 3

  140.         "movq   1(%1, %%"REG_a"), %%mm4 \n\t"

  141.         "movq   %%mm2, %%mm3            \n\t"

  142.         "movq   %%mm4, %%mm5            \n\t"

  143.         "punpcklbw %%mm7, %%mm2         \n\t"

  144.         "punpcklbw %%mm7, %%mm4         \n\t"

  145.         "punpckhbw %%mm7, %%mm3         \n\t"

  146.         "punpckhbw %%mm7, %%mm5         \n\t"

  147.         "paddusw %%mm2, %%mm4           \n\t"

  148.         "paddusw %%mm3, %%mm5           \n\t"

  149.         "paddusw %%mm6, %%mm0           \n\t"

  150.         "paddusw %%mm6, %%mm1           \n\t"

  151.         "paddusw %%mm4, %%mm0           \n\t"

  152.         "paddusw %%mm5, %%mm1           \n\t"

  153.         "psrlw  $2, %%mm0               \n\t"

  154.         "psrlw  $2, %%mm1               \n\t"

  155.                 "movq   (%2, %%"REG_a"), %%mm3  \n\t"

  156.         "packuswb  %%mm1, %%mm0         \n\t"

  157.                 "pcmpeqd %%mm2, %%mm2   \n\t"

  158.                 "paddb %%mm2, %%mm2     \n\t"

  159.                 OP_AVG(%%mm3, %%mm0, %%mm1, %%mm2)

  160.                 "movq   %%mm1, (%2, %%"REG_a")  \n\t"

  161.         "add    %3, %%"REG_a"           \n\t"

  162. 

  163.         "subl   $2, %0                  \n\t"

  164.         "jnz    1b                      \n\t"

  165.         :"+g"(h), "+S"(pixels)

  166.         :"D"(block), "r"((x86_reg)line_size)

  167.         :REG_a, "memory");

  168. }

  169. 

  170. //FIXME optimize

  171. static void DEF(put, pixels16_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){

  172.     DEF(put, pixels8_xy2)(block  , pixels  , line_size, h);

  173.     DEF(put, pixels8_xy2)(block+8, pixels+8, line_size, h);

  174. }

  175. 

  176. static void DEF(avg, pixels16_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){

  177.     DEF(avg, pixels8_xy2)(block  , pixels  , line_size, h);

  178.     DEF(avg, pixels8_xy2)(block+8, pixels+8, line_size, h);

  179. }