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.
35015 Commits
32 Branches
582MB
Tree: 90c02ae142
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '90c02ae142'
${ noResults }
FFmpeg/libswscale/x86/swscale_template.c

1967 lines
87KB
Raw Blame History 

  1. /*

  2.  * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>

  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. #undef REAL_MOVNTQ

  22. #undef MOVNTQ

  23. #undef MOVNTQ2

  24. #undef PREFETCH

  25. 

  26. #if COMPILE_TEMPLATE_MMX2

  27. #define PREFETCH "prefetchnta"

  28. #else

  29. #define PREFETCH  " # nop"

  30. #endif

  31. 

  32. #if COMPILE_TEMPLATE_MMX2

  33. #define REAL_MOVNTQ(a,b) "movntq " #a ", " #b " \n\t"

  34. #define MOVNTQ2 "movntq "

  35. #else

  36. #define REAL_MOVNTQ(a,b) "movq " #a ", " #b " \n\t"

  37. #define MOVNTQ2 "movq "

  38. #endif

  39. #define MOVNTQ(a,b)  REAL_MOVNTQ(a,b)

  40. 

  41. #if !COMPILE_TEMPLATE_MMX2

  42. static av_always_inline void

  43. dither_8to16(const uint8_t *srcDither, int rot)

  44. {

  45.     if (rot) {

  46.         __asm__ volatile("pxor      %%mm0, %%mm0\n\t"

  47.                          "movq       (%0), %%mm3\n\t"

  48.                          "movq      %%mm3, %%mm4\n\t"

  49.                          "psrlq       $24, %%mm3\n\t"

  50.                          "psllq       $40, %%mm4\n\t"

  51.                          "por       %%mm4, %%mm3\n\t"

  52.                          "movq      %%mm3, %%mm4\n\t"

  53.                          "punpcklbw %%mm0, %%mm3\n\t"

  54.                          "punpckhbw %%mm0, %%mm4\n\t"

  55.                          :: "r"(srcDither)

  56.                          );

  57.     } else {

  58.         __asm__ volatile("pxor      %%mm0, %%mm0\n\t"

  59.                          "movq       (%0), %%mm3\n\t"

  60.                          "movq      %%mm3, %%mm4\n\t"

  61.                          "punpcklbw %%mm0, %%mm3\n\t"

  62.                          "punpckhbw %%mm0, %%mm4\n\t"

  63.                          :: "r"(srcDither)

  64.                          );

  65.     }

  66. }

  67. #endif

  68. 

  69. static void RENAME(yuv2yuvX)(const int16_t *filter, int filterSize,

  70.                            const int16_t **src, uint8_t *dest, int dstW,

  71.                            const uint8_t *dither, int offset)

  72. {

  73.     dither_8to16(dither, offset);

  74.     __asm__ volatile(\

  75.         "psraw        $4, %%mm3\n\t"

  76.         "psraw        $4, %%mm4\n\t"

  77.         "movq    %%mm3, %%mm6\n\t"

  78.         "movq    %%mm4, %%mm7\n\t"

  79.         "movl %3, %%ecx\n\t"

  80.         "mov                                 %0, %%"REG_d"  \n\t"\

  81.         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\

  82.         ".p2align                             4             \n\t" /* FIXME Unroll? */\

  83.         "1:                                                 \n\t"\

  84.         "movq                      8(%%"REG_d"), %%mm0      \n\t" /* filterCoeff */\

  85.         "movq                (%%"REG_S", %%"REG_c", 2), %%mm2      \n\t" /* srcData */\

  86.         "movq               8(%%"REG_S", %%"REG_c", 2), %%mm5      \n\t" /* srcData */\

  87.         "add                                $16, %%"REG_d"  \n\t"\

  88.         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\

  89.         "test                         %%"REG_S", %%"REG_S"  \n\t"\

  90.         "pmulhw                           %%mm0, %%mm2      \n\t"\

  91.         "pmulhw                           %%mm0, %%mm5      \n\t"\

  92.         "paddw                            %%mm2, %%mm3      \n\t"\

  93.         "paddw                            %%mm5, %%mm4      \n\t"\

  94.         " jnz                                1b             \n\t"\

  95.         "psraw                               $3, %%mm3      \n\t"\

  96.         "psraw                               $3, %%mm4      \n\t"\

  97.         "packuswb                         %%mm4, %%mm3      \n\t"

  98.         MOVNTQ2 "                         %%mm3, (%1, %%"REG_c")\n\t"

  99.         "add                          $8, %%"REG_c"         \n\t"\

  100.         "cmp                          %2, %%"REG_c"         \n\t"\

  101.         "movq    %%mm6, %%mm3\n\t"

  102.         "movq    %%mm7, %%mm4\n\t"

  103.         "mov                                 %0, %%"REG_d"  \n\t"\

  104.         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\

  105.         "jb                                  1b             \n\t"\

  106.         :: "g" (filter),

  107.            "r" (dest-offset), "g" ((x86_reg)(dstW+offset)), "m" (offset)

  108.         : "%"REG_d, "%"REG_S, "%"REG_c

  109.     );

  110. }

  111. 

  112. #define YSCALEYUV2PACKEDX_UV \

  113.     __asm__ volatile(\

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

  115.         ".p2align                      4                \n\t"\

  116.         "nop                                            \n\t"\

  117.         "1:                                             \n\t"\

  118.         "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"     \n\t"\

  119.         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\

  120.         "movq      "VROUNDER_OFFSET"(%0), %%mm3         \n\t"\

  121.         "movq                      %%mm3, %%mm4         \n\t"\

  122.         ".p2align                      4                \n\t"\

  123.         "2:                                             \n\t"\

  124.         "movq               8(%%"REG_d"), %%mm0         \n\t" /* filterCoeff */\

  125.         "movq     (%%"REG_S", %%"REG_a"), %%mm2         \n\t" /* UsrcData */\

  126.         "add                          %6, %%"REG_S"     \n\t" \

  127.         "movq     (%%"REG_S", %%"REG_a"), %%mm5         \n\t" /* VsrcData */\

  128.         "add                         $16, %%"REG_d"     \n\t"\

  129.         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\

  130.         "pmulhw                    %%mm0, %%mm2         \n\t"\

  131.         "pmulhw                    %%mm0, %%mm5         \n\t"\

  132.         "paddw                     %%mm2, %%mm3         \n\t"\

  133.         "paddw                     %%mm5, %%mm4         \n\t"\

  134.         "test                  %%"REG_S", %%"REG_S"     \n\t"\

  135.         " jnz                         2b                \n\t"\

  136. 

  137. #define YSCALEYUV2PACKEDX_YA(offset,coeff,src1,src2,dst1,dst2) \

  138.     "lea                "offset"(%0), %%"REG_d"     \n\t"\

  139.     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\

  140.     "movq      "VROUNDER_OFFSET"(%0), "#dst1"       \n\t"\

  141.     "movq                    "#dst1", "#dst2"       \n\t"\

  142.     ".p2align                      4                \n\t"\

  143.     "2:                                             \n\t"\

  144.     "movq               8(%%"REG_d"), "#coeff"      \n\t" /* filterCoeff */\

  145.     "movq  (%%"REG_S", %%"REG_a", 2), "#src1"       \n\t" /* Y1srcData */\

  146.     "movq 8(%%"REG_S", %%"REG_a", 2), "#src2"       \n\t" /* Y2srcData */\

  147.     "add                         $16, %%"REG_d"            \n\t"\

  148.     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\

  149.     "pmulhw                 "#coeff", "#src1"       \n\t"\

  150.     "pmulhw                 "#coeff", "#src2"       \n\t"\

  151.     "paddw                   "#src1", "#dst1"       \n\t"\

  152.     "paddw                   "#src2", "#dst2"       \n\t"\

  153.     "test                  %%"REG_S", %%"REG_S"     \n\t"\

  154.     " jnz                         2b                \n\t"\

  155. 

  156. #define YSCALEYUV2PACKEDX \

  157.     YSCALEYUV2PACKEDX_UV \

  158.     YSCALEYUV2PACKEDX_YA(LUM_MMX_FILTER_OFFSET,%%mm0,%%mm2,%%mm5,%%mm1,%%mm7) \

  159. 

  160. #define YSCALEYUV2PACKEDX_END                     \

  161.         :: "r" (&c->redDither),                   \

  162.             "m" (dummy), "m" (dummy), "m" (dummy),\

  163.             "r" (dest), "m" (dstW_reg), "m"(uv_off) \

  164.         : "%"REG_a, "%"REG_d, "%"REG_S            \

  165.     );

  166. 

  167. #define YSCALEYUV2PACKEDX_ACCURATE_UV \

  168.     __asm__ volatile(\

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

  170.         ".p2align                      4                \n\t"\

  171.         "nop                                            \n\t"\

  172.         "1:                                             \n\t"\

  173.         "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"     \n\t"\

  174.         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\

  175.         "pxor                      %%mm4, %%mm4         \n\t"\

  176.         "pxor                      %%mm5, %%mm5         \n\t"\

  177.         "pxor                      %%mm6, %%mm6         \n\t"\

  178.         "pxor                      %%mm7, %%mm7         \n\t"\

  179.         ".p2align                      4                \n\t"\

  180.         "2:                                             \n\t"\

  181.         "movq     (%%"REG_S", %%"REG_a"), %%mm0         \n\t" /* UsrcData */\

  182.         "add                          %6, %%"REG_S"      \n\t" \

  183.         "movq     (%%"REG_S", %%"REG_a"), %%mm2         \n\t" /* VsrcData */\

  184.         "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S"     \n\t"\

  185.         "movq     (%%"REG_S", %%"REG_a"), %%mm1         \n\t" /* UsrcData */\

  186.         "movq                      %%mm0, %%mm3         \n\t"\

  187.         "punpcklwd                 %%mm1, %%mm0         \n\t"\

  188.         "punpckhwd                 %%mm1, %%mm3         \n\t"\

  189.         "movq "STR(APCK_COEF)"(%%"REG_d"),%%mm1         \n\t" /* filterCoeff */\

  190.         "pmaddwd                   %%mm1, %%mm0         \n\t"\

  191.         "pmaddwd                   %%mm1, %%mm3         \n\t"\

  192.         "paddd                     %%mm0, %%mm4         \n\t"\

  193.         "paddd                     %%mm3, %%mm5         \n\t"\

  194.         "add                          %6, %%"REG_S"      \n\t" \

  195.         "movq     (%%"REG_S", %%"REG_a"), %%mm3         \n\t" /* VsrcData */\

  196.         "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S"     \n\t"\

  197.         "add           $"STR(APCK_SIZE)", %%"REG_d"     \n\t"\

  198.         "test                  %%"REG_S", %%"REG_S"     \n\t"\

  199.         "movq                      %%mm2, %%mm0         \n\t"\

  200.         "punpcklwd                 %%mm3, %%mm2         \n\t"\

  201.         "punpckhwd                 %%mm3, %%mm0         \n\t"\

  202.         "pmaddwd                   %%mm1, %%mm2         \n\t"\

  203.         "pmaddwd                   %%mm1, %%mm0         \n\t"\

  204.         "paddd                     %%mm2, %%mm6         \n\t"\

  205.         "paddd                     %%mm0, %%mm7         \n\t"\

  206.         " jnz                         2b                \n\t"\

  207.         "psrad                       $16, %%mm4         \n\t"\

  208.         "psrad                       $16, %%mm5         \n\t"\

  209.         "psrad                       $16, %%mm6         \n\t"\

  210.         "psrad                       $16, %%mm7         \n\t"\

  211.         "movq      "VROUNDER_OFFSET"(%0), %%mm0         \n\t"\

  212.         "packssdw                  %%mm5, %%mm4         \n\t"\

  213.         "packssdw                  %%mm7, %%mm6         \n\t"\

  214.         "paddw                     %%mm0, %%mm4         \n\t"\

  215.         "paddw                     %%mm0, %%mm6         \n\t"\

  216.         "movq                      %%mm4, "U_TEMP"(%0)  \n\t"\

  217.         "movq                      %%mm6, "V_TEMP"(%0)  \n\t"\

  218. 

  219. #define YSCALEYUV2PACKEDX_ACCURATE_YA(offset) \

  220.     "lea                "offset"(%0), %%"REG_d"     \n\t"\

  221.     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\

  222.     "pxor                      %%mm1, %%mm1         \n\t"\

  223.     "pxor                      %%mm5, %%mm5         \n\t"\

  224.     "pxor                      %%mm7, %%mm7         \n\t"\

  225.     "pxor                      %%mm6, %%mm6         \n\t"\

  226.     ".p2align                      4                \n\t"\

  227.     "2:                                             \n\t"\

  228.     "movq  (%%"REG_S", %%"REG_a", 2), %%mm0         \n\t" /* Y1srcData */\

  229.     "movq 8(%%"REG_S", %%"REG_a", 2), %%mm2         \n\t" /* Y2srcData */\

  230.     "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S"     \n\t"\

  231.     "movq  (%%"REG_S", %%"REG_a", 2), %%mm4         \n\t" /* Y1srcData */\

  232.     "movq                      %%mm0, %%mm3         \n\t"\

  233.     "punpcklwd                 %%mm4, %%mm0         \n\t"\

  234.     "punpckhwd                 %%mm4, %%mm3         \n\t"\

  235.     "movq "STR(APCK_COEF)"(%%"REG_d"), %%mm4         \n\t" /* filterCoeff */\

  236.     "pmaddwd                   %%mm4, %%mm0         \n\t"\

  237.     "pmaddwd                   %%mm4, %%mm3         \n\t"\

  238.     "paddd                     %%mm0, %%mm1         \n\t"\

  239.     "paddd                     %%mm3, %%mm5         \n\t"\

  240.     "movq 8(%%"REG_S", %%"REG_a", 2), %%mm3         \n\t" /* Y2srcData */\

  241.     "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S"     \n\t"\

  242.     "add           $"STR(APCK_SIZE)", %%"REG_d"     \n\t"\

  243.     "test                  %%"REG_S", %%"REG_S"     \n\t"\

  244.     "movq                      %%mm2, %%mm0         \n\t"\

  245.     "punpcklwd                 %%mm3, %%mm2         \n\t"\

  246.     "punpckhwd                 %%mm3, %%mm0         \n\t"\

  247.     "pmaddwd                   %%mm4, %%mm2         \n\t"\

  248.     "pmaddwd                   %%mm4, %%mm0         \n\t"\

  249.     "paddd                     %%mm2, %%mm7         \n\t"\

  250.     "paddd                     %%mm0, %%mm6         \n\t"\

  251.     " jnz                         2b                \n\t"\

  252.     "psrad                       $16, %%mm1         \n\t"\

  253.     "psrad                       $16, %%mm5         \n\t"\

  254.     "psrad                       $16, %%mm7         \n\t"\

  255.     "psrad                       $16, %%mm6         \n\t"\

  256.     "movq      "VROUNDER_OFFSET"(%0), %%mm0         \n\t"\

  257.     "packssdw                  %%mm5, %%mm1         \n\t"\

  258.     "packssdw                  %%mm6, %%mm7         \n\t"\

  259.     "paddw                     %%mm0, %%mm1         \n\t"\

  260.     "paddw                     %%mm0, %%mm7         \n\t"\

  261.     "movq               "U_TEMP"(%0), %%mm3         \n\t"\

  262.     "movq               "V_TEMP"(%0), %%mm4         \n\t"\

  263. 

  264. #define YSCALEYUV2PACKEDX_ACCURATE \

  265.     YSCALEYUV2PACKEDX_ACCURATE_UV \

  266.     YSCALEYUV2PACKEDX_ACCURATE_YA(LUM_MMX_FILTER_OFFSET)

  267. 

  268. #define YSCALEYUV2RGBX \

  269.     "psubw  "U_OFFSET"(%0), %%mm3       \n\t" /* (U-128)8*/\

  270.     "psubw  "V_OFFSET"(%0), %%mm4       \n\t" /* (V-128)8*/\

  271.     "movq            %%mm3, %%mm2       \n\t" /* (U-128)8*/\

  272.     "movq            %%mm4, %%mm5       \n\t" /* (V-128)8*/\

  273.     "pmulhw "UG_COEFF"(%0), %%mm3       \n\t"\

  274.     "pmulhw "VG_COEFF"(%0), %%mm4       \n\t"\

  275.     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\

  276.     "pmulhw "UB_COEFF"(%0), %%mm2       \n\t"\

  277.     "pmulhw "VR_COEFF"(%0), %%mm5       \n\t"\

  278.     "psubw  "Y_OFFSET"(%0), %%mm1       \n\t" /* 8(Y-16)*/\

  279.     "psubw  "Y_OFFSET"(%0), %%mm7       \n\t" /* 8(Y-16)*/\

  280.     "pmulhw  "Y_COEFF"(%0), %%mm1       \n\t"\

  281.     "pmulhw  "Y_COEFF"(%0), %%mm7       \n\t"\

  282.     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\

  283.     "paddw           %%mm3, %%mm4       \n\t"\

  284.     "movq            %%mm2, %%mm0       \n\t"\

  285.     "movq            %%mm5, %%mm6       \n\t"\

  286.     "movq            %%mm4, %%mm3       \n\t"\

  287.     "punpcklwd       %%mm2, %%mm2       \n\t"\

  288.     "punpcklwd       %%mm5, %%mm5       \n\t"\

  289.     "punpcklwd       %%mm4, %%mm4       \n\t"\

  290.     "paddw           %%mm1, %%mm2       \n\t"\

  291.     "paddw           %%mm1, %%mm5       \n\t"\

  292.     "paddw           %%mm1, %%mm4       \n\t"\

  293.     "punpckhwd       %%mm0, %%mm0       \n\t"\

  294.     "punpckhwd       %%mm6, %%mm6       \n\t"\

  295.     "punpckhwd       %%mm3, %%mm3       \n\t"\

  296.     "paddw           %%mm7, %%mm0       \n\t"\

  297.     "paddw           %%mm7, %%mm6       \n\t"\

  298.     "paddw           %%mm7, %%mm3       \n\t"\

  299.     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\

  300.     "packuswb        %%mm0, %%mm2       \n\t"\

  301.     "packuswb        %%mm6, %%mm5       \n\t"\

  302.     "packuswb        %%mm3, %%mm4       \n\t"\

  303. 

  304. #define REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) \

  305.     "movq       "#b", "#q2"     \n\t" /* B */\

  306.     "movq       "#r", "#t"      \n\t" /* R */\

  307.     "punpcklbw  "#g", "#b"      \n\t" /* GBGBGBGB 0 */\

  308.     "punpcklbw  "#a", "#r"      \n\t" /* ARARARAR 0 */\

  309.     "punpckhbw  "#g", "#q2"     \n\t" /* GBGBGBGB 2 */\

  310.     "punpckhbw  "#a", "#t"      \n\t" /* ARARARAR 2 */\

  311.     "movq       "#b", "#q0"     \n\t" /* GBGBGBGB 0 */\

  312.     "movq      "#q2", "#q3"     \n\t" /* GBGBGBGB 2 */\

  313.     "punpcklwd  "#r", "#q0"     \n\t" /* ARGBARGB 0 */\

  314.     "punpckhwd  "#r", "#b"      \n\t" /* ARGBARGB 1 */\

  315.     "punpcklwd  "#t", "#q2"     \n\t" /* ARGBARGB 2 */\

  316.     "punpckhwd  "#t", "#q3"     \n\t" /* ARGBARGB 3 */\

  317. \

  318.     MOVNTQ(   q0,   (dst, index, 4))\

  319.     MOVNTQ(    b,  8(dst, index, 4))\

  320.     MOVNTQ(   q2, 16(dst, index, 4))\

  321.     MOVNTQ(   q3, 24(dst, index, 4))\

  322. \

  323.     "add      $8, "#index"      \n\t"\

  324.     "cmp "#dstw", "#index"      \n\t"\

  325.     " jb      1b                \n\t"

  326. #define WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t)  REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t)

  327. 

  328. static void RENAME(yuv2rgb32_X_ar)(SwsContext *c, const int16_t *lumFilter,

  329.                                    const int16_t **lumSrc, int lumFilterSize,

  330.                                    const int16_t *chrFilter, const int16_t **chrUSrc,

  331.                                    const int16_t **chrVSrc,

  332.                                    int chrFilterSize, const int16_t **alpSrc,

  333.                                    uint8_t *dest, int dstW, int dstY)

  334. {

  335.     x86_reg dummy=0;

  336.     x86_reg dstW_reg = dstW;

  337.     x86_reg uv_off = c->uv_offx2;

  338. 

  339.     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  340.         YSCALEYUV2PACKEDX_ACCURATE

  341.         YSCALEYUV2RGBX

  342.         "movq                      %%mm2, "U_TEMP"(%0)  \n\t"

  343.         "movq                      %%mm4, "V_TEMP"(%0)  \n\t"

  344.         "movq                      %%mm5, "Y_TEMP"(%0)  \n\t"

  345.         YSCALEYUV2PACKEDX_ACCURATE_YA(ALP_MMX_FILTER_OFFSET)

  346.         "movq               "Y_TEMP"(%0), %%mm5         \n\t"

  347.         "psraw                        $3, %%mm1         \n\t"

  348.         "psraw                        $3, %%mm7         \n\t"

  349.         "packuswb                  %%mm7, %%mm1         \n\t"

  350.         WRITEBGR32(%4, %5, %%REGa, %%mm3, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm2, %%mm6)

  351.         YSCALEYUV2PACKEDX_END

  352.     } else {

  353.         YSCALEYUV2PACKEDX_ACCURATE

  354.         YSCALEYUV2RGBX

  355.         "pcmpeqd %%mm7, %%mm7 \n\t"

  356.         WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)

  357.         YSCALEYUV2PACKEDX_END

  358.     }

  359. }

  360. 

  361. static void RENAME(yuv2rgb32_X)(SwsContext *c, const int16_t *lumFilter,

  362.                                 const int16_t **lumSrc, int lumFilterSize,

  363.                                 const int16_t *chrFilter, const int16_t **chrUSrc,

  364.                                 const int16_t **chrVSrc,

  365.                                 int chrFilterSize, const int16_t **alpSrc,

  366.                                 uint8_t *dest, int dstW, int dstY)

  367. {

  368.     x86_reg dummy=0;

  369.     x86_reg dstW_reg = dstW;

  370.     x86_reg uv_off = c->uv_offx2;

  371. 

  372.     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  373.         YSCALEYUV2PACKEDX

  374.         YSCALEYUV2RGBX

  375.         YSCALEYUV2PACKEDX_YA(ALP_MMX_FILTER_OFFSET, %%mm0, %%mm3, %%mm6, %%mm1, %%mm7)

  376.         "psraw                        $3, %%mm1         \n\t"

  377.         "psraw                        $3, %%mm7         \n\t"

  378.         "packuswb                  %%mm7, %%mm1         \n\t"

  379.         WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)

  380.         YSCALEYUV2PACKEDX_END

  381.     } else {

  382.         YSCALEYUV2PACKEDX

  383.         YSCALEYUV2RGBX

  384.         "pcmpeqd %%mm7, %%mm7 \n\t"

  385.         WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)

  386.         YSCALEYUV2PACKEDX_END

  387.     }

  388. }

  389. 

  390. #define REAL_WRITERGB16(dst, dstw, index) \

  391.     "pand "MANGLE(bF8)", %%mm2  \n\t" /* B */\

  392.     "pand "MANGLE(bFC)", %%mm4  \n\t" /* G */\

  393.     "pand "MANGLE(bF8)", %%mm5  \n\t" /* R */\

  394.     "psrlq           $3, %%mm2  \n\t"\

  395. \

  396.     "movq         %%mm2, %%mm1  \n\t"\

  397.     "movq         %%mm4, %%mm3  \n\t"\

  398. \

  399.     "punpcklbw    %%mm7, %%mm3  \n\t"\

  400.     "punpcklbw    %%mm5, %%mm2  \n\t"\

  401.     "punpckhbw    %%mm7, %%mm4  \n\t"\

  402.     "punpckhbw    %%mm5, %%mm1  \n\t"\

  403. \

  404.     "psllq           $3, %%mm3  \n\t"\

  405.     "psllq           $3, %%mm4  \n\t"\

  406. \

  407.     "por          %%mm3, %%mm2  \n\t"\

  408.     "por          %%mm4, %%mm1  \n\t"\

  409. \

  410.     MOVNTQ(%%mm2,  (dst, index, 2))\

  411.     MOVNTQ(%%mm1, 8(dst, index, 2))\

  412. \

  413.     "add             $8, "#index"   \n\t"\

  414.     "cmp        "#dstw", "#index"   \n\t"\

  415.     " jb             1b             \n\t"

  416. #define WRITERGB16(dst, dstw, index)  REAL_WRITERGB16(dst, dstw, index)

  417. 

  418. static void RENAME(yuv2rgb565_X_ar)(SwsContext *c, const int16_t *lumFilter,

  419.                                     const int16_t **lumSrc, int lumFilterSize,

  420.                                     const int16_t *chrFilter, const int16_t **chrUSrc,

  421.                                     const int16_t **chrVSrc,

  422.                                     int chrFilterSize, const int16_t **alpSrc,

  423.                                     uint8_t *dest, int dstW, int dstY)

  424. {

  425.     x86_reg dummy=0;

  426.     x86_reg dstW_reg = dstW;

  427.     x86_reg uv_off = c->uv_offx2;

  428. 

  429.     YSCALEYUV2PACKEDX_ACCURATE

  430.     YSCALEYUV2RGBX

  431.     "pxor %%mm7, %%mm7 \n\t"

  432.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  433. #ifdef DITHER1XBPP

  434.     "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"

  435.     "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"

  436.     "paddusb "RED_DITHER"(%0), %%mm5\n\t"

  437. #endif

  438.     WRITERGB16(%4, %5, %%REGa)

  439.     YSCALEYUV2PACKEDX_END

  440. }

  441. 

  442. static void RENAME(yuv2rgb565_X)(SwsContext *c, const int16_t *lumFilter,

  443.                                  const int16_t **lumSrc, int lumFilterSize,

  444.                                  const int16_t *chrFilter, const int16_t **chrUSrc,

  445.                                  const int16_t **chrVSrc,

  446.                                  int chrFilterSize, const int16_t **alpSrc,

  447.                                  uint8_t *dest, int dstW, int dstY)

  448. {

  449.     x86_reg dummy=0;

  450.     x86_reg dstW_reg = dstW;

  451.     x86_reg uv_off = c->uv_offx2;

  452. 

  453.     YSCALEYUV2PACKEDX

  454.     YSCALEYUV2RGBX

  455.     "pxor %%mm7, %%mm7 \n\t"

  456.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  457. #ifdef DITHER1XBPP

  458.     "paddusb "BLUE_DITHER"(%0), %%mm2  \n\t"

  459.     "paddusb "GREEN_DITHER"(%0), %%mm4  \n\t"

  460.     "paddusb "RED_DITHER"(%0), %%mm5  \n\t"

  461. #endif

  462.     WRITERGB16(%4, %5, %%REGa)

  463.     YSCALEYUV2PACKEDX_END

  464. }

  465. 

  466. #define REAL_WRITERGB15(dst, dstw, index) \

  467.     "pand "MANGLE(bF8)", %%mm2  \n\t" /* B */\

  468.     "pand "MANGLE(bF8)", %%mm4  \n\t" /* G */\

  469.     "pand "MANGLE(bF8)", %%mm5  \n\t" /* R */\

  470.     "psrlq           $3, %%mm2  \n\t"\

  471.     "psrlq           $1, %%mm5  \n\t"\

  472. \

  473.     "movq         %%mm2, %%mm1  \n\t"\

  474.     "movq         %%mm4, %%mm3  \n\t"\

  475. \

  476.     "punpcklbw    %%mm7, %%mm3  \n\t"\

  477.     "punpcklbw    %%mm5, %%mm2  \n\t"\

  478.     "punpckhbw    %%mm7, %%mm4  \n\t"\

  479.     "punpckhbw    %%mm5, %%mm1  \n\t"\

  480. \

  481.     "psllq           $2, %%mm3  \n\t"\

  482.     "psllq           $2, %%mm4  \n\t"\

  483. \

  484.     "por          %%mm3, %%mm2  \n\t"\

  485.     "por          %%mm4, %%mm1  \n\t"\

  486. \

  487.     MOVNTQ(%%mm2,  (dst, index, 2))\

  488.     MOVNTQ(%%mm1, 8(dst, index, 2))\

  489. \

  490.     "add             $8, "#index"   \n\t"\

  491.     "cmp        "#dstw", "#index"   \n\t"\

  492.     " jb             1b             \n\t"

  493. #define WRITERGB15(dst, dstw, index)  REAL_WRITERGB15(dst, dstw, index)

  494. 

  495. static void RENAME(yuv2rgb555_X_ar)(SwsContext *c, const int16_t *lumFilter,

  496.                                     const int16_t **lumSrc, int lumFilterSize,

  497.                                     const int16_t *chrFilter, const int16_t **chrUSrc,

  498.                                     const int16_t **chrVSrc,

  499.                                     int chrFilterSize, const int16_t **alpSrc,

  500.                                     uint8_t *dest, int dstW, int dstY)

  501. {

  502.     x86_reg dummy=0;

  503.     x86_reg dstW_reg = dstW;

  504.     x86_reg uv_off = c->uv_offx2;

  505. 

  506.     YSCALEYUV2PACKEDX_ACCURATE

  507.     YSCALEYUV2RGBX

  508.     "pxor %%mm7, %%mm7 \n\t"

  509.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  510. #ifdef DITHER1XBPP

  511.     "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"

  512.     "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"

  513.     "paddusb "RED_DITHER"(%0), %%mm5\n\t"

  514. #endif

  515.     WRITERGB15(%4, %5, %%REGa)

  516.     YSCALEYUV2PACKEDX_END

  517. }

  518. 

  519. static void RENAME(yuv2rgb555_X)(SwsContext *c, const int16_t *lumFilter,

  520.                                  const int16_t **lumSrc, int lumFilterSize,

  521.                                  const int16_t *chrFilter, const int16_t **chrUSrc,

  522.                                  const int16_t **chrVSrc,

  523.                                  int chrFilterSize, const int16_t **alpSrc,

  524.                                  uint8_t *dest, int dstW, int dstY)

  525. {

  526.     x86_reg dummy=0;

  527.     x86_reg dstW_reg = dstW;

  528.     x86_reg uv_off = c->uv_offx2;

  529. 

  530.     YSCALEYUV2PACKEDX

  531.     YSCALEYUV2RGBX

  532.     "pxor %%mm7, %%mm7 \n\t"

  533.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  534. #ifdef DITHER1XBPP

  535.     "paddusb "BLUE_DITHER"(%0), %%mm2  \n\t"

  536.     "paddusb "GREEN_DITHER"(%0), %%mm4  \n\t"

  537.     "paddusb "RED_DITHER"(%0), %%mm5  \n\t"

  538. #endif

  539.     WRITERGB15(%4, %5, %%REGa)

  540.     YSCALEYUV2PACKEDX_END

  541. }

  542. 

  543. #define WRITEBGR24MMX(dst, dstw, index) \

  544.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\

  545.     "movq      %%mm2, %%mm1     \n\t" /* B */\

  546.     "movq      %%mm5, %%mm6     \n\t" /* R */\

  547.     "punpcklbw %%mm4, %%mm2     \n\t" /* GBGBGBGB 0 */\

  548.     "punpcklbw %%mm7, %%mm5     \n\t" /* 0R0R0R0R 0 */\

  549.     "punpckhbw %%mm4, %%mm1     \n\t" /* GBGBGBGB 2 */\

  550.     "punpckhbw %%mm7, %%mm6     \n\t" /* 0R0R0R0R 2 */\

  551.     "movq      %%mm2, %%mm0     \n\t" /* GBGBGBGB 0 */\

  552.     "movq      %%mm1, %%mm3     \n\t" /* GBGBGBGB 2 */\

  553.     "punpcklwd %%mm5, %%mm0     \n\t" /* 0RGB0RGB 0 */\

  554.     "punpckhwd %%mm5, %%mm2     \n\t" /* 0RGB0RGB 1 */\

  555.     "punpcklwd %%mm6, %%mm1     \n\t" /* 0RGB0RGB 2 */\

  556.     "punpckhwd %%mm6, %%mm3     \n\t" /* 0RGB0RGB 3 */\

  557. \

  558.     "movq      %%mm0, %%mm4     \n\t" /* 0RGB0RGB 0 */\

  559.     "movq      %%mm2, %%mm6     \n\t" /* 0RGB0RGB 1 */\

  560.     "movq      %%mm1, %%mm5     \n\t" /* 0RGB0RGB 2 */\

  561.     "movq      %%mm3, %%mm7     \n\t" /* 0RGB0RGB 3 */\

  562. \

  563.     "psllq       $40, %%mm0     \n\t" /* RGB00000 0 */\

  564.     "psllq       $40, %%mm2     \n\t" /* RGB00000 1 */\

  565.     "psllq       $40, %%mm1     \n\t" /* RGB00000 2 */\

  566.     "psllq       $40, %%mm3     \n\t" /* RGB00000 3 */\

  567. \

  568.     "punpckhdq %%mm4, %%mm0     \n\t" /* 0RGBRGB0 0 */\

  569.     "punpckhdq %%mm6, %%mm2     \n\t" /* 0RGBRGB0 1 */\

  570.     "punpckhdq %%mm5, %%mm1     \n\t" /* 0RGBRGB0 2 */\

  571.     "punpckhdq %%mm7, %%mm3     \n\t" /* 0RGBRGB0 3 */\

  572. \

  573.     "psrlq        $8, %%mm0     \n\t" /* 00RGBRGB 0 */\

  574.     "movq      %%mm2, %%mm6     \n\t" /* 0RGBRGB0 1 */\

  575.     "psllq       $40, %%mm2     \n\t" /* GB000000 1 */\

  576.     "por       %%mm2, %%mm0     \n\t" /* GBRGBRGB 0 */\

  577.     MOVNTQ(%%mm0, (dst))\

  578. \

  579.     "psrlq       $24, %%mm6     \n\t" /* 0000RGBR 1 */\

  580.     "movq      %%mm1, %%mm5     \n\t" /* 0RGBRGB0 2 */\

  581.     "psllq       $24, %%mm1     \n\t" /* BRGB0000 2 */\

  582.     "por       %%mm1, %%mm6     \n\t" /* BRGBRGBR 1 */\

  583.     MOVNTQ(%%mm6, 8(dst))\

  584. \

  585.     "psrlq       $40, %%mm5     \n\t" /* 000000RG 2 */\

  586.     "psllq        $8, %%mm3     \n\t" /* RGBRGB00 3 */\

  587.     "por       %%mm3, %%mm5     \n\t" /* RGBRGBRG 2 */\

  588.     MOVNTQ(%%mm5, 16(dst))\

  589. \

  590.     "add         $24, "#dst"    \n\t"\

  591. \

  592.     "add          $8, "#index"  \n\t"\

  593.     "cmp     "#dstw", "#index"  \n\t"\

  594.     " jb          1b            \n\t"

  595. 

  596. #define WRITEBGR24MMX2(dst, dstw, index) \

  597.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\

  598.     "movq "MANGLE(ff_M24A)", %%mm0 \n\t"\

  599.     "movq "MANGLE(ff_M24C)", %%mm7 \n\t"\

  600.     "pshufw $0x50, %%mm2, %%mm1 \n\t" /* B3 B2 B3 B2  B1 B0 B1 B0 */\

  601.     "pshufw $0x50, %%mm4, %%mm3 \n\t" /* G3 G2 G3 G2  G1 G0 G1 G0 */\

  602.     "pshufw $0x00, %%mm5, %%mm6 \n\t" /* R1 R0 R1 R0  R1 R0 R1 R0 */\

  603. \

  604.     "pand   %%mm0, %%mm1        \n\t" /*    B2        B1       B0 */\

  605.     "pand   %%mm0, %%mm3        \n\t" /*    G2        G1       G0 */\

  606.     "pand   %%mm7, %%mm6        \n\t" /*       R1        R0       */\

  607. \

  608.     "psllq     $8, %%mm3        \n\t" /* G2        G1       G0    */\

  609.     "por    %%mm1, %%mm6        \n\t"\

  610.     "por    %%mm3, %%mm6        \n\t"\

  611.     MOVNTQ(%%mm6, (dst))\

  612. \

  613.     "psrlq     $8, %%mm4        \n\t" /* 00 G7 G6 G5  G4 G3 G2 G1 */\

  614.     "pshufw $0xA5, %%mm2, %%mm1 \n\t" /* B5 B4 B5 B4  B3 B2 B3 B2 */\

  615.     "pshufw $0x55, %%mm4, %%mm3 \n\t" /* G4 G3 G4 G3  G4 G3 G4 G3 */\

  616.     "pshufw $0xA5, %%mm5, %%mm6 \n\t" /* R5 R4 R5 R4  R3 R2 R3 R2 */\

  617. \

  618.     "pand "MANGLE(ff_M24B)", %%mm1 \n\t" /* B5       B4        B3    */\

  619.     "pand   %%mm7, %%mm3        \n\t" /*       G4        G3       */\

  620.     "pand   %%mm0, %%mm6        \n\t" /*    R4        R3       R2 */\

  621. \

  622.     "por    %%mm1, %%mm3        \n\t" /* B5    G4 B4     G3 B3    */\

  623.     "por    %%mm3, %%mm6        \n\t"\

  624.     MOVNTQ(%%mm6, 8(dst))\

  625. \

  626.     "pshufw $0xFF, %%mm2, %%mm1 \n\t" /* B7 B6 B7 B6  B7 B6 B6 B7 */\

  627.     "pshufw $0xFA, %%mm4, %%mm3 \n\t" /* 00 G7 00 G7  G6 G5 G6 G5 */\

  628.     "pshufw $0xFA, %%mm5, %%mm6 \n\t" /* R7 R6 R7 R6  R5 R4 R5 R4 */\

  629. \

  630.     "pand   %%mm7, %%mm1        \n\t" /*       B7        B6       */\

  631.     "pand   %%mm0, %%mm3        \n\t" /*    G7        G6       G5 */\

  632.     "pand "MANGLE(ff_M24B)", %%mm6 \n\t" /* R7       R6        R5    */\

  633. \

  634.     "por    %%mm1, %%mm3        \n\t"\

  635.     "por    %%mm3, %%mm6        \n\t"\

  636.     MOVNTQ(%%mm6, 16(dst))\

  637. \

  638.     "add      $24, "#dst"       \n\t"\

  639. \

  640.     "add       $8, "#index"     \n\t"\

  641.     "cmp  "#dstw", "#index"     \n\t"\

  642.     " jb       1b               \n\t"

  643. 

  644. #if COMPILE_TEMPLATE_MMX2

  645. #undef WRITEBGR24

  646. #define WRITEBGR24(dst, dstw, index)  WRITEBGR24MMX2(dst, dstw, index)

  647. #else

  648. #undef WRITEBGR24

  649. #define WRITEBGR24(dst, dstw, index)  WRITEBGR24MMX(dst, dstw, index)

  650. #endif

  651. 

  652. static void RENAME(yuv2bgr24_X_ar)(SwsContext *c, const int16_t *lumFilter,

  653.                                    const int16_t **lumSrc, int lumFilterSize,

  654.                                    const int16_t *chrFilter, const int16_t **chrUSrc,

  655.                                    const int16_t **chrVSrc,

  656.                                    int chrFilterSize, const int16_t **alpSrc,

  657.                                    uint8_t *dest, int dstW, int dstY)

  658. {

  659.     x86_reg dummy=0;

  660.     x86_reg dstW_reg = dstW;

  661.     x86_reg uv_off = c->uv_offx2;

  662. 

  663.     YSCALEYUV2PACKEDX_ACCURATE

  664.     YSCALEYUV2RGBX

  665.     "pxor %%mm7, %%mm7 \n\t"

  666.     "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"\n\t" //FIXME optimize

  667.     "add %4, %%"REG_c"                        \n\t"

  668.     WRITEBGR24(%%REGc, %5, %%REGa)

  669.     :: "r" (&c->redDither),

  670.        "m" (dummy), "m" (dummy), "m" (dummy),

  671.        "r" (dest), "m" (dstW_reg), "m"(uv_off)

  672.     : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S

  673.     );

  674. }

  675. 

  676. static void RENAME(yuv2bgr24_X)(SwsContext *c, const int16_t *lumFilter,

  677.                                 const int16_t **lumSrc, int lumFilterSize,

  678.                                 const int16_t *chrFilter, const int16_t **chrUSrc,

  679.                                 const int16_t **chrVSrc,

  680.                                 int chrFilterSize, const int16_t **alpSrc,

  681.                                 uint8_t *dest, int dstW, int dstY)

  682. {

  683.     x86_reg dummy=0;

  684.     x86_reg dstW_reg = dstW;

  685.     x86_reg uv_off = c->uv_offx2;

  686. 

  687.     YSCALEYUV2PACKEDX

  688.     YSCALEYUV2RGBX

  689.     "pxor                    %%mm7, %%mm7       \n\t"

  690.     "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"   \n\t" //FIXME optimize

  691.     "add                        %4, %%"REG_c"   \n\t"

  692.     WRITEBGR24(%%REGc, %5, %%REGa)

  693.     :: "r" (&c->redDither),

  694.        "m" (dummy), "m" (dummy), "m" (dummy),

  695.        "r" (dest),  "m" (dstW_reg), "m"(uv_off)

  696.     : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S

  697.     );

  698. }

  699. 

  700. #define REAL_WRITEYUY2(dst, dstw, index) \

  701.     "packuswb  %%mm3, %%mm3     \n\t"\

  702.     "packuswb  %%mm4, %%mm4     \n\t"\

  703.     "packuswb  %%mm7, %%mm1     \n\t"\

  704.     "punpcklbw %%mm4, %%mm3     \n\t"\

  705.     "movq      %%mm1, %%mm7     \n\t"\

  706.     "punpcklbw %%mm3, %%mm1     \n\t"\

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

  708. \

  709.     MOVNTQ(%%mm1, (dst, index, 2))\

  710.     MOVNTQ(%%mm7, 8(dst, index, 2))\

  711. \

  712.     "add          $8, "#index"  \n\t"\

  713.     "cmp     "#dstw", "#index"  \n\t"\

  714.     " jb          1b            \n\t"

  715. #define WRITEYUY2(dst, dstw, index)  REAL_WRITEYUY2(dst, dstw, index)

  716. 

  717. static void RENAME(yuv2yuyv422_X_ar)(SwsContext *c, const int16_t *lumFilter,

  718.                                      const int16_t **lumSrc, int lumFilterSize,

  719.                                      const int16_t *chrFilter, const int16_t **chrUSrc,

  720.                                      const int16_t **chrVSrc,

  721.                                      int chrFilterSize, const int16_t **alpSrc,

  722.                                      uint8_t *dest, int dstW, int dstY)

  723. {

  724.     x86_reg dummy=0;

  725.     x86_reg dstW_reg = dstW;

  726.     x86_reg uv_off = c->uv_offx2;

  727. 

  728.     YSCALEYUV2PACKEDX_ACCURATE

  729.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  730.     "psraw $3, %%mm3    \n\t"

  731.     "psraw $3, %%mm4    \n\t"

  732.     "psraw $3, %%mm1    \n\t"

  733.     "psraw $3, %%mm7    \n\t"

  734.     WRITEYUY2(%4, %5, %%REGa)

  735.     YSCALEYUV2PACKEDX_END

  736. }

  737. 

  738. static void RENAME(yuv2yuyv422_X)(SwsContext *c, const int16_t *lumFilter,

  739.                                   const int16_t **lumSrc, int lumFilterSize,

  740.                                   const int16_t *chrFilter, const int16_t **chrUSrc,

  741.                                   const int16_t **chrVSrc,

  742.                                   int chrFilterSize, const int16_t **alpSrc,

  743.                                   uint8_t *dest, int dstW, int dstY)

  744. {

  745.     x86_reg dummy=0;

  746.     x86_reg dstW_reg = dstW;

  747.     x86_reg uv_off = c->uv_offx2;

  748. 

  749.     YSCALEYUV2PACKEDX

  750.     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  751.     "psraw $3, %%mm3    \n\t"

  752.     "psraw $3, %%mm4    \n\t"

  753.     "psraw $3, %%mm1    \n\t"

  754.     "psraw $3, %%mm7    \n\t"

  755.     WRITEYUY2(%4, %5, %%REGa)

  756.     YSCALEYUV2PACKEDX_END

  757. }

  758. 

  759. #define REAL_YSCALEYUV2RGB_UV(index, c) \

  760.     "xor            "#index", "#index"  \n\t"\

  761.     ".p2align              4            \n\t"\

  762.     "1:                                 \n\t"\

  763.     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\

  764.     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\

  765.     "add           "UV_OFFx2"("#c"), "#index"  \n\t" \

  766.     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\

  767.     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\

  768.     "sub           "UV_OFFx2"("#c"), "#index"  \n\t" \

  769.     "psubw             %%mm3, %%mm2     \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\

  770.     "psubw             %%mm4, %%mm5     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\

  771.     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0    \n\t"\

  772.     "pmulhw            %%mm0, %%mm2     \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\

  773.     "pmulhw            %%mm0, %%mm5     \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\

  774.     "psraw                $4, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\

  775.     "psraw                $4, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\

  776.     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\

  777.     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\

  778.     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\

  779.     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\

  780.     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\

  781.     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\

  782.     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\

  783.     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\

  784.     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\

  785. 

  786. #define REAL_YSCALEYUV2RGB_YA(index, c, b1, b2) \

  787.     "movq  ("#b1", "#index", 2), %%mm0     \n\t" /*buf0[eax]*/\

  788.     "movq  ("#b2", "#index", 2), %%mm1     \n\t" /*buf1[eax]*/\

  789.     "movq 8("#b1", "#index", 2), %%mm6     \n\t" /*buf0[eax]*/\

  790.     "movq 8("#b2", "#index", 2), %%mm7     \n\t" /*buf1[eax]*/\

  791.     "psubw             %%mm1, %%mm0     \n\t" /* buf0[eax] - buf1[eax]*/\

  792.     "psubw             %%mm7, %%mm6     \n\t" /* buf0[eax] - buf1[eax]*/\

  793.     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\

  794.     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\

  795.     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  796.     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  797.     "paddw             %%mm0, %%mm1     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\

  798.     "paddw             %%mm6, %%mm7     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\

  799. 

  800. #define REAL_YSCALEYUV2RGB_COEFF(c) \

  801.     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\

  802.     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\

  803.     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\

  804.     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\

  805.     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\

  806.     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\

  807.     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\

  808.     "paddw             %%mm3, %%mm4     \n\t"\

  809.     "movq              %%mm2, %%mm0     \n\t"\

  810.     "movq              %%mm5, %%mm6     \n\t"\

  811.     "movq              %%mm4, %%mm3     \n\t"\

  812.     "punpcklwd         %%mm2, %%mm2     \n\t"\

  813.     "punpcklwd         %%mm5, %%mm5     \n\t"\

  814.     "punpcklwd         %%mm4, %%mm4     \n\t"\

  815.     "paddw             %%mm1, %%mm2     \n\t"\

  816.     "paddw             %%mm1, %%mm5     \n\t"\

  817.     "paddw             %%mm1, %%mm4     \n\t"\

  818.     "punpckhwd         %%mm0, %%mm0     \n\t"\

  819.     "punpckhwd         %%mm6, %%mm6     \n\t"\

  820.     "punpckhwd         %%mm3, %%mm3     \n\t"\

  821.     "paddw             %%mm7, %%mm0     \n\t"\

  822.     "paddw             %%mm7, %%mm6     \n\t"\

  823.     "paddw             %%mm7, %%mm3     \n\t"\

  824.     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\

  825.     "packuswb          %%mm0, %%mm2     \n\t"\

  826.     "packuswb          %%mm6, %%mm5     \n\t"\

  827.     "packuswb          %%mm3, %%mm4     \n\t"\

  828. 

  829. #define YSCALEYUV2RGB_YA(index, c, b1, b2) REAL_YSCALEYUV2RGB_YA(index, c, b1, b2)

  830. 

  831. #define YSCALEYUV2RGB(index, c) \

  832.     REAL_YSCALEYUV2RGB_UV(index, c) \

  833.     REAL_YSCALEYUV2RGB_YA(index, c, %0, %1) \

  834.     REAL_YSCALEYUV2RGB_COEFF(c)

  835. 

  836. /**

  837.  * vertical bilinear scale YV12 to RGB

  838.  */

  839. static void RENAME(yuv2rgb32_2)(SwsContext *c, const int16_t *buf[2],

  840.                                 const int16_t *ubuf[2], const int16_t *vbuf[2],

  841.                                 const int16_t *abuf[2], uint8_t *dest,

  842.                                 int dstW, int yalpha, int uvalpha, int y)

  843. {

  844.     const int16_t *buf0  = buf[0],  *buf1  = buf[1],

  845.                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  846. 

  847.     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  848.         const int16_t *abuf0 = abuf[0], *abuf1 = abuf[1];

  849. #if ARCH_X86_64

  850.         __asm__ volatile(

  851.             YSCALEYUV2RGB(%%r8, %5)

  852.             YSCALEYUV2RGB_YA(%%r8, %5, %6, %7)

  853.             "psraw                  $3, %%mm1       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/

  854.             "psraw                  $3, %%mm7       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/

  855.             "packuswb            %%mm7, %%mm1       \n\t"

  856.             WRITEBGR32(%4, 8280(%5), %%r8, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)

  857.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "r" (dest),

  858.                "a" (&c->redDither),

  859.                "r" (abuf0), "r" (abuf1)

  860.             : "%r8"

  861.         );

  862. #else

  863.         c->u_temp=(intptr_t)abuf0;

  864.         c->v_temp=(intptr_t)abuf1;

  865.         __asm__ volatile(

  866.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  867.             "mov        %4, %%"REG_b"               \n\t"

  868.             "push %%"REG_BP"                        \n\t"

  869.             YSCALEYUV2RGB(%%REGBP, %5)

  870.             "push                   %0              \n\t"

  871.             "push                   %1              \n\t"

  872.             "mov          "U_TEMP"(%5), %0          \n\t"

  873.             "mov          "V_TEMP"(%5), %1          \n\t"

  874.             YSCALEYUV2RGB_YA(%%REGBP, %5, %0, %1)

  875.             "psraw                  $3, %%mm1       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/

  876.             "psraw                  $3, %%mm7       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/

  877.             "packuswb            %%mm7, %%mm1       \n\t"

  878.             "pop                    %1              \n\t"

  879.             "pop                    %0              \n\t"

  880.             WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)

  881.             "pop %%"REG_BP"                         \n\t"

  882.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  883.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  884.                "a" (&c->redDither)

  885.         );

  886. #endif

  887.     } else {

  888.         __asm__ volatile(

  889.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  890.             "mov        %4, %%"REG_b"               \n\t"

  891.             "push %%"REG_BP"                        \n\t"

  892.             YSCALEYUV2RGB(%%REGBP, %5)

  893.             "pcmpeqd %%mm7, %%mm7                   \n\t"

  894.             WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)

  895.             "pop %%"REG_BP"                         \n\t"

  896.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  897.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  898.                "a" (&c->redDither)

  899.         );

  900.     }

  901. }

  902. 

  903. static void RENAME(yuv2bgr24_2)(SwsContext *c, const int16_t *buf[2],

  904.                                 const int16_t *ubuf[2], const int16_t *vbuf[2],

  905.                                 const int16_t *abuf[2], uint8_t *dest,

  906.                                 int dstW, int yalpha, int uvalpha, int y)

  907. {

  908.     const int16_t *buf0  = buf[0],  *buf1  = buf[1],

  909.                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  910. 

  911.     //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(

  912.     __asm__ volatile(

  913.         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  914.         "mov        %4, %%"REG_b"               \n\t"

  915.         "push %%"REG_BP"                        \n\t"

  916.         YSCALEYUV2RGB(%%REGBP, %5)

  917.         "pxor    %%mm7, %%mm7                   \n\t"

  918.         WRITEBGR24(%%REGb, 8280(%5), %%REGBP)

  919.         "pop %%"REG_BP"                         \n\t"

  920.         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  921.         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  922.            "a" (&c->redDither)

  923.     );

  924. }

  925. 

  926. static void RENAME(yuv2rgb555_2)(SwsContext *c, const int16_t *buf[2],

  927.                                  const int16_t *ubuf[2], const int16_t *vbuf[2],

  928.                                  const int16_t *abuf[2], uint8_t *dest,

  929.                                  int dstW, int yalpha, int uvalpha, int y)

  930. {

  931.     const int16_t *buf0  = buf[0],  *buf1  = buf[1],

  932.                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  933. 

  934.     //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(

  935.     __asm__ volatile(

  936.         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  937.         "mov        %4, %%"REG_b"               \n\t"

  938.         "push %%"REG_BP"                        \n\t"

  939.         YSCALEYUV2RGB(%%REGBP, %5)

  940.         "pxor    %%mm7, %%mm7                   \n\t"

  941.         /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  942. #ifdef DITHER1XBPP

  943.         "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"

  944.         "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"

  945.         "paddusb "RED_DITHER"(%5), %%mm5      \n\t"

  946. #endif

  947.         WRITERGB15(%%REGb, 8280(%5), %%REGBP)

  948.         "pop %%"REG_BP"                         \n\t"

  949.         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  950.         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  951.            "a" (&c->redDither)

  952.     );

  953. }

  954. 

  955. static void RENAME(yuv2rgb565_2)(SwsContext *c, const int16_t *buf[2],

  956.                                  const int16_t *ubuf[2], const int16_t *vbuf[2],

  957.                                  const int16_t *abuf[2], uint8_t *dest,

  958.                                  int dstW, int yalpha, int uvalpha, int y)

  959. {

  960.     const int16_t *buf0  = buf[0],  *buf1  = buf[1],

  961.                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  962. 

  963.     //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(

  964.     __asm__ volatile(

  965.         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  966.         "mov        %4, %%"REG_b"               \n\t"

  967.         "push %%"REG_BP"                        \n\t"

  968.         YSCALEYUV2RGB(%%REGBP, %5)

  969.         "pxor    %%mm7, %%mm7                   \n\t"

  970.         /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  971. #ifdef DITHER1XBPP

  972.         "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"

  973.         "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"

  974.         "paddusb "RED_DITHER"(%5), %%mm5      \n\t"

  975. #endif

  976.         WRITERGB16(%%REGb, 8280(%5), %%REGBP)

  977.         "pop %%"REG_BP"                         \n\t"

  978.         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  979.         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  980.            "a" (&c->redDither)

  981.     );

  982. }

  983. 

  984. #define REAL_YSCALEYUV2PACKED(index, c) \

  985.     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0              \n\t"\

  986.     "movq "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm1              \n\t"\

  987.     "psraw                $3, %%mm0                           \n\t"\

  988.     "psraw                $3, %%mm1                           \n\t"\

  989.     "movq              %%mm0, "CHR_MMX_FILTER_OFFSET"+8("#c") \n\t"\

  990.     "movq              %%mm1, "LUM_MMX_FILTER_OFFSET"+8("#c") \n\t"\

  991.     "xor            "#index", "#index"                        \n\t"\

  992.     ".p2align              4            \n\t"\

  993.     "1:                                 \n\t"\

  994.     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\

  995.     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\

  996.     "add           "UV_OFFx2"("#c"), "#index"  \n\t" \

  997.     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\

  998.     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\

  999.     "sub           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1000.     "psubw             %%mm3, %%mm2     \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\

  1001.     "psubw             %%mm4, %%mm5     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\

  1002.     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0    \n\t"\

  1003.     "pmulhw            %%mm0, %%mm2     \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\

  1004.     "pmulhw            %%mm0, %%mm5     \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\

  1005.     "psraw                $7, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\

  1006.     "psraw                $7, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\

  1007.     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\

  1008.     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\

  1009.     "movq  (%0, "#index", 2), %%mm0     \n\t" /*buf0[eax]*/\

  1010.     "movq  (%1, "#index", 2), %%mm1     \n\t" /*buf1[eax]*/\

  1011.     "movq 8(%0, "#index", 2), %%mm6     \n\t" /*buf0[eax]*/\

  1012.     "movq 8(%1, "#index", 2), %%mm7     \n\t" /*buf1[eax]*/\

  1013.     "psubw             %%mm1, %%mm0     \n\t" /* buf0[eax] - buf1[eax]*/\

  1014.     "psubw             %%mm7, %%mm6     \n\t" /* buf0[eax] - buf1[eax]*/\

  1015.     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\

  1016.     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\

  1017.     "psraw                $7, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  1018.     "psraw                $7, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  1019.     "paddw             %%mm0, %%mm1     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\

  1020.     "paddw             %%mm6, %%mm7     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\

  1021. 

  1022. #define YSCALEYUV2PACKED(index, c)  REAL_YSCALEYUV2PACKED(index, c)

  1023. 

  1024. static void RENAME(yuv2yuyv422_2)(SwsContext *c, const int16_t *buf[2],

  1025.                                   const int16_t *ubuf[2], const int16_t *vbuf[2],

  1026.                                   const int16_t *abuf[2], uint8_t *dest,

  1027.                                   int dstW, int yalpha, int uvalpha, int y)

  1028. {

  1029.     const int16_t *buf0  = buf[0],  *buf1  = buf[1],

  1030.                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  1031. 

  1032.     //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(

  1033.     __asm__ volatile(

  1034.         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1035.         "mov %4, %%"REG_b"                        \n\t"

  1036.         "push %%"REG_BP"                        \n\t"

  1037.         YSCALEYUV2PACKED(%%REGBP, %5)

  1038.         WRITEYUY2(%%REGb, 8280(%5), %%REGBP)

  1039.         "pop %%"REG_BP"                         \n\t"

  1040.         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1041.         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1042.            "a" (&c->redDither)

  1043.     );

  1044. }

  1045. 

  1046. #define REAL_YSCALEYUV2RGB1(index, c) \

  1047.     "xor            "#index", "#index"  \n\t"\

  1048.     ".p2align              4            \n\t"\

  1049.     "1:                                 \n\t"\

  1050.     "movq     (%2, "#index"), %%mm3     \n\t" /* uvbuf0[eax]*/\

  1051.     "add           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1052.     "movq     (%2, "#index"), %%mm4     \n\t" /* uvbuf0[eax+2048]*/\

  1053.     "sub           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1054.     "psraw                $4, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\

  1055.     "psraw                $4, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\

  1056.     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\

  1057.     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\

  1058.     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\

  1059.     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\

  1060.     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\

  1061.     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\

  1062.     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\

  1063.     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\

  1064.     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\

  1065.     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  1066.     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  1067.     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\

  1068.     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\

  1069.     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\

  1070.     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\

  1071.     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\

  1072.     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\

  1073.     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\

  1074.     "paddw             %%mm3, %%mm4     \n\t"\

  1075.     "movq              %%mm2, %%mm0     \n\t"\

  1076.     "movq              %%mm5, %%mm6     \n\t"\

  1077.     "movq              %%mm4, %%mm3     \n\t"\

  1078.     "punpcklwd         %%mm2, %%mm2     \n\t"\

  1079.     "punpcklwd         %%mm5, %%mm5     \n\t"\

  1080.     "punpcklwd         %%mm4, %%mm4     \n\t"\

  1081.     "paddw             %%mm1, %%mm2     \n\t"\

  1082.     "paddw             %%mm1, %%mm5     \n\t"\

  1083.     "paddw             %%mm1, %%mm4     \n\t"\

  1084.     "punpckhwd         %%mm0, %%mm0     \n\t"\

  1085.     "punpckhwd         %%mm6, %%mm6     \n\t"\

  1086.     "punpckhwd         %%mm3, %%mm3     \n\t"\

  1087.     "paddw             %%mm7, %%mm0     \n\t"\

  1088.     "paddw             %%mm7, %%mm6     \n\t"\

  1089.     "paddw             %%mm7, %%mm3     \n\t"\

  1090.     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\

  1091.     "packuswb          %%mm0, %%mm2     \n\t"\

  1092.     "packuswb          %%mm6, %%mm5     \n\t"\

  1093.     "packuswb          %%mm3, %%mm4     \n\t"\

  1094. 

  1095. #define YSCALEYUV2RGB1(index, c)  REAL_YSCALEYUV2RGB1(index, c)

  1096. 

  1097. // do vertical chrominance interpolation

  1098. #define REAL_YSCALEYUV2RGB1b(index, c) \

  1099.     "xor            "#index", "#index"  \n\t"\

  1100.     ".p2align              4            \n\t"\

  1101.     "1:                                 \n\t"\

  1102.     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\

  1103.     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\

  1104.     "add           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1105.     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\

  1106.     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\

  1107.     "sub           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1108.     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\

  1109.     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\

  1110.     "psrlw                $5, %%mm3     \n\t" /*FIXME might overflow*/\

  1111.     "psrlw                $5, %%mm4     \n\t" /*FIXME might overflow*/\

  1112.     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\

  1113.     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\

  1114.     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\

  1115.     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\

  1116.     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\

  1117.     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\

  1118.     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\

  1119.     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\

  1120.     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\

  1121.     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  1122.     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\

  1123.     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\

  1124.     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\

  1125.     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\

  1126.     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\

  1127.     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\

  1128.     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\

  1129.     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\

  1130.     "paddw             %%mm3, %%mm4     \n\t"\

  1131.     "movq              %%mm2, %%mm0     \n\t"\

  1132.     "movq              %%mm5, %%mm6     \n\t"\

  1133.     "movq              %%mm4, %%mm3     \n\t"\

  1134.     "punpcklwd         %%mm2, %%mm2     \n\t"\

  1135.     "punpcklwd         %%mm5, %%mm5     \n\t"\

  1136.     "punpcklwd         %%mm4, %%mm4     \n\t"\

  1137.     "paddw             %%mm1, %%mm2     \n\t"\

  1138.     "paddw             %%mm1, %%mm5     \n\t"\

  1139.     "paddw             %%mm1, %%mm4     \n\t"\

  1140.     "punpckhwd         %%mm0, %%mm0     \n\t"\

  1141.     "punpckhwd         %%mm6, %%mm6     \n\t"\

  1142.     "punpckhwd         %%mm3, %%mm3     \n\t"\

  1143.     "paddw             %%mm7, %%mm0     \n\t"\

  1144.     "paddw             %%mm7, %%mm6     \n\t"\

  1145.     "paddw             %%mm7, %%mm3     \n\t"\

  1146.     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\

  1147.     "packuswb          %%mm0, %%mm2     \n\t"\

  1148.     "packuswb          %%mm6, %%mm5     \n\t"\

  1149.     "packuswb          %%mm3, %%mm4     \n\t"\

  1150. 

  1151. #define YSCALEYUV2RGB1b(index, c)  REAL_YSCALEYUV2RGB1b(index, c)

  1152. 

  1153. #define REAL_YSCALEYUV2RGB1_ALPHA(index) \

  1154.     "movq  (%1, "#index", 2), %%mm7     \n\t" /* abuf0[index  ]     */\

  1155.     "movq 8(%1, "#index", 2), %%mm1     \n\t" /* abuf0[index+4]     */\

  1156.     "psraw                $7, %%mm7     \n\t" /* abuf0[index  ] >>7 */\

  1157.     "psraw                $7, %%mm1     \n\t" /* abuf0[index+4] >>7 */\

  1158.     "packuswb          %%mm1, %%mm7     \n\t"

  1159. #define YSCALEYUV2RGB1_ALPHA(index) REAL_YSCALEYUV2RGB1_ALPHA(index)

  1160. 

  1161. /**

  1162.  * YV12 to RGB without scaling or interpolating

  1163.  */

  1164. static void RENAME(yuv2rgb32_1)(SwsContext *c, const int16_t *buf0,

  1165.                                 const int16_t *ubuf[2], const int16_t *bguf[2],

  1166.                                 const int16_t *abuf0, uint8_t *dest,

  1167.                                 int dstW, int uvalpha, int y)

  1168. {

  1169.     const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  1170.     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1

  1171. 

  1172.     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster

  1173.         if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  1174.             __asm__ volatile(

  1175.                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1176.                 "mov        %4, %%"REG_b"               \n\t"

  1177.                 "push %%"REG_BP"                        \n\t"

  1178.                 YSCALEYUV2RGB1(%%REGBP, %5)

  1179.                 YSCALEYUV2RGB1_ALPHA(%%REGBP)

  1180.                 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)

  1181.                 "pop %%"REG_BP"                         \n\t"

  1182.                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1183.                 :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1184.                    "a" (&c->redDither)

  1185.             );

  1186.         } else {

  1187.             __asm__ volatile(

  1188.                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1189.                 "mov        %4, %%"REG_b"               \n\t"

  1190.                 "push %%"REG_BP"                        \n\t"

  1191.                 YSCALEYUV2RGB1(%%REGBP, %5)

  1192.                 "pcmpeqd %%mm7, %%mm7                   \n\t"

  1193.                 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)

  1194.                 "pop %%"REG_BP"                         \n\t"

  1195.                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1196.                 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1197.                    "a" (&c->redDither)

  1198.             );

  1199.         }

  1200.     } else {

  1201.         if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  1202.             __asm__ volatile(

  1203.                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1204.                 "mov        %4, %%"REG_b"               \n\t"

  1205.                 "push %%"REG_BP"                        \n\t"

  1206.                 YSCALEYUV2RGB1b(%%REGBP, %5)

  1207.                 YSCALEYUV2RGB1_ALPHA(%%REGBP)

  1208.                 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)

  1209.                 "pop %%"REG_BP"                         \n\t"

  1210.                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1211.                 :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1212.                    "a" (&c->redDither)

  1213.             );

  1214.         } else {

  1215.             __asm__ volatile(

  1216.                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1217.                 "mov        %4, %%"REG_b"               \n\t"

  1218.                 "push %%"REG_BP"                        \n\t"

  1219.                 YSCALEYUV2RGB1b(%%REGBP, %5)

  1220.                 "pcmpeqd %%mm7, %%mm7                   \n\t"

  1221.                 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)

  1222.                 "pop %%"REG_BP"                         \n\t"

  1223.                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1224.                 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1225.                    "a" (&c->redDither)

  1226.             );

  1227.         }

  1228.     }

  1229. }

  1230. 

  1231. static void RENAME(yuv2bgr24_1)(SwsContext *c, const int16_t *buf0,

  1232.                                 const int16_t *ubuf[2], const int16_t *bguf[2],

  1233.                                 const int16_t *abuf0, uint8_t *dest,

  1234.                                 int dstW, int uvalpha, int y)

  1235. {

  1236.     const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  1237.     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1

  1238. 

  1239.     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster

  1240.         __asm__ volatile(

  1241.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1242.             "mov        %4, %%"REG_b"               \n\t"

  1243.             "push %%"REG_BP"                        \n\t"

  1244.             YSCALEYUV2RGB1(%%REGBP, %5)

  1245.             "pxor    %%mm7, %%mm7                   \n\t"

  1246.             WRITEBGR24(%%REGb, 8280(%5), %%REGBP)

  1247.             "pop %%"REG_BP"                         \n\t"

  1248.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1249.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1250.                "a" (&c->redDither)

  1251.         );

  1252.     } else {

  1253.         __asm__ volatile(

  1254.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1255.             "mov        %4, %%"REG_b"               \n\t"

  1256.             "push %%"REG_BP"                        \n\t"

  1257.             YSCALEYUV2RGB1b(%%REGBP, %5)

  1258.             "pxor    %%mm7, %%mm7                   \n\t"

  1259.             WRITEBGR24(%%REGb, 8280(%5), %%REGBP)

  1260.             "pop %%"REG_BP"                         \n\t"

  1261.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1262.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1263.                "a" (&c->redDither)

  1264.         );

  1265.     }

  1266. }

  1267. 

  1268. static void RENAME(yuv2rgb555_1)(SwsContext *c, const int16_t *buf0,

  1269.                                  const int16_t *ubuf[2], const int16_t *bguf[2],

  1270.                                  const int16_t *abuf0, uint8_t *dest,

  1271.                                  int dstW, int uvalpha, int y)

  1272. {

  1273.     const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  1274.     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1

  1275. 

  1276.     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster

  1277.         __asm__ volatile(

  1278.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1279.             "mov        %4, %%"REG_b"               \n\t"

  1280.             "push %%"REG_BP"                        \n\t"

  1281.             YSCALEYUV2RGB1(%%REGBP, %5)

  1282.             "pxor    %%mm7, %%mm7                   \n\t"

  1283.             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  1284. #ifdef DITHER1XBPP

  1285.             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"

  1286.             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"

  1287.             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"

  1288. #endif

  1289.             WRITERGB15(%%REGb, 8280(%5), %%REGBP)

  1290.             "pop %%"REG_BP"                         \n\t"

  1291.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1292.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1293.                "a" (&c->redDither)

  1294.         );

  1295.     } else {

  1296.         __asm__ volatile(

  1297.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1298.             "mov        %4, %%"REG_b"               \n\t"

  1299.             "push %%"REG_BP"                        \n\t"

  1300.             YSCALEYUV2RGB1b(%%REGBP, %5)

  1301.             "pxor    %%mm7, %%mm7                   \n\t"

  1302.             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  1303. #ifdef DITHER1XBPP

  1304.             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"

  1305.             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"

  1306.             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"

  1307. #endif

  1308.             WRITERGB15(%%REGb, 8280(%5), %%REGBP)

  1309.             "pop %%"REG_BP"                         \n\t"

  1310.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1311.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1312.                "a" (&c->redDither)

  1313.         );

  1314.     }

  1315. }

  1316. 

  1317. static void RENAME(yuv2rgb565_1)(SwsContext *c, const int16_t *buf0,

  1318.                                  const int16_t *ubuf[2], const int16_t *bguf[2],

  1319.                                  const int16_t *abuf0, uint8_t *dest,

  1320.                                  int dstW, int uvalpha, int y)

  1321. {

  1322.     const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  1323.     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1

  1324. 

  1325.     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster

  1326.         __asm__ volatile(

  1327.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1328.             "mov        %4, %%"REG_b"               \n\t"

  1329.             "push %%"REG_BP"                        \n\t"

  1330.             YSCALEYUV2RGB1(%%REGBP, %5)

  1331.             "pxor    %%mm7, %%mm7                   \n\t"

  1332.             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  1333. #ifdef DITHER1XBPP

  1334.             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"

  1335.             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"

  1336.             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"

  1337. #endif

  1338.             WRITERGB16(%%REGb, 8280(%5), %%REGBP)

  1339.             "pop %%"REG_BP"                         \n\t"

  1340.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1341.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1342.                "a" (&c->redDither)

  1343.         );

  1344.     } else {

  1345.         __asm__ volatile(

  1346.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1347.             "mov        %4, %%"REG_b"               \n\t"

  1348.             "push %%"REG_BP"                        \n\t"

  1349.             YSCALEYUV2RGB1b(%%REGBP, %5)

  1350.             "pxor    %%mm7, %%mm7                   \n\t"

  1351.             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */

  1352. #ifdef DITHER1XBPP

  1353.             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"

  1354.             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"

  1355.             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"

  1356. #endif

  1357.             WRITERGB16(%%REGb, 8280(%5), %%REGBP)

  1358.             "pop %%"REG_BP"                         \n\t"

  1359.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1360.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1361.                "a" (&c->redDither)

  1362.         );

  1363.     }

  1364. }

  1365. 

  1366. #define REAL_YSCALEYUV2PACKED1(index, c) \

  1367.     "xor            "#index", "#index"  \n\t"\

  1368.     ".p2align              4            \n\t"\

  1369.     "1:                                 \n\t"\

  1370.     "movq     (%2, "#index"), %%mm3     \n\t" /* uvbuf0[eax]*/\

  1371.     "add           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1372.     "movq     (%2, "#index"), %%mm4     \n\t" /* uvbuf0[eax+2048]*/\

  1373.     "sub           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1374.     "psraw                $7, %%mm3     \n\t" \

  1375.     "psraw                $7, %%mm4     \n\t" \

  1376.     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\

  1377.     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\

  1378.     "psraw                $7, %%mm1     \n\t" \

  1379.     "psraw                $7, %%mm7     \n\t" \

  1380. 

  1381. #define YSCALEYUV2PACKED1(index, c)  REAL_YSCALEYUV2PACKED1(index, c)

  1382. 

  1383. #define REAL_YSCALEYUV2PACKED1b(index, c) \

  1384.     "xor "#index", "#index"             \n\t"\

  1385.     ".p2align              4            \n\t"\

  1386.     "1:                                 \n\t"\

  1387.     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\

  1388.     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\

  1389.     "add           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1390.     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\

  1391.     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\

  1392.     "sub           "UV_OFFx2"("#c"), "#index"  \n\t" \

  1393.     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\

  1394.     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\

  1395.     "psrlw                $8, %%mm3     \n\t" \

  1396.     "psrlw                $8, %%mm4     \n\t" \

  1397.     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\

  1398.     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\

  1399.     "psraw                $7, %%mm1     \n\t" \

  1400.     "psraw                $7, %%mm7     \n\t"

  1401. #define YSCALEYUV2PACKED1b(index, c)  REAL_YSCALEYUV2PACKED1b(index, c)

  1402. 

  1403. static void RENAME(yuv2yuyv422_1)(SwsContext *c, const int16_t *buf0,

  1404.                                   const int16_t *ubuf[2], const int16_t *bguf[2],

  1405.                                   const int16_t *abuf0, uint8_t *dest,

  1406.                                   int dstW, int uvalpha, int y)

  1407. {

  1408.     const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];

  1409.     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1

  1410. 

  1411.     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster

  1412.         __asm__ volatile(

  1413.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1414.             "mov        %4, %%"REG_b"               \n\t"

  1415.             "push %%"REG_BP"                        \n\t"

  1416.             YSCALEYUV2PACKED1(%%REGBP, %5)

  1417.             WRITEYUY2(%%REGb, 8280(%5), %%REGBP)

  1418.             "pop %%"REG_BP"                         \n\t"

  1419.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1420.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1421.                "a" (&c->redDither)

  1422.         );

  1423.     } else {

  1424.         __asm__ volatile(

  1425.             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"

  1426.             "mov        %4, %%"REG_b"               \n\t"

  1427.             "push %%"REG_BP"                        \n\t"

  1428.             YSCALEYUV2PACKED1b(%%REGBP, %5)

  1429.             WRITEYUY2(%%REGb, 8280(%5), %%REGBP)

  1430.             "pop %%"REG_BP"                         \n\t"

  1431.             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"

  1432.             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),

  1433.                "a" (&c->redDither)

  1434.         );

  1435.     }

  1436. }

  1437. 

  1438. #if !COMPILE_TEMPLATE_MMX2

  1439. //FIXME yuy2* can read up to 7 samples too much

  1440. 

  1441. static void RENAME(yuy2ToY)(uint8_t *dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2,

  1442.                             int width, uint32_t *unused)

  1443. {

  1444.     __asm__ volatile(

  1445.         "movq "MANGLE(bm01010101)", %%mm2           \n\t"

  1446.         "mov                    %0, %%"REG_a"       \n\t"

  1447.         "1:                                         \n\t"

  1448.         "movq    (%1, %%"REG_a",2), %%mm0           \n\t"

  1449.         "movq   8(%1, %%"REG_a",2), %%mm1           \n\t"

  1450.         "pand                %%mm2, %%mm0           \n\t"

  1451.         "pand                %%mm2, %%mm1           \n\t"

  1452.         "packuswb            %%mm1, %%mm0           \n\t"

  1453.         "movq                %%mm0, (%2, %%"REG_a") \n\t"

  1454.         "add                    $8, %%"REG_a"       \n\t"

  1455.         " js                    1b                  \n\t"

  1456.         : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst+width)

  1457.         : "%"REG_a

  1458.     );

  1459. }

  1460. 

  1461. static void RENAME(yuy2ToUV)(uint8_t *dstU, uint8_t *dstV,

  1462.                              const uint8_t *unused1, const uint8_t *src1, const uint8_t *src2,

  1463.                              int width, uint32_t *unused)

  1464. {

  1465.     __asm__ volatile(

  1466.         "movq "MANGLE(bm01010101)", %%mm4           \n\t"

  1467.         "mov                    %0, %%"REG_a"       \n\t"

  1468.         "1:                                         \n\t"

  1469.         "movq    (%1, %%"REG_a",4), %%mm0           \n\t"

  1470.         "movq   8(%1, %%"REG_a",4), %%mm1           \n\t"

  1471.         "psrlw                  $8, %%mm0           \n\t"

  1472.         "psrlw                  $8, %%mm1           \n\t"

  1473.         "packuswb            %%mm1, %%mm0           \n\t"

  1474.         "movq                %%mm0, %%mm1           \n\t"

  1475.         "psrlw                  $8, %%mm0           \n\t"

  1476.         "pand                %%mm4, %%mm1           \n\t"

  1477.         "packuswb            %%mm0, %%mm0           \n\t"

  1478.         "packuswb            %%mm1, %%mm1           \n\t"

  1479.         "movd                %%mm0, (%3, %%"REG_a") \n\t"

  1480.         "movd                %%mm1, (%2, %%"REG_a") \n\t"

  1481.         "add                    $4, %%"REG_a"       \n\t"

  1482.         " js                    1b                  \n\t"

  1483.         : : "g" ((x86_reg)-width), "r" (src1+width*4), "r" (dstU+width), "r" (dstV+width)

  1484.         : "%"REG_a

  1485.     );

  1486.     assert(src1 == src2);

  1487. }

  1488. 

  1489. /* This is almost identical to the previous, end exists only because

  1490.  * yuy2ToY/UV)(dst, src+1, ...) would have 100% unaligned accesses. */

  1491. static void RENAME(uyvyToY)(uint8_t *dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2,

  1492.                             int width, uint32_t *unused)

  1493. {

  1494.     __asm__ volatile(

  1495.         "mov                  %0, %%"REG_a"         \n\t"

  1496.         "1:                                         \n\t"

  1497.         "movq  (%1, %%"REG_a",2), %%mm0             \n\t"

  1498.         "movq 8(%1, %%"REG_a",2), %%mm1             \n\t"

  1499.         "psrlw                $8, %%mm0             \n\t"

  1500.         "psrlw                $8, %%mm1             \n\t"

  1501.         "packuswb          %%mm1, %%mm0             \n\t"

  1502.         "movq              %%mm0, (%2, %%"REG_a")   \n\t"

  1503.         "add                  $8, %%"REG_a"         \n\t"

  1504.         " js                  1b                    \n\t"

  1505.         : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst+width)

  1506.         : "%"REG_a

  1507.     );

  1508. }

  1509. 

  1510. static void RENAME(uyvyToUV)(uint8_t *dstU, uint8_t *dstV,

  1511.                              const uint8_t *unused1, const uint8_t *src1, const uint8_t *src2,

  1512.                              int width, uint32_t *unused)

  1513. {

  1514.     __asm__ volatile(

  1515.         "movq "MANGLE(bm01010101)", %%mm4           \n\t"

  1516.         "mov                    %0, %%"REG_a"       \n\t"

  1517.         "1:                                         \n\t"

  1518.         "movq    (%1, %%"REG_a",4), %%mm0           \n\t"

  1519.         "movq   8(%1, %%"REG_a",4), %%mm1           \n\t"

  1520.         "pand                %%mm4, %%mm0           \n\t"

  1521.         "pand                %%mm4, %%mm1           \n\t"

  1522.         "packuswb            %%mm1, %%mm0           \n\t"

  1523.         "movq                %%mm0, %%mm1           \n\t"

  1524.         "psrlw                  $8, %%mm0           \n\t"

  1525.         "pand                %%mm4, %%mm1           \n\t"

  1526.         "packuswb            %%mm0, %%mm0           \n\t"

  1527.         "packuswb            %%mm1, %%mm1           \n\t"

  1528.         "movd                %%mm0, (%3, %%"REG_a") \n\t"

  1529.         "movd                %%mm1, (%2, %%"REG_a") \n\t"

  1530.         "add                    $4, %%"REG_a"       \n\t"

  1531.         " js                    1b                  \n\t"

  1532.         : : "g" ((x86_reg)-width), "r" (src1+width*4), "r" (dstU+width), "r" (dstV+width)

  1533.         : "%"REG_a

  1534.     );

  1535.     assert(src1 == src2);

  1536. }

  1537. 

  1538. static av_always_inline void RENAME(nvXXtoUV)(uint8_t *dst1, uint8_t *dst2,

  1539.                                               const uint8_t *src, int width)

  1540. {

  1541.     __asm__ volatile(

  1542.         "movq "MANGLE(bm01010101)", %%mm4           \n\t"

  1543.         "mov                    %0, %%"REG_a"       \n\t"

  1544.         "1:                                         \n\t"

  1545.         "movq    (%1, %%"REG_a",2), %%mm0           \n\t"

  1546.         "movq   8(%1, %%"REG_a",2), %%mm1           \n\t"

  1547.         "movq                %%mm0, %%mm2           \n\t"

  1548.         "movq                %%mm1, %%mm3           \n\t"

  1549.         "pand                %%mm4, %%mm0           \n\t"

  1550.         "pand                %%mm4, %%mm1           \n\t"

  1551.         "psrlw                  $8, %%mm2           \n\t"

  1552.         "psrlw                  $8, %%mm3           \n\t"

  1553.         "packuswb            %%mm1, %%mm0           \n\t"

  1554.         "packuswb            %%mm3, %%mm2           \n\t"

  1555.         "movq                %%mm0, (%2, %%"REG_a") \n\t"

  1556.         "movq                %%mm2, (%3, %%"REG_a") \n\t"

  1557.         "add                    $8, %%"REG_a"       \n\t"

  1558.         " js                    1b                  \n\t"

  1559.         : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst1+width), "r" (dst2+width)

  1560.         : "%"REG_a

  1561.     );

  1562. }

  1563. 

  1564. static void RENAME(nv12ToUV)(uint8_t *dstU, uint8_t *dstV,

  1565.                              const uint8_t *unused1, const uint8_t *src1, const uint8_t *src2,

  1566.                              int width, uint32_t *unused)

  1567. {

  1568.     RENAME(nvXXtoUV)(dstU, dstV, src1, width);

  1569. }

  1570. 

  1571. static void RENAME(nv21ToUV)(uint8_t *dstU, uint8_t *dstV,

  1572.                              const uint8_t *unused1, const uint8_t *src1, const uint8_t *src2,

  1573.                              int width, uint32_t *unused)

  1574. {

  1575.     RENAME(nvXXtoUV)(dstV, dstU, src1, width);

  1576. }

  1577. #endif /* !COMPILE_TEMPLATE_MMX2 */

  1578. 

  1579. static av_always_inline void RENAME(bgr24ToY_mmx)(int16_t *dst, const uint8_t *src,

  1580.                                                   int width, enum PixelFormat srcFormat)

  1581. {

  1582. 

  1583.     if(srcFormat == PIX_FMT_BGR24) {

  1584.         __asm__ volatile(

  1585.             "movq  "MANGLE(ff_bgr24toY1Coeff)", %%mm5       \n\t"

  1586.             "movq  "MANGLE(ff_bgr24toY2Coeff)", %%mm6       \n\t"

  1587.             :

  1588.         );

  1589.     } else {

  1590.         __asm__ volatile(

  1591.             "movq  "MANGLE(ff_rgb24toY1Coeff)", %%mm5       \n\t"

  1592.             "movq  "MANGLE(ff_rgb24toY2Coeff)", %%mm6       \n\t"

  1593.             :

  1594.         );

  1595.     }

  1596. 

  1597.     __asm__ volatile(

  1598.         "movq  "MANGLE(ff_bgr24toYOffset)", %%mm4   \n\t"

  1599.         "mov                        %2, %%"REG_a"   \n\t"

  1600.         "pxor                    %%mm7, %%mm7       \n\t"

  1601.         "1:                                         \n\t"

  1602.         PREFETCH"               64(%0)              \n\t"

  1603.         "movd                     (%0), %%mm0       \n\t"

  1604.         "movd                    2(%0), %%mm1       \n\t"

  1605.         "movd                    6(%0), %%mm2       \n\t"

  1606.         "movd                    8(%0), %%mm3       \n\t"

  1607.         "add                       $12, %0          \n\t"

  1608.         "punpcklbw               %%mm7, %%mm0       \n\t"

  1609.         "punpcklbw               %%mm7, %%mm1       \n\t"

  1610.         "punpcklbw               %%mm7, %%mm2       \n\t"

  1611.         "punpcklbw               %%mm7, %%mm3       \n\t"

  1612.         "pmaddwd                 %%mm5, %%mm0       \n\t"

  1613.         "pmaddwd                 %%mm6, %%mm1       \n\t"

  1614.         "pmaddwd                 %%mm5, %%mm2       \n\t"

  1615.         "pmaddwd                 %%mm6, %%mm3       \n\t"

  1616.         "paddd                   %%mm1, %%mm0       \n\t"

  1617.         "paddd                   %%mm3, %%mm2       \n\t"

  1618.         "paddd                   %%mm4, %%mm0       \n\t"

  1619.         "paddd                   %%mm4, %%mm2       \n\t"

  1620.         "psrad                     $9, %%mm0       \n\t"

  1621.         "psrad                     $9, %%mm2       \n\t"

  1622.         "packssdw                %%mm2, %%mm0       \n\t"

  1623.         "movq                %%mm0, (%1, %%"REG_a") \n\t"

  1624.         "add                        $8, %%"REG_a"   \n\t"

  1625.         " js                        1b              \n\t"

  1626.     : "+r" (src)

  1627.     : "r" (dst+width), "g" ((x86_reg)-2*width)

  1628.     : "%"REG_a

  1629.     );

  1630. }

  1631. 

  1632. static void RENAME(bgr24ToY)(int16_t *dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2,

  1633.                              int width, uint32_t *unused)

  1634. {

  1635.     RENAME(bgr24ToY_mmx)(dst, src, width, PIX_FMT_BGR24);

  1636. }

  1637. 

  1638. static void RENAME(rgb24ToY)(int16_t *dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2,

  1639.                              int width, uint32_t *unused)

  1640. {

  1641.     RENAME(bgr24ToY_mmx)(dst, src, width, PIX_FMT_RGB24);

  1642. }

  1643. 

  1644. static av_always_inline void RENAME(bgr24ToUV_mmx)(int16_t *dstU, int16_t *dstV,

  1645.                                                    const uint8_t *src, int width,

  1646.                                                    enum PixelFormat srcFormat)

  1647. {

  1648.     __asm__ volatile(

  1649.         "movq                    24(%4), %%mm6       \n\t"

  1650.         "mov                        %3, %%"REG_a"   \n\t"

  1651.         "pxor                    %%mm7, %%mm7       \n\t"

  1652.         "1:                                         \n\t"

  1653.         PREFETCH"               64(%0)              \n\t"

  1654.         "movd                     (%0), %%mm0       \n\t"

  1655.         "movd                    2(%0), %%mm1       \n\t"

  1656.         "punpcklbw               %%mm7, %%mm0       \n\t"

  1657.         "punpcklbw               %%mm7, %%mm1       \n\t"

  1658.         "movq                    %%mm0, %%mm2       \n\t"

  1659.         "movq                    %%mm1, %%mm3       \n\t"

  1660.         "pmaddwd                  (%4), %%mm0       \n\t"

  1661.         "pmaddwd                 8(%4), %%mm1       \n\t"

  1662.         "pmaddwd                16(%4), %%mm2       \n\t"

  1663.         "pmaddwd                 %%mm6, %%mm3       \n\t"

  1664.         "paddd                   %%mm1, %%mm0       \n\t"

  1665.         "paddd                   %%mm3, %%mm2       \n\t"

  1666. 

  1667.         "movd                    6(%0), %%mm1       \n\t"

  1668.         "movd                    8(%0), %%mm3       \n\t"

  1669.         "add                       $12, %0          \n\t"

  1670.         "punpcklbw               %%mm7, %%mm1       \n\t"

  1671.         "punpcklbw               %%mm7, %%mm3       \n\t"

  1672.         "movq                    %%mm1, %%mm4       \n\t"

  1673.         "movq                    %%mm3, %%mm5       \n\t"

  1674.         "pmaddwd                  (%4), %%mm1       \n\t"

  1675.         "pmaddwd                 8(%4), %%mm3       \n\t"

  1676.         "pmaddwd                16(%4), %%mm4       \n\t"

  1677.         "pmaddwd                 %%mm6, %%mm5       \n\t"

  1678.         "paddd                   %%mm3, %%mm1       \n\t"

  1679.         "paddd                   %%mm5, %%mm4       \n\t"

  1680. 

  1681.         "movq "MANGLE(ff_bgr24toUVOffset)", %%mm3       \n\t"

  1682.         "paddd                   %%mm3, %%mm0       \n\t"

  1683.         "paddd                   %%mm3, %%mm2       \n\t"

  1684.         "paddd                   %%mm3, %%mm1       \n\t"

  1685.         "paddd                   %%mm3, %%mm4       \n\t"

  1686.         "psrad                     $9, %%mm0       \n\t"

  1687.         "psrad                     $9, %%mm2       \n\t"

  1688.         "psrad                     $9, %%mm1       \n\t"

  1689.         "psrad                     $9, %%mm4       \n\t"

  1690.         "packssdw                %%mm1, %%mm0       \n\t"

  1691.         "packssdw                %%mm4, %%mm2       \n\t"

  1692.         "movq                %%mm0, (%1, %%"REG_a") \n\t"

  1693.         "movq                %%mm2, (%2, %%"REG_a") \n\t"

  1694.         "add                        $8, %%"REG_a"   \n\t"

  1695.         " js                        1b              \n\t"

  1696.     : "+r" (src)

  1697.     : "r" (dstU+width), "r" (dstV+width), "g" ((x86_reg)-2*width), "r"(ff_bgr24toUV[srcFormat == PIX_FMT_RGB24])

  1698.     : "%"REG_a

  1699.     );

  1700. }

  1701. 

  1702. static void RENAME(bgr24ToUV)(int16_t *dstU, int16_t *dstV,

  1703.                               const uint8_t *unused1, const uint8_t *src1, const uint8_t *src2,

  1704.                               int width, uint32_t *unused)

  1705. {

  1706.     RENAME(bgr24ToUV_mmx)(dstU, dstV, src1, width, PIX_FMT_BGR24);

  1707.     assert(src1 == src2);

  1708. }

  1709. 

  1710. static void RENAME(rgb24ToUV)(int16_t *dstU, int16_t *dstV,

  1711.                               const uint8_t *unused1, const uint8_t *src1, const uint8_t *src2,

  1712.                               int width, uint32_t *unused)

  1713. {

  1714.     assert(src1==src2);

  1715.     RENAME(bgr24ToUV_mmx)(dstU, dstV, src1, width, PIX_FMT_RGB24);

  1716. }

  1717. 

  1718. #if COMPILE_TEMPLATE_MMX2

  1719. static void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,

  1720.                                  int dstWidth, const uint8_t *src,

  1721.                                  int srcW, int xInc)

  1722. {

  1723.     int16_t *filterPos = c->hLumFilterPos;

  1724.     int16_t *filter    = c->hLumFilter;

  1725.     void    *mmx2FilterCode= c->lumMmx2FilterCode;

  1726.     int i;

  1727. #if defined(PIC)

  1728.     DECLARE_ALIGNED(8, uint64_t, ebxsave);

  1729. #endif

  1730. 

  1731.     __asm__ volatile(

  1732. #if defined(PIC)

  1733.         "mov               %%"REG_b", %5        \n\t"

  1734. #endif

  1735.         "pxor                  %%mm7, %%mm7     \n\t"

  1736.         "mov                      %0, %%"REG_c" \n\t"

  1737.         "mov                      %1, %%"REG_D" \n\t"

  1738.         "mov                      %2, %%"REG_d" \n\t"

  1739.         "mov                      %3, %%"REG_b" \n\t"

  1740.         "xor               %%"REG_a", %%"REG_a" \n\t" // i

  1741.         PREFETCH"        (%%"REG_c")            \n\t"

  1742.         PREFETCH"      32(%%"REG_c")            \n\t"

  1743.         PREFETCH"      64(%%"REG_c")            \n\t"

  1744. 

  1745. #if ARCH_X86_64

  1746. #define CALL_MMX2_FILTER_CODE \

  1747.         "movl            (%%"REG_b"), %%esi     \n\t"\

  1748.         "call                    *%4            \n\t"\

  1749.         "movl (%%"REG_b", %%"REG_a"), %%esi     \n\t"\

  1750.         "add               %%"REG_S", %%"REG_c" \n\t"\

  1751.         "add               %%"REG_a", %%"REG_D" \n\t"\

  1752.         "xor               %%"REG_a", %%"REG_a" \n\t"\

  1753. 

  1754. #else

  1755. #define CALL_MMX2_FILTER_CODE \

  1756.         "movl (%%"REG_b"), %%esi        \n\t"\

  1757.         "call         *%4                       \n\t"\

  1758.         "addl (%%"REG_b", %%"REG_a"), %%"REG_c" \n\t"\

  1759.         "add               %%"REG_a", %%"REG_D" \n\t"\

  1760.         "xor               %%"REG_a", %%"REG_a" \n\t"\

  1761. 

  1762. #endif /* ARCH_X86_64 */

  1763. 

  1764.         CALL_MMX2_FILTER_CODE

  1765.         CALL_MMX2_FILTER_CODE

  1766.         CALL_MMX2_FILTER_CODE

  1767.         CALL_MMX2_FILTER_CODE

  1768.         CALL_MMX2_FILTER_CODE

  1769.         CALL_MMX2_FILTER_CODE

  1770.         CALL_MMX2_FILTER_CODE

  1771.         CALL_MMX2_FILTER_CODE

  1772. 

  1773. #if defined(PIC)

  1774.         "mov                      %5, %%"REG_b" \n\t"

  1775. #endif

  1776.         :: "m" (src), "m" (dst), "m" (filter), "m" (filterPos),

  1777.            "m" (mmx2FilterCode)

  1778. #if defined(PIC)

  1779.           ,"m" (ebxsave)

  1780. #endif

  1781.         : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D

  1782. #if !defined(PIC)

  1783.          ,"%"REG_b

  1784. #endif

  1785.     );

  1786. 

  1787.     for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--)

  1788.         dst[i] = src[srcW-1]*128;

  1789. }

  1790. 

  1791. static void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst1, int16_t *dst2,

  1792.                                  int dstWidth, const uint8_t *src1,

  1793.                                  const uint8_t *src2, int srcW, int xInc)

  1794. {

  1795.     int16_t *filterPos = c->hChrFilterPos;

  1796.     int16_t *filter    = c->hChrFilter;

  1797.     void    *mmx2FilterCode= c->chrMmx2FilterCode;

  1798.     int i;

  1799. #if defined(PIC)

  1800.     DECLARE_ALIGNED(8, uint64_t, ebxsave);

  1801. #endif

  1802. 

  1803.     __asm__ volatile(

  1804. #if defined(PIC)

  1805.         "mov          %%"REG_b", %7         \n\t"

  1806. #endif

  1807.         "pxor             %%mm7, %%mm7      \n\t"

  1808.         "mov                 %0, %%"REG_c"  \n\t"

  1809.         "mov                 %1, %%"REG_D"  \n\t"

  1810.         "mov                 %2, %%"REG_d"  \n\t"

  1811.         "mov                 %3, %%"REG_b"  \n\t"

  1812.         "xor          %%"REG_a", %%"REG_a"  \n\t" // i

  1813.         PREFETCH"   (%%"REG_c")             \n\t"

  1814.         PREFETCH" 32(%%"REG_c")             \n\t"

  1815.         PREFETCH" 64(%%"REG_c")             \n\t"

  1816. 

  1817.         CALL_MMX2_FILTER_CODE

  1818.         CALL_MMX2_FILTER_CODE

  1819.         CALL_MMX2_FILTER_CODE

  1820.         CALL_MMX2_FILTER_CODE

  1821.         "xor          %%"REG_a", %%"REG_a"  \n\t" // i

  1822.         "mov                 %5, %%"REG_c"  \n\t" // src

  1823.         "mov                 %6, %%"REG_D"  \n\t" // buf2

  1824.         PREFETCH"   (%%"REG_c")             \n\t"

  1825.         PREFETCH" 32(%%"REG_c")             \n\t"

  1826.         PREFETCH" 64(%%"REG_c")             \n\t"

  1827. 

  1828.         CALL_MMX2_FILTER_CODE

  1829.         CALL_MMX2_FILTER_CODE

  1830.         CALL_MMX2_FILTER_CODE

  1831.         CALL_MMX2_FILTER_CODE

  1832. 

  1833. #if defined(PIC)

  1834.         "mov %7, %%"REG_b"    \n\t"

  1835. #endif

  1836.         :: "m" (src1), "m" (dst1), "m" (filter), "m" (filterPos),

  1837.            "m" (mmx2FilterCode), "m" (src2), "m"(dst2)

  1838. #if defined(PIC)

  1839.           ,"m" (ebxsave)

  1840. #endif

  1841.         : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D

  1842. #if !defined(PIC)

  1843.          ,"%"REG_b

  1844. #endif

  1845.     );

  1846. 

  1847.     for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) {

  1848.         dst1[i] = src1[srcW-1]*128;

  1849.         dst2[i] = src2[srcW-1]*128;

  1850.     }

  1851. }

  1852. #endif /* COMPILE_TEMPLATE_MMX2 */

  1853. 

  1854. static av_cold void RENAME(sws_init_swScale)(SwsContext *c)

  1855. {

  1856.     enum PixelFormat srcFormat = c->srcFormat,

  1857.                      dstFormat = c->dstFormat;

  1858.     c->use_mmx_vfilter= 0;

  1859.     if (!is16BPS(dstFormat) && !is9_OR_10BPS(dstFormat) && dstFormat != PIX_FMT_NV12

  1860.         && dstFormat != PIX_FMT_NV21 && !(c->flags & SWS_BITEXACT)) {

  1861.             if (c->flags & SWS_ACCURATE_RND) {

  1862.                 if (!(c->flags & SWS_FULL_CHR_H_INT)) {

  1863.                     switch (c->dstFormat) {

  1864.                     case PIX_FMT_RGB32:   c->yuv2packedX = RENAME(yuv2rgb32_X_ar);   break;

  1865.                     case PIX_FMT_BGR24:   c->yuv2packedX = RENAME(yuv2bgr24_X_ar);   break;

  1866.                     case PIX_FMT_RGB555:  c->yuv2packedX = RENAME(yuv2rgb555_X_ar);  break;

  1867.                     case PIX_FMT_RGB565:  c->yuv2packedX = RENAME(yuv2rgb565_X_ar);  break;

  1868.                     case PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X_ar); break;

  1869.                     default: break;

  1870.                     }

  1871.                 }

  1872.             } else {

  1873.                 int should_dither= isNBPS(c->srcFormat) || is16BPS(c->srcFormat);

  1874.                 c->use_mmx_vfilter= 1;

  1875.                 c->yuv2planeX = RENAME(yuv2yuvX    );

  1876.                 if (!(c->flags & SWS_FULL_CHR_H_INT)) {

  1877.                     switch (c->dstFormat) {

  1878.                     case PIX_FMT_RGB32:   c->yuv2packedX = RENAME(yuv2rgb32_X);   break;

  1879.                     case PIX_FMT_BGR24:   c->yuv2packedX = RENAME(yuv2bgr24_X);   break;

  1880.                     case PIX_FMT_RGB555:  c->yuv2packedX = RENAME(yuv2rgb555_X);  break;

  1881.                     case PIX_FMT_RGB565:  c->yuv2packedX = RENAME(yuv2rgb565_X);  break;

  1882.                     case PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X); break;

  1883.                     default: break;

  1884.                     }

  1885.                 }

  1886.             }

  1887.         if (!(c->flags & SWS_FULL_CHR_H_INT)) {

  1888.             switch (c->dstFormat) {

  1889.             case PIX_FMT_RGB32:

  1890.                 c->yuv2packed1 = RENAME(yuv2rgb32_1);

  1891.                 c->yuv2packed2 = RENAME(yuv2rgb32_2);

  1892.                 break;

  1893.             case PIX_FMT_BGR24:

  1894.                 c->yuv2packed1 = RENAME(yuv2bgr24_1);

  1895.                 c->yuv2packed2 = RENAME(yuv2bgr24_2);

  1896.                 break;

  1897.             case PIX_FMT_RGB555:

  1898.                 c->yuv2packed1 = RENAME(yuv2rgb555_1);

  1899.                 c->yuv2packed2 = RENAME(yuv2rgb555_2);

  1900.                 break;

  1901.             case PIX_FMT_RGB565:

  1902.                 c->yuv2packed1 = RENAME(yuv2rgb565_1);

  1903.                 c->yuv2packed2 = RENAME(yuv2rgb565_2);

  1904.                 break;

  1905.             case PIX_FMT_YUYV422:

  1906.                 c->yuv2packed1 = RENAME(yuv2yuyv422_1);

  1907.                 c->yuv2packed2 = RENAME(yuv2yuyv422_2);

  1908.                 break;

  1909.             default:

  1910.                 break;

  1911.             }

  1912.         }

  1913.     }

  1914. 

  1915.     if (c->srcBpc == 8 && c->dstBpc <= 10) {

  1916.     // Use the new MMX scaler if the MMX2 one can't be used (it is faster than the x86 ASM one).

  1917. #if COMPILE_TEMPLATE_MMX2

  1918.     if (c->flags & SWS_FAST_BILINEAR && c->canMMX2BeUsed)

  1919.     {

  1920.         c->hyscale_fast = RENAME(hyscale_fast);

  1921.         c->hcscale_fast = RENAME(hcscale_fast);

  1922.     } else {

  1923. #endif /* COMPILE_TEMPLATE_MMX2 */

  1924.         c->hyscale_fast = NULL;

  1925.         c->hcscale_fast = NULL;

  1926. #if COMPILE_TEMPLATE_MMX2

  1927.     }

  1928. #endif /* COMPILE_TEMPLATE_MMX2 */

  1929.     }

  1930. 

  1931. #if !COMPILE_TEMPLATE_MMX2

  1932.     switch(srcFormat) {

  1933.         case PIX_FMT_YUYV422  : c->chrToYV12 = RENAME(yuy2ToUV); break;

  1934.         case PIX_FMT_UYVY422  : c->chrToYV12 = RENAME(uyvyToUV); break;

  1935.         case PIX_FMT_NV12     : c->chrToYV12 = RENAME(nv12ToUV); break;

  1936.         case PIX_FMT_NV21     : c->chrToYV12 = RENAME(nv21ToUV); break;

  1937.         default: break;

  1938.     }

  1939. #endif /* !COMPILE_TEMPLATE_MMX2 */

  1940.     if (!c->chrSrcHSubSample) {

  1941.         switch(srcFormat) {

  1942.         case PIX_FMT_BGR24  : c->chrToYV12 = RENAME(bgr24ToUV); break;

  1943.         case PIX_FMT_RGB24  : c->chrToYV12 = RENAME(rgb24ToUV); break;

  1944.         default: break;

  1945.         }

  1946.     }

  1947. 

  1948.     switch (srcFormat) {

  1949. #if !COMPILE_TEMPLATE_MMX2

  1950.     case PIX_FMT_YUYV422  :

  1951.     case PIX_FMT_Y400A    : c->lumToYV12 = RENAME(yuy2ToY); break;

  1952.     case PIX_FMT_UYVY422  : c->lumToYV12 = RENAME(uyvyToY); break;

  1953. #endif /* !COMPILE_TEMPLATE_MMX2 */

  1954.     case PIX_FMT_BGR24    : c->lumToYV12 = RENAME(bgr24ToY); break;

  1955.     case PIX_FMT_RGB24    : c->lumToYV12 = RENAME(rgb24ToY); break;

  1956.     default: break;

  1957.     }

  1958. #if !COMPILE_TEMPLATE_MMX2

  1959.     if (c->alpPixBuf) {

  1960.         switch (srcFormat) {

  1961.         case PIX_FMT_Y400A  : c->alpToYV12 = RENAME(yuy2ToY); break;

  1962.         default: break;

  1963.         }

  1964.     }

  1965. #endif /* !COMPILE_TEMPLATE_MMX2 */

  1966. }