git.sesse.net Git - ffmpeg/blob - libswscale/x86/swscale_template.c

   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 #include <stdint.h>
  22
  23 #include "libavutil/x86/asm.h"
  24 #include "libswscale/swscale_internal.h"
  25
  26 #undef REAL_MOVNTQ
  27 #undef MOVNTQ
  28 #undef MOVNTQ2
  29 #undef PREFETCH
  30
  31
  32 #if COMPILE_TEMPLATE_MMXEXT
  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_MMXEXT
  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     filterSize--;
  75     __asm__ volatile(
  76         "movd %0, %%mm1\n\t"
  77         "punpcklwd %%mm1, %%mm1\n\t"
  78         "punpckldq %%mm1, %%mm1\n\t"
  79         "psllw        $3, %%mm1\n\t"
  80         "paddw     %%mm1, %%mm3\n\t"
  81         "paddw     %%mm1, %%mm4\n\t"
  82         "psraw        $4, %%mm3\n\t"
  83         "psraw        $4, %%mm4\n\t"
  84         ::"m"(filterSize)
  85      );
  86
  87     __asm__ volatile(\
  88         "movq    %%mm3, %%mm6\n\t"
  89         "movq    %%mm4, %%mm7\n\t"
  90         "movl %3, %%ecx\n\t"
  91         "mov                                 %0, %%"REG_d"  \n\t"\
  92         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
  93         ".p2align                             4             \n\t" /* FIXME Unroll? */\
  94         "1:                                                 \n\t"\
  95         "movq                      8(%%"REG_d"), %%mm0      \n\t" /* filterCoeff */\
  96         "movq                (%%"REG_S", %%"REG_c", 2), %%mm2      \n\t" /* srcData */\
  97         "movq               8(%%"REG_S", %%"REG_c", 2), %%mm5      \n\t" /* srcData */\
  98         "add                                $16, %%"REG_d"  \n\t"\
  99         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
 100         "test                         %%"REG_S", %%"REG_S"  \n\t"\
 101         "pmulhw                           %%mm0, %%mm2      \n\t"\
 102         "pmulhw                           %%mm0, %%mm5      \n\t"\
 103         "paddw                            %%mm2, %%mm3      \n\t"\
 104         "paddw                            %%mm5, %%mm4      \n\t"\
 105         " jnz                                1b             \n\t"\
 106         "psraw                               $3, %%mm3      \n\t"\
 107         "psraw                               $3, %%mm4      \n\t"\
 108         "packuswb                         %%mm4, %%mm3      \n\t"
 109         MOVNTQ2 "                         %%mm3, (%1, %%"REG_c")\n\t"
 110         "add                          $8, %%"REG_c"         \n\t"\
 111         "cmp                          %2, %%"REG_c"         \n\t"\
 112         "movq    %%mm6, %%mm3\n\t"
 113         "movq    %%mm7, %%mm4\n\t"
 114         "mov                                 %0, %%"REG_d"  \n\t"\
 115         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
 116         "jb                                  1b             \n\t"\
 117         :: "g" (filter),
 118            "r" (dest-offset), "g" ((x86_reg)(dstW+offset)), "m" (offset)
 119         : "%"REG_d, "%"REG_S, "%"REG_c
 120     );
 121 }
 122
 123 #define YSCALEYUV2PACKEDX_UV \
 124     __asm__ volatile(\
 125         "xor                   %%"REG_a", %%"REG_a"     \n\t"\
 126         ".p2align                      4                \n\t"\
 127         "nop                                            \n\t"\
 128         "1:                                             \n\t"\
 129         "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"     \n\t"\
 130         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 131         "movq      "VROUNDER_OFFSET"(%0), %%mm3         \n\t"\
 132         "movq                      %%mm3, %%mm4         \n\t"\
 133         ".p2align                      4                \n\t"\
 134         "2:                                             \n\t"\
 135         "movq               8(%%"REG_d"), %%mm0         \n\t" /* filterCoeff */\
 136         "movq     (%%"REG_S", %%"REG_a"), %%mm2         \n\t" /* UsrcData */\
 137         "add                          %6, %%"REG_S"     \n\t" \
 138         "movq     (%%"REG_S", %%"REG_a"), %%mm5         \n\t" /* VsrcData */\
 139         "add                         $16, %%"REG_d"     \n\t"\
 140         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 141         "pmulhw                    %%mm0, %%mm2         \n\t"\
 142         "pmulhw                    %%mm0, %%mm5         \n\t"\
 143         "paddw                     %%mm2, %%mm3         \n\t"\
 144         "paddw                     %%mm5, %%mm4         \n\t"\
 145         "test                  %%"REG_S", %%"REG_S"     \n\t"\
 146         " jnz                         2b                \n\t"\
 147
 148 #define YSCALEYUV2PACKEDX_YA(offset,coeff,src1,src2,dst1,dst2) \
 149     "lea                "offset"(%0), %%"REG_d"     \n\t"\
 150     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 151     "movq      "VROUNDER_OFFSET"(%0), "#dst1"       \n\t"\
 152     "movq                    "#dst1", "#dst2"       \n\t"\
 153     ".p2align                      4                \n\t"\
 154     "2:                                             \n\t"\
 155     "movq               8(%%"REG_d"), "#coeff"      \n\t" /* filterCoeff */\
 156     "movq  (%%"REG_S", %%"REG_a", 2), "#src1"       \n\t" /* Y1srcData */\
 157     "movq 8(%%"REG_S", %%"REG_a", 2), "#src2"       \n\t" /* Y2srcData */\
 158     "add                         $16, %%"REG_d"            \n\t"\
 159     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 160     "pmulhw                 "#coeff", "#src1"       \n\t"\
 161     "pmulhw                 "#coeff", "#src2"       \n\t"\
 162     "paddw                   "#src1", "#dst1"       \n\t"\
 163     "paddw                   "#src2", "#dst2"       \n\t"\
 164     "test                  %%"REG_S", %%"REG_S"     \n\t"\
 165     " jnz                         2b                \n\t"\
 166
 167 #define YSCALEYUV2PACKEDX \
 168     YSCALEYUV2PACKEDX_UV \
 169     YSCALEYUV2PACKEDX_YA(LUM_MMX_FILTER_OFFSET,%%mm0,%%mm2,%%mm5,%%mm1,%%mm7) \
 170
 171 #define YSCALEYUV2PACKEDX_END                     \
 172         :: "r" (&c->redDither),                   \
 173             "m" (dummy), "m" (dummy), "m" (dummy),\
 174             "r" (dest), "m" (dstW_reg), "m"(uv_off) \
 175             NAMED_CONSTRAINTS_ADD(bF8,bFC) \
 176         : "%"REG_a, "%"REG_d, "%"REG_S            \
 177     );
 178
 179 #define YSCALEYUV2PACKEDX_ACCURATE_UV \
 180     __asm__ volatile(\
 181         "xor %%"REG_a", %%"REG_a"                       \n\t"\
 182         ".p2align                      4                \n\t"\
 183         "nop                                            \n\t"\
 184         "1:                                             \n\t"\
 185         "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"     \n\t"\
 186         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 187         "pxor                      %%mm4, %%mm4         \n\t"\
 188         "pxor                      %%mm5, %%mm5         \n\t"\
 189         "pxor                      %%mm6, %%mm6         \n\t"\
 190         "pxor                      %%mm7, %%mm7         \n\t"\
 191         ".p2align                      4                \n\t"\
 192         "2:                                             \n\t"\
 193         "movq     (%%"REG_S", %%"REG_a"), %%mm0         \n\t" /* UsrcData */\
 194         "add                          %6, %%"REG_S"      \n\t" \
 195         "movq     (%%"REG_S", %%"REG_a"), %%mm2         \n\t" /* VsrcData */\
 196         "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S"     \n\t"\
 197         "movq     (%%"REG_S", %%"REG_a"), %%mm1         \n\t" /* UsrcData */\
 198         "movq                      %%mm0, %%mm3         \n\t"\
 199         "punpcklwd                 %%mm1, %%mm0         \n\t"\
 200         "punpckhwd                 %%mm1, %%mm3         \n\t"\
 201         "movq "STR(APCK_COEF)"(%%"REG_d"),%%mm1         \n\t" /* filterCoeff */\
 202         "pmaddwd                   %%mm1, %%mm0         \n\t"\
 203         "pmaddwd                   %%mm1, %%mm3         \n\t"\
 204         "paddd                     %%mm0, %%mm4         \n\t"\
 205         "paddd                     %%mm3, %%mm5         \n\t"\
 206         "add                          %6, %%"REG_S"      \n\t" \
 207         "movq     (%%"REG_S", %%"REG_a"), %%mm3         \n\t" /* VsrcData */\
 208         "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S"     \n\t"\
 209         "add           $"STR(APCK_SIZE)", %%"REG_d"     \n\t"\
 210         "test                  %%"REG_S", %%"REG_S"     \n\t"\
 211         "movq                      %%mm2, %%mm0         \n\t"\
 212         "punpcklwd                 %%mm3, %%mm2         \n\t"\
 213         "punpckhwd                 %%mm3, %%mm0         \n\t"\
 214         "pmaddwd                   %%mm1, %%mm2         \n\t"\
 215         "pmaddwd                   %%mm1, %%mm0         \n\t"\
 216         "paddd                     %%mm2, %%mm6         \n\t"\
 217         "paddd                     %%mm0, %%mm7         \n\t"\
 218         " jnz                         2b                \n\t"\
 219         "psrad                       $16, %%mm4         \n\t"\
 220         "psrad                       $16, %%mm5         \n\t"\
 221         "psrad                       $16, %%mm6         \n\t"\
 222         "psrad                       $16, %%mm7         \n\t"\
 223         "movq      "VROUNDER_OFFSET"(%0), %%mm0         \n\t"\
 224         "packssdw                  %%mm5, %%mm4         \n\t"\
 225         "packssdw                  %%mm7, %%mm6         \n\t"\
 226         "paddw                     %%mm0, %%mm4         \n\t"\
 227         "paddw                     %%mm0, %%mm6         \n\t"\
 228         "movq                      %%mm4, "U_TEMP"(%0)  \n\t"\
 229         "movq                      %%mm6, "V_TEMP"(%0)  \n\t"\
 230
 231 #define YSCALEYUV2PACKEDX_ACCURATE_YA(offset) \
 232     "lea                "offset"(%0), %%"REG_d"     \n\t"\
 233     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 234     "pxor                      %%mm1, %%mm1         \n\t"\
 235     "pxor                      %%mm5, %%mm5         \n\t"\
 236     "pxor                      %%mm7, %%mm7         \n\t"\
 237     "pxor                      %%mm6, %%mm6         \n\t"\
 238     ".p2align                      4                \n\t"\
 239     "2:                                             \n\t"\
 240     "movq  (%%"REG_S", %%"REG_a", 2), %%mm0         \n\t" /* Y1srcData */\
 241     "movq 8(%%"REG_S", %%"REG_a", 2), %%mm2         \n\t" /* Y2srcData */\
 242     "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S"     \n\t"\
 243     "movq  (%%"REG_S", %%"REG_a", 2), %%mm4         \n\t" /* Y1srcData */\
 244     "movq                      %%mm0, %%mm3         \n\t"\
 245     "punpcklwd                 %%mm4, %%mm0         \n\t"\
 246     "punpckhwd                 %%mm4, %%mm3         \n\t"\
 247     "movq "STR(APCK_COEF)"(%%"REG_d"), %%mm4         \n\t" /* filterCoeff */\
 248     "pmaddwd                   %%mm4, %%mm0         \n\t"\
 249     "pmaddwd                   %%mm4, %%mm3         \n\t"\
 250     "paddd                     %%mm0, %%mm1         \n\t"\
 251     "paddd                     %%mm3, %%mm5         \n\t"\
 252     "movq 8(%%"REG_S", %%"REG_a", 2), %%mm3         \n\t" /* Y2srcData */\
 253     "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S"     \n\t"\
 254     "add           $"STR(APCK_SIZE)", %%"REG_d"     \n\t"\
 255     "test                  %%"REG_S", %%"REG_S"     \n\t"\
 256     "movq                      %%mm2, %%mm0         \n\t"\
 257     "punpcklwd                 %%mm3, %%mm2         \n\t"\
 258     "punpckhwd                 %%mm3, %%mm0         \n\t"\
 259     "pmaddwd                   %%mm4, %%mm2         \n\t"\
 260     "pmaddwd                   %%mm4, %%mm0         \n\t"\
 261     "paddd                     %%mm2, %%mm7         \n\t"\
 262     "paddd                     %%mm0, %%mm6         \n\t"\
 263     " jnz                         2b                \n\t"\
 264     "psrad                       $16, %%mm1         \n\t"\
 265     "psrad                       $16, %%mm5         \n\t"\
 266     "psrad                       $16, %%mm7         \n\t"\
 267     "psrad                       $16, %%mm6         \n\t"\
 268     "movq      "VROUNDER_OFFSET"(%0), %%mm0         \n\t"\
 269     "packssdw                  %%mm5, %%mm1         \n\t"\
 270     "packssdw                  %%mm6, %%mm7         \n\t"\
 271     "paddw                     %%mm0, %%mm1         \n\t"\
 272     "paddw                     %%mm0, %%mm7         \n\t"\
 273     "movq               "U_TEMP"(%0), %%mm3         \n\t"\
 274     "movq               "V_TEMP"(%0), %%mm4         \n\t"\
 275
 276 #define YSCALEYUV2PACKEDX_ACCURATE \
 277     YSCALEYUV2PACKEDX_ACCURATE_UV \
 278     YSCALEYUV2PACKEDX_ACCURATE_YA(LUM_MMX_FILTER_OFFSET)
 279
 280 #define YSCALEYUV2RGBX \
 281     "psubw  "U_OFFSET"(%0), %%mm3       \n\t" /* (U-128)8*/\
 282     "psubw  "V_OFFSET"(%0), %%mm4       \n\t" /* (V-128)8*/\
 283     "movq            %%mm3, %%mm2       \n\t" /* (U-128)8*/\
 284     "movq            %%mm4, %%mm5       \n\t" /* (V-128)8*/\
 285     "pmulhw "UG_COEFF"(%0), %%mm3       \n\t"\
 286     "pmulhw "VG_COEFF"(%0), %%mm4       \n\t"\
 287     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
 288     "pmulhw "UB_COEFF"(%0), %%mm2       \n\t"\
 289     "pmulhw "VR_COEFF"(%0), %%mm5       \n\t"\
 290     "psubw  "Y_OFFSET"(%0), %%mm1       \n\t" /* 8(Y-16)*/\
 291     "psubw  "Y_OFFSET"(%0), %%mm7       \n\t" /* 8(Y-16)*/\
 292     "pmulhw  "Y_COEFF"(%0), %%mm1       \n\t"\
 293     "pmulhw  "Y_COEFF"(%0), %%mm7       \n\t"\
 294     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
 295     "paddw           %%mm3, %%mm4       \n\t"\
 296     "movq            %%mm2, %%mm0       \n\t"\
 297     "movq            %%mm5, %%mm6       \n\t"\
 298     "movq            %%mm4, %%mm3       \n\t"\
 299     "punpcklwd       %%mm2, %%mm2       \n\t"\
 300     "punpcklwd       %%mm5, %%mm5       \n\t"\
 301     "punpcklwd       %%mm4, %%mm4       \n\t"\
 302     "paddw           %%mm1, %%mm2       \n\t"\
 303     "paddw           %%mm1, %%mm5       \n\t"\
 304     "paddw           %%mm1, %%mm4       \n\t"\
 305     "punpckhwd       %%mm0, %%mm0       \n\t"\
 306     "punpckhwd       %%mm6, %%mm6       \n\t"\
 307     "punpckhwd       %%mm3, %%mm3       \n\t"\
 308     "paddw           %%mm7, %%mm0       \n\t"\
 309     "paddw           %%mm7, %%mm6       \n\t"\
 310     "paddw           %%mm7, %%mm3       \n\t"\
 311     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
 312     "packuswb        %%mm0, %%mm2       \n\t"\
 313     "packuswb        %%mm6, %%mm5       \n\t"\
 314     "packuswb        %%mm3, %%mm4       \n\t"\
 315
 316 #define REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) \
 317     "movq       "#b", "#q2"     \n\t" /* B */\
 318     "movq       "#r", "#t"      \n\t" /* R */\
 319     "punpcklbw  "#g", "#b"      \n\t" /* GBGBGBGB 0 */\
 320     "punpcklbw  "#a", "#r"      \n\t" /* ARARARAR 0 */\
 321     "punpckhbw  "#g", "#q2"     \n\t" /* GBGBGBGB 2 */\
 322     "punpckhbw  "#a", "#t"      \n\t" /* ARARARAR 2 */\
 323     "movq       "#b", "#q0"     \n\t" /* GBGBGBGB 0 */\
 324     "movq      "#q2", "#q3"     \n\t" /* GBGBGBGB 2 */\
 325     "punpcklwd  "#r", "#q0"     \n\t" /* ARGBARGB 0 */\
 326     "punpckhwd  "#r", "#b"      \n\t" /* ARGBARGB 1 */\
 327     "punpcklwd  "#t", "#q2"     \n\t" /* ARGBARGB 2 */\
 328     "punpckhwd  "#t", "#q3"     \n\t" /* ARGBARGB 3 */\
 329 \
 330     MOVNTQ(   q0,   (dst, index, 4))\
 331     MOVNTQ(    b,  8(dst, index, 4))\
 332     MOVNTQ(   q2, 16(dst, index, 4))\
 333     MOVNTQ(   q3, 24(dst, index, 4))\
 334 \
 335     "add      $8, "#index"      \n\t"\
 336     "cmp  "dstw", "#index"      \n\t"\
 337     " jb      1b                \n\t"
 338 #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)
 339
 340 static void RENAME(yuv2rgb32_X_ar)(SwsContext *c, const int16_t *lumFilter,
 341                                    const int16_t **lumSrc, int lumFilterSize,
 342                                    const int16_t *chrFilter, const int16_t **chrUSrc,
 343                                    const int16_t **chrVSrc,
 344                                    int chrFilterSize, const int16_t **alpSrc,
 345                                    uint8_t *dest, int dstW, int dstY)
 346 {
 347     x86_reg dummy=0;
 348     x86_reg dstW_reg = dstW;
 349     x86_reg uv_off = c->uv_offx2;
 350
 351     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
 352         YSCALEYUV2PACKEDX_ACCURATE
 353         YSCALEYUV2RGBX
 354         "movq                      %%mm2, "U_TEMP"(%0)  \n\t"
 355         "movq                      %%mm4, "V_TEMP"(%0)  \n\t"
 356         "movq                      %%mm5, "Y_TEMP"(%0)  \n\t"
 357         YSCALEYUV2PACKEDX_ACCURATE_YA(ALP_MMX_FILTER_OFFSET)
 358         "movq               "Y_TEMP"(%0), %%mm5         \n\t"
 359         "psraw                        $3, %%mm1         \n\t"
 360         "psraw                        $3, %%mm7         \n\t"
 361         "packuswb                  %%mm7, %%mm1         \n\t"
 362         WRITEBGR32(%4, "%5", %%REGa, %%mm3, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm2, %%mm6)
 363         YSCALEYUV2PACKEDX_END
 364     } else {
 365         YSCALEYUV2PACKEDX_ACCURATE
 366         YSCALEYUV2RGBX
 367         "pcmpeqd %%mm7, %%mm7 \n\t"
 368         WRITEBGR32(%4, "%5", %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
 369         YSCALEYUV2PACKEDX_END
 370     }
 371 }
 372
 373 static void RENAME(yuv2rgb32_X)(SwsContext *c, const int16_t *lumFilter,
 374                                 const int16_t **lumSrc, int lumFilterSize,
 375                                 const int16_t *chrFilter, const int16_t **chrUSrc,
 376                                 const int16_t **chrVSrc,
 377                                 int chrFilterSize, const int16_t **alpSrc,
 378                                 uint8_t *dest, int dstW, int dstY)
 379 {
 380     x86_reg dummy=0;
 381     x86_reg dstW_reg = dstW;
 382     x86_reg uv_off = c->uv_offx2;
 383
 384     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
 385         YSCALEYUV2PACKEDX
 386         YSCALEYUV2RGBX
 387         YSCALEYUV2PACKEDX_YA(ALP_MMX_FILTER_OFFSET, %%mm0, %%mm3, %%mm6, %%mm1, %%mm7)
 388         "psraw                        $3, %%mm1         \n\t"
 389         "psraw                        $3, %%mm7         \n\t"
 390         "packuswb                  %%mm7, %%mm1         \n\t"
 391         WRITEBGR32(%4, "%5", %%REGa, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
 392         YSCALEYUV2PACKEDX_END
 393     } else {
 394         YSCALEYUV2PACKEDX
 395         YSCALEYUV2RGBX
 396         "pcmpeqd %%mm7, %%mm7 \n\t"
 397         WRITEBGR32(%4, "%5", %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
 398         YSCALEYUV2PACKEDX_END
 399     }
 400 }
 401
 402 #define REAL_WRITERGB16(dst, dstw, index) \
 403     "pand "MANGLE(bF8)", %%mm2  \n\t" /* B */\
 404     "pand "MANGLE(bFC)", %%mm4  \n\t" /* G */\
 405     "pand "MANGLE(bF8)", %%mm5  \n\t" /* R */\
 406     "psrlq           $3, %%mm2  \n\t"\
 407 \
 408     "movq         %%mm2, %%mm1  \n\t"\
 409     "movq         %%mm4, %%mm3  \n\t"\
 410 \
 411     "punpcklbw    %%mm7, %%mm3  \n\t"\
 412     "punpcklbw    %%mm5, %%mm2  \n\t"\
 413     "punpckhbw    %%mm7, %%mm4  \n\t"\
 414     "punpckhbw    %%mm5, %%mm1  \n\t"\
 415 \
 416     "psllq           $3, %%mm3  \n\t"\
 417     "psllq           $3, %%mm4  \n\t"\
 418 \
 419     "por          %%mm3, %%mm2  \n\t"\
 420     "por          %%mm4, %%mm1  \n\t"\
 421 \
 422     MOVNTQ(%%mm2,  (dst, index, 2))\
 423     MOVNTQ(%%mm1, 8(dst, index, 2))\
 424 \
 425     "add             $8, "#index"   \n\t"\
 426     "cmp         "dstw", "#index"   \n\t"\
 427     " jb             1b             \n\t"
 428 #define WRITERGB16(dst, dstw, index)  REAL_WRITERGB16(dst, dstw, index)
 429
 430 static void RENAME(yuv2rgb565_X_ar)(SwsContext *c, const int16_t *lumFilter,
 431                                     const int16_t **lumSrc, int lumFilterSize,
 432                                     const int16_t *chrFilter, const int16_t **chrUSrc,
 433                                     const int16_t **chrVSrc,
 434                                     int chrFilterSize, const int16_t **alpSrc,
 435                                     uint8_t *dest, int dstW, int dstY)
 436 {
 437     x86_reg dummy=0;
 438     x86_reg dstW_reg = dstW;
 439     x86_reg uv_off = c->uv_offx2;
 440
 441     YSCALEYUV2PACKEDX_ACCURATE
 442     YSCALEYUV2RGBX
 443     "pxor %%mm7, %%mm7 \n\t"
 444     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 445 #ifdef DITHER1XBPP
 446     "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"
 447     "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"
 448     "paddusb "RED_DITHER"(%0), %%mm5\n\t"
 449 #endif
 450     WRITERGB16(%4, "%5", %%REGa)
 451     YSCALEYUV2PACKEDX_END
 452 }
 453
 454 static void RENAME(yuv2rgb565_X)(SwsContext *c, const int16_t *lumFilter,
 455                                  const int16_t **lumSrc, int lumFilterSize,
 456                                  const int16_t *chrFilter, const int16_t **chrUSrc,
 457                                  const int16_t **chrVSrc,
 458                                  int chrFilterSize, const int16_t **alpSrc,
 459                                  uint8_t *dest, int dstW, int dstY)
 460 {
 461     x86_reg dummy=0;
 462     x86_reg dstW_reg = dstW;
 463     x86_reg uv_off = c->uv_offx2;
 464
 465     YSCALEYUV2PACKEDX
 466     YSCALEYUV2RGBX
 467     "pxor %%mm7, %%mm7 \n\t"
 468     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 469 #ifdef DITHER1XBPP
 470     "paddusb "BLUE_DITHER"(%0), %%mm2  \n\t"
 471     "paddusb "GREEN_DITHER"(%0), %%mm4  \n\t"
 472     "paddusb "RED_DITHER"(%0), %%mm5  \n\t"
 473 #endif
 474     WRITERGB16(%4, "%5", %%REGa)
 475     YSCALEYUV2PACKEDX_END
 476 }
 477
 478 #define REAL_WRITERGB15(dst, dstw, index) \
 479     "pand "MANGLE(bF8)", %%mm2  \n\t" /* B */\
 480     "pand "MANGLE(bF8)", %%mm4  \n\t" /* G */\
 481     "pand "MANGLE(bF8)", %%mm5  \n\t" /* R */\
 482     "psrlq           $3, %%mm2  \n\t"\
 483     "psrlq           $1, %%mm5  \n\t"\
 484 \
 485     "movq         %%mm2, %%mm1  \n\t"\
 486     "movq         %%mm4, %%mm3  \n\t"\
 487 \
 488     "punpcklbw    %%mm7, %%mm3  \n\t"\
 489     "punpcklbw    %%mm5, %%mm2  \n\t"\
 490     "punpckhbw    %%mm7, %%mm4  \n\t"\
 491     "punpckhbw    %%mm5, %%mm1  \n\t"\
 492 \
 493     "psllq           $2, %%mm3  \n\t"\
 494     "psllq           $2, %%mm4  \n\t"\
 495 \
 496     "por          %%mm3, %%mm2  \n\t"\
 497     "por          %%mm4, %%mm1  \n\t"\
 498 \
 499     MOVNTQ(%%mm2,  (dst, index, 2))\
 500     MOVNTQ(%%mm1, 8(dst, index, 2))\
 501 \
 502     "add             $8, "#index"   \n\t"\
 503     "cmp         "dstw", "#index"   \n\t"\
 504     " jb             1b             \n\t"
 505 #define WRITERGB15(dst, dstw, index)  REAL_WRITERGB15(dst, dstw, index)
 506
 507 static void RENAME(yuv2rgb555_X_ar)(SwsContext *c, const int16_t *lumFilter,
 508                                     const int16_t **lumSrc, int lumFilterSize,
 509                                     const int16_t *chrFilter, const int16_t **chrUSrc,
 510                                     const int16_t **chrVSrc,
 511                                     int chrFilterSize, const int16_t **alpSrc,
 512                                     uint8_t *dest, int dstW, int dstY)
 513 {
 514     x86_reg dummy=0;
 515     x86_reg dstW_reg = dstW;
 516     x86_reg uv_off = c->uv_offx2;
 517
 518     YSCALEYUV2PACKEDX_ACCURATE
 519     YSCALEYUV2RGBX
 520     "pxor %%mm7, %%mm7 \n\t"
 521     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 522 #ifdef DITHER1XBPP
 523     "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"
 524     "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"
 525     "paddusb "RED_DITHER"(%0), %%mm5\n\t"
 526 #endif
 527     WRITERGB15(%4, "%5", %%REGa)
 528     YSCALEYUV2PACKEDX_END
 529 }
 530
 531 static void RENAME(yuv2rgb555_X)(SwsContext *c, const int16_t *lumFilter,
 532                                  const int16_t **lumSrc, int lumFilterSize,
 533                                  const int16_t *chrFilter, const int16_t **chrUSrc,
 534                                  const int16_t **chrVSrc,
 535                                  int chrFilterSize, const int16_t **alpSrc,
 536                                  uint8_t *dest, int dstW, int dstY)
 537 {
 538     x86_reg dummy=0;
 539     x86_reg dstW_reg = dstW;
 540     x86_reg uv_off = c->uv_offx2;
 541
 542     YSCALEYUV2PACKEDX
 543     YSCALEYUV2RGBX
 544     "pxor %%mm7, %%mm7 \n\t"
 545     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 546 #ifdef DITHER1XBPP
 547     "paddusb "BLUE_DITHER"(%0), %%mm2  \n\t"
 548     "paddusb "GREEN_DITHER"(%0), %%mm4  \n\t"
 549     "paddusb "RED_DITHER"(%0), %%mm5  \n\t"
 550 #endif
 551     WRITERGB15(%4, "%5", %%REGa)
 552     YSCALEYUV2PACKEDX_END
 553 }
 554
 555 #define WRITEBGR24MMX(dst, dstw, index) \
 556     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\
 557     "movq      %%mm2, %%mm1     \n\t" /* B */\
 558     "movq      %%mm5, %%mm6     \n\t" /* R */\
 559     "punpcklbw %%mm4, %%mm2     \n\t" /* GBGBGBGB 0 */\
 560     "punpcklbw %%mm7, %%mm5     \n\t" /* 0R0R0R0R 0 */\
 561     "punpckhbw %%mm4, %%mm1     \n\t" /* GBGBGBGB 2 */\
 562     "punpckhbw %%mm7, %%mm6     \n\t" /* 0R0R0R0R 2 */\
 563     "movq      %%mm2, %%mm0     \n\t" /* GBGBGBGB 0 */\
 564     "movq      %%mm1, %%mm3     \n\t" /* GBGBGBGB 2 */\
 565     "punpcklwd %%mm5, %%mm0     \n\t" /* 0RGB0RGB 0 */\
 566     "punpckhwd %%mm5, %%mm2     \n\t" /* 0RGB0RGB 1 */\
 567     "punpcklwd %%mm6, %%mm1     \n\t" /* 0RGB0RGB 2 */\
 568     "punpckhwd %%mm6, %%mm3     \n\t" /* 0RGB0RGB 3 */\
 569 \
 570     "movq      %%mm0, %%mm4     \n\t" /* 0RGB0RGB 0 */\
 571     "movq      %%mm2, %%mm6     \n\t" /* 0RGB0RGB 1 */\
 572     "movq      %%mm1, %%mm5     \n\t" /* 0RGB0RGB 2 */\
 573     "movq      %%mm3, %%mm7     \n\t" /* 0RGB0RGB 3 */\
 574 \
 575     "psllq       $40, %%mm0     \n\t" /* RGB00000 0 */\
 576     "psllq       $40, %%mm2     \n\t" /* RGB00000 1 */\
 577     "psllq       $40, %%mm1     \n\t" /* RGB00000 2 */\
 578     "psllq       $40, %%mm3     \n\t" /* RGB00000 3 */\
 579 \
 580     "punpckhdq %%mm4, %%mm0     \n\t" /* 0RGBRGB0 0 */\
 581     "punpckhdq %%mm6, %%mm2     \n\t" /* 0RGBRGB0 1 */\
 582     "punpckhdq %%mm5, %%mm1     \n\t" /* 0RGBRGB0 2 */\
 583     "punpckhdq %%mm7, %%mm3     \n\t" /* 0RGBRGB0 3 */\
 584 \
 585     "psrlq        $8, %%mm0     \n\t" /* 00RGBRGB 0 */\
 586     "movq      %%mm2, %%mm6     \n\t" /* 0RGBRGB0 1 */\
 587     "psllq       $40, %%mm2     \n\t" /* GB000000 1 */\
 588     "por       %%mm2, %%mm0     \n\t" /* GBRGBRGB 0 */\
 589     MOVNTQ(%%mm0, (dst))\
 590 \
 591     "psrlq       $24, %%mm6     \n\t" /* 0000RGBR 1 */\
 592     "movq      %%mm1, %%mm5     \n\t" /* 0RGBRGB0 2 */\
 593     "psllq       $24, %%mm1     \n\t" /* BRGB0000 2 */\
 594     "por       %%mm1, %%mm6     \n\t" /* BRGBRGBR 1 */\
 595     MOVNTQ(%%mm6, 8(dst))\
 596 \
 597     "psrlq       $40, %%mm5     \n\t" /* 000000RG 2 */\
 598     "psllq        $8, %%mm3     \n\t" /* RGBRGB00 3 */\
 599     "por       %%mm3, %%mm5     \n\t" /* RGBRGBRG 2 */\
 600     MOVNTQ(%%mm5, 16(dst))\
 601 \
 602     "add         $24, "#dst"    \n\t"\
 603 \
 604     "add          $8, "#index"  \n\t"\
 605     "cmp      "dstw", "#index"  \n\t"\
 606     " jb          1b            \n\t"
 607
 608 #define WRITEBGR24MMXEXT(dst, dstw, index) \
 609     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\
 610     "movq "MANGLE(ff_M24A)", %%mm0 \n\t"\
 611     "movq "MANGLE(ff_M24C)", %%mm7 \n\t"\
 612     "pshufw $0x50, %%mm2, %%mm1 \n\t" /* B3 B2 B3 B2  B1 B0 B1 B0 */\
 613     "pshufw $0x50, %%mm4, %%mm3 \n\t" /* G3 G2 G3 G2  G1 G0 G1 G0 */\
 614     "pshufw $0x00, %%mm5, %%mm6 \n\t" /* R1 R0 R1 R0  R1 R0 R1 R0 */\
 615 \
 616     "pand   %%mm0, %%mm1        \n\t" /*    B2        B1       B0 */\
 617     "pand   %%mm0, %%mm3        \n\t" /*    G2        G1       G0 */\
 618     "pand   %%mm7, %%mm6        \n\t" /*       R1        R0       */\
 619 \
 620     "psllq     $8, %%mm3        \n\t" /* G2        G1       G0    */\
 621     "por    %%mm1, %%mm6        \n\t"\
 622     "por    %%mm3, %%mm6        \n\t"\
 623     MOVNTQ(%%mm6, (dst))\
 624 \
 625     "psrlq     $8, %%mm4        \n\t" /* 00 G7 G6 G5  G4 G3 G2 G1 */\
 626     "pshufw $0xA5, %%mm2, %%mm1 \n\t" /* B5 B4 B5 B4  B3 B2 B3 B2 */\
 627     "pshufw $0x55, %%mm4, %%mm3 \n\t" /* G4 G3 G4 G3  G4 G3 G4 G3 */\
 628     "pshufw $0xA5, %%mm5, %%mm6 \n\t" /* R5 R4 R5 R4  R3 R2 R3 R2 */\
 629 \
 630     "pand "MANGLE(ff_M24B)", %%mm1 \n\t" /* B5       B4        B3    */\
 631     "pand   %%mm7, %%mm3        \n\t" /*       G4        G3       */\
 632     "pand   %%mm0, %%mm6        \n\t" /*    R4        R3       R2 */\
 633 \
 634     "por    %%mm1, %%mm3        \n\t" /* B5    G4 B4     G3 B3    */\
 635     "por    %%mm3, %%mm6        \n\t"\
 636     MOVNTQ(%%mm6, 8(dst))\
 637 \
 638     "pshufw $0xFF, %%mm2, %%mm1 \n\t" /* B7 B6 B7 B6  B7 B6 B6 B7 */\
 639     "pshufw $0xFA, %%mm4, %%mm3 \n\t" /* 00 G7 00 G7  G6 G5 G6 G5 */\
 640     "pshufw $0xFA, %%mm5, %%mm6 \n\t" /* R7 R6 R7 R6  R5 R4 R5 R4 */\
 641 \
 642     "pand   %%mm7, %%mm1        \n\t" /*       B7        B6       */\
 643     "pand   %%mm0, %%mm3        \n\t" /*    G7        G6       G5 */\
 644     "pand "MANGLE(ff_M24B)", %%mm6 \n\t" /* R7       R6        R5    */\
 645 \
 646     "por    %%mm1, %%mm3        \n\t"\
 647     "por    %%mm3, %%mm6        \n\t"\
 648     MOVNTQ(%%mm6, 16(dst))\
 649 \
 650     "add      $24, "#dst"       \n\t"\
 651 \
 652     "add       $8, "#index"     \n\t"\
 653     "cmp   "dstw", "#index"     \n\t"\
 654     " jb       1b               \n\t"
 655
 656 #if COMPILE_TEMPLATE_MMXEXT
 657 #undef WRITEBGR24
 658 #define WRITEBGR24(dst, dstw, index)  WRITEBGR24MMXEXT(dst, dstw, index)
 659 #else
 660 #undef WRITEBGR24
 661 #define WRITEBGR24(dst, dstw, index)  WRITEBGR24MMX(dst, dstw, index)
 662 #endif
 663
 664 #if HAVE_6REGS
 665 static void RENAME(yuv2bgr24_X_ar)(SwsContext *c, const int16_t *lumFilter,
 666                                    const int16_t **lumSrc, int lumFilterSize,
 667                                    const int16_t *chrFilter, const int16_t **chrUSrc,
 668                                    const int16_t **chrVSrc,
 669                                    int chrFilterSize, const int16_t **alpSrc,
 670                                    uint8_t *dest, int dstW, int dstY)
 671 {
 672     x86_reg dummy=0;
 673     x86_reg dstW_reg = dstW;
 674     x86_reg uv_off = c->uv_offx2;
 675
 676     YSCALEYUV2PACKEDX_ACCURATE
 677     YSCALEYUV2RGBX
 678     "pxor %%mm7, %%mm7 \n\t"
 679     "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"\n\t" //FIXME optimize
 680     "add %4, %%"REG_c"                        \n\t"
 681     WRITEBGR24(%%REGc, "%5", %%REGa)
 682     :: "r" (&c->redDither),
 683        "m" (dummy), "m" (dummy), "m" (dummy),
 684        "r" (dest), "m" (dstW_reg), "m"(uv_off)
 685        NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B)
 686     : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S
 687     );
 688 }
 689
 690 static void RENAME(yuv2bgr24_X)(SwsContext *c, const int16_t *lumFilter,
 691                                 const int16_t **lumSrc, int lumFilterSize,
 692                                 const int16_t *chrFilter, const int16_t **chrUSrc,
 693                                 const int16_t **chrVSrc,
 694                                 int chrFilterSize, const int16_t **alpSrc,
 695                                 uint8_t *dest, int dstW, int dstY)
 696 {
 697     x86_reg dummy=0;
 698     x86_reg dstW_reg = dstW;
 699     x86_reg uv_off = c->uv_offx2;
 700
 701     YSCALEYUV2PACKEDX
 702     YSCALEYUV2RGBX
 703     "pxor                    %%mm7, %%mm7       \n\t"
 704     "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"   \n\t" //FIXME optimize
 705     "add                        %4, %%"REG_c"   \n\t"
 706     WRITEBGR24(%%REGc, "%5", %%REGa)
 707     :: "r" (&c->redDither),
 708        "m" (dummy), "m" (dummy), "m" (dummy),
 709        "r" (dest),  "m" (dstW_reg), "m"(uv_off)
 710        NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B)
 711     : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S
 712     );
 713 }
 714 #endif /* HAVE_6REGS */
 715
 716 #define REAL_WRITEYUY2(dst, dstw, index) \
 717     "packuswb  %%mm3, %%mm3     \n\t"\
 718     "packuswb  %%mm4, %%mm4     \n\t"\
 719     "packuswb  %%mm7, %%mm1     \n\t"\
 720     "punpcklbw %%mm4, %%mm3     \n\t"\
 721     "movq      %%mm1, %%mm7     \n\t"\
 722     "punpcklbw %%mm3, %%mm1     \n\t"\
 723     "punpckhbw %%mm3, %%mm7     \n\t"\
 724 \
 725     MOVNTQ(%%mm1, (dst, index, 2))\
 726     MOVNTQ(%%mm7, 8(dst, index, 2))\
 727 \
 728     "add          $8, "#index"  \n\t"\
 729     "cmp      "dstw", "#index"  \n\t"\
 730     " jb          1b            \n\t"
 731 #define WRITEYUY2(dst, dstw, index)  REAL_WRITEYUY2(dst, dstw, index)
 732
 733 static void RENAME(yuv2yuyv422_X_ar)(SwsContext *c, const int16_t *lumFilter,
 734                                      const int16_t **lumSrc, int lumFilterSize,
 735                                      const int16_t *chrFilter, const int16_t **chrUSrc,
 736                                      const int16_t **chrVSrc,
 737                                      int chrFilterSize, const int16_t **alpSrc,
 738                                      uint8_t *dest, int dstW, int dstY)
 739 {
 740     x86_reg dummy=0;
 741     x86_reg dstW_reg = dstW;
 742     x86_reg uv_off = c->uv_offx2;
 743
 744     YSCALEYUV2PACKEDX_ACCURATE
 745     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 746     "psraw $3, %%mm3    \n\t"
 747     "psraw $3, %%mm4    \n\t"
 748     "psraw $3, %%mm1    \n\t"
 749     "psraw $3, %%mm7    \n\t"
 750     WRITEYUY2(%4, "%5", %%REGa)
 751     YSCALEYUV2PACKEDX_END
 752 }
 753
 754 static void RENAME(yuv2yuyv422_X)(SwsContext *c, const int16_t *lumFilter,
 755                                   const int16_t **lumSrc, int lumFilterSize,
 756                                   const int16_t *chrFilter, const int16_t **chrUSrc,
 757                                   const int16_t **chrVSrc,
 758                                   int chrFilterSize, const int16_t **alpSrc,
 759                                   uint8_t *dest, int dstW, int dstY)
 760 {
 761     x86_reg dummy=0;
 762     x86_reg dstW_reg = dstW;
 763     x86_reg uv_off = c->uv_offx2;
 764
 765     YSCALEYUV2PACKEDX
 766     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 767     "psraw $3, %%mm3    \n\t"
 768     "psraw $3, %%mm4    \n\t"
 769     "psraw $3, %%mm1    \n\t"
 770     "psraw $3, %%mm7    \n\t"
 771     WRITEYUY2(%4, "%5", %%REGa)
 772     YSCALEYUV2PACKEDX_END
 773 }
 774
 775 #define REAL_YSCALEYUV2RGB_UV(index, c) \
 776     "xor            "#index", "#index"  \n\t"\
 777     ".p2align              4            \n\t"\
 778     "1:                                 \n\t"\
 779     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
 780     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
 781     "add "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
 782     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
 783     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
 784     "sub "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
 785     "psubw             %%mm3, %%mm2     \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
 786     "psubw             %%mm4, %%mm5     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
 787     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0    \n\t"\
 788     "pmulhw            %%mm0, %%mm2     \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
 789     "pmulhw            %%mm0, %%mm5     \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\
 790     "psraw                $4, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
 791     "psraw                $4, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
 792     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
 793     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\
 794     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\
 795     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\
 796     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\
 797     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\
 798     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\
 799     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\
 800     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
 801
 802 #define REAL_YSCALEYUV2RGB_YA(index, c, b1, b2) \
 803     "movq  ("#b1", "#index", 2), %%mm0     \n\t" /*buf0[eax]*/\
 804     "movq  ("#b2", "#index", 2), %%mm1     \n\t" /*buf1[eax]*/\
 805     "movq 8("#b1", "#index", 2), %%mm6     \n\t" /*buf0[eax]*/\
 806     "movq 8("#b2", "#index", 2), %%mm7     \n\t" /*buf1[eax]*/\
 807     "psubw             %%mm1, %%mm0     \n\t" /* buf0[eax] - buf1[eax]*/\
 808     "psubw             %%mm7, %%mm6     \n\t" /* buf0[eax] - buf1[eax]*/\
 809     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
 810     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
 811     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 812     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 813     "paddw             %%mm0, %%mm1     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
 814     "paddw             %%mm6, %%mm7     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
 815
 816 #define REAL_YSCALEYUV2RGB_COEFF(c) \
 817     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\
 818     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\
 819     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\
 820     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\
 821     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\
 822     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\
 823     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
 824     "paddw             %%mm3, %%mm4     \n\t"\
 825     "movq              %%mm2, %%mm0     \n\t"\
 826     "movq              %%mm5, %%mm6     \n\t"\
 827     "movq              %%mm4, %%mm3     \n\t"\
 828     "punpcklwd         %%mm2, %%mm2     \n\t"\
 829     "punpcklwd         %%mm5, %%mm5     \n\t"\
 830     "punpcklwd         %%mm4, %%mm4     \n\t"\
 831     "paddw             %%mm1, %%mm2     \n\t"\
 832     "paddw             %%mm1, %%mm5     \n\t"\
 833     "paddw             %%mm1, %%mm4     \n\t"\
 834     "punpckhwd         %%mm0, %%mm0     \n\t"\
 835     "punpckhwd         %%mm6, %%mm6     \n\t"\
 836     "punpckhwd         %%mm3, %%mm3     \n\t"\
 837     "paddw             %%mm7, %%mm0     \n\t"\
 838     "paddw             %%mm7, %%mm6     \n\t"\
 839     "paddw             %%mm7, %%mm3     \n\t"\
 840     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
 841     "packuswb          %%mm0, %%mm2     \n\t"\
 842     "packuswb          %%mm6, %%mm5     \n\t"\
 843     "packuswb          %%mm3, %%mm4     \n\t"\
 844
 845 #define YSCALEYUV2RGB_YA(index, c, b1, b2) REAL_YSCALEYUV2RGB_YA(index, c, b1, b2)
 846
 847 #define YSCALEYUV2RGB(index, c) \
 848     REAL_YSCALEYUV2RGB_UV(index, c) \
 849     REAL_YSCALEYUV2RGB_YA(index, c, %0, %1) \
 850     REAL_YSCALEYUV2RGB_COEFF(c)
 851
 852 /**
 853  * vertical bilinear scale YV12 to RGB
 854  */
 855 static void RENAME(yuv2rgb32_2)(SwsContext *c, const int16_t *buf[2],
 856                                 const int16_t *ubuf[2], const int16_t *vbuf[2],
 857                                 const int16_t *abuf[2], uint8_t *dest,
 858                                 int dstW, int yalpha, int uvalpha, int y)
 859 {
 860     const int16_t *buf0  = buf[0],  *buf1  = buf[1],
 861                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
 862
 863     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
 864         const int16_t *abuf0 = abuf[0], *abuf1 = abuf[1];
 865 #if ARCH_X86_64
 866         __asm__ volatile(
 867             YSCALEYUV2RGB(%%r8, %5)
 868             YSCALEYUV2RGB_YA(%%r8, %5, %6, %7)
 869             "psraw                  $3, %%mm1       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
 870             "psraw                  $3, %%mm7       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
 871             "packuswb            %%mm7, %%mm1       \n\t"
 872             WRITEBGR32(%4, DSTW_OFFSET"(%5)", %%r8, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
 873             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "r" (dest),
 874                "a" (&c->redDither),
 875                "r" (abuf0), "r" (abuf1)
 876             : "%r8"
 877         );
 878 #else
 879         c->u_temp=(intptr_t)abuf0;
 880         c->v_temp=(intptr_t)abuf1;
 881         __asm__ volatile(
 882             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
 883             "mov        %4, %%"REG_b"               \n\t"
 884             "push %%"REG_BP"                        \n\t"
 885             YSCALEYUV2RGB(%%REGBP, %5)
 886             "push                   %0              \n\t"
 887             "push                   %1              \n\t"
 888             "mov          "U_TEMP"(%5), %0          \n\t"
 889             "mov          "V_TEMP"(%5), %1          \n\t"
 890             YSCALEYUV2RGB_YA(%%REGBP, %5, %0, %1)
 891             "psraw                  $3, %%mm1       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
 892             "psraw                  $3, %%mm7       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
 893             "packuswb            %%mm7, %%mm1       \n\t"
 894             "pop                    %1              \n\t"
 895             "pop                    %0              \n\t"
 896             WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
 897             "pop %%"REG_BP"                         \n\t"
 898             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
 899             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
 900                "a" (&c->redDither)
 901         );
 902 #endif
 903     } else {
 904         __asm__ volatile(
 905             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
 906             "mov        %4, %%"REG_b"               \n\t"
 907             "push %%"REG_BP"                        \n\t"
 908             YSCALEYUV2RGB(%%REGBP, %5)
 909             "pcmpeqd %%mm7, %%mm7                   \n\t"
 910             WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
 911             "pop %%"REG_BP"                         \n\t"
 912             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
 913             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
 914                "a" (&c->redDither)
 915         );
 916     }
 917 }
 918
 919 static void RENAME(yuv2bgr24_2)(SwsContext *c, const int16_t *buf[2],
 920                                 const int16_t *ubuf[2], const int16_t *vbuf[2],
 921                                 const int16_t *abuf[2], uint8_t *dest,
 922                                 int dstW, int yalpha, int uvalpha, int y)
 923 {
 924     const int16_t *buf0  = buf[0],  *buf1  = buf[1],
 925                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
 926
 927     __asm__ volatile(
 928         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
 929         "mov        %4, %%"REG_b"               \n\t"
 930         "push %%"REG_BP"                        \n\t"
 931         YSCALEYUV2RGB(%%REGBP, %5)
 932         "pxor    %%mm7, %%mm7                   \n\t"
 933         WRITEBGR24(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
 934         "pop %%"REG_BP"                         \n\t"
 935         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
 936         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
 937            "a" (&c->redDither)
 938            NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B)
 939     );
 940 }
 941
 942 static void RENAME(yuv2rgb555_2)(SwsContext *c, const int16_t *buf[2],
 943                                  const int16_t *ubuf[2], const int16_t *vbuf[2],
 944                                  const int16_t *abuf[2], uint8_t *dest,
 945                                  int dstW, int yalpha, int uvalpha, int y)
 946 {
 947     const int16_t *buf0  = buf[0],  *buf1  = buf[1],
 948                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
 949
 950     __asm__ volatile(
 951         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
 952         "mov        %4, %%"REG_b"               \n\t"
 953         "push %%"REG_BP"                        \n\t"
 954         YSCALEYUV2RGB(%%REGBP, %5)
 955         "pxor    %%mm7, %%mm7                   \n\t"
 956         /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 957 #ifdef DITHER1XBPP
 958         "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
 959         "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
 960         "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
 961 #endif
 962         WRITERGB15(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
 963         "pop %%"REG_BP"                         \n\t"
 964         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
 965         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
 966            "a" (&c->redDither)
 967            NAMED_CONSTRAINTS_ADD(bF8)
 968     );
 969 }
 970
 971 static void RENAME(yuv2rgb565_2)(SwsContext *c, const int16_t *buf[2],
 972                                  const int16_t *ubuf[2], const int16_t *vbuf[2],
 973                                  const int16_t *abuf[2], uint8_t *dest,
 974                                  int dstW, int yalpha, int uvalpha, int y)
 975 {
 976     const int16_t *buf0  = buf[0],  *buf1  = buf[1],
 977                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
 978
 979     __asm__ volatile(
 980         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
 981         "mov        %4, %%"REG_b"               \n\t"
 982         "push %%"REG_BP"                        \n\t"
 983         YSCALEYUV2RGB(%%REGBP, %5)
 984         "pxor    %%mm7, %%mm7                   \n\t"
 985         /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
 986 #ifdef DITHER1XBPP
 987         "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
 988         "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
 989         "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
 990 #endif
 991         WRITERGB16(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
 992         "pop %%"REG_BP"                         \n\t"
 993         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
 994         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
 995            "a" (&c->redDither)
 996            NAMED_CONSTRAINTS_ADD(bF8,bFC)
 997     );
 998 }
 999
1000 #define REAL_YSCALEYUV2PACKED(index, c) \
1001     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0              \n\t"\
1002     "movq "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm1              \n\t"\
1003     "psraw                $3, %%mm0                           \n\t"\
1004     "psraw                $3, %%mm1                           \n\t"\
1005     "movq              %%mm0, "CHR_MMX_FILTER_OFFSET"+8("#c") \n\t"\
1006     "movq              %%mm1, "LUM_MMX_FILTER_OFFSET"+8("#c") \n\t"\
1007     "xor            "#index", "#index"                        \n\t"\
1008     ".p2align              4            \n\t"\
1009     "1:                                 \n\t"\
1010     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
1011     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
1012     "add "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1013     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
1014     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
1015     "sub "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1016     "psubw             %%mm3, %%mm2     \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
1017     "psubw             %%mm4, %%mm5     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
1018     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0    \n\t"\
1019     "pmulhw            %%mm0, %%mm2     \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
1020     "pmulhw            %%mm0, %%mm5     \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\
1021     "psraw                $7, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
1022     "psraw                $7, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
1023     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
1024     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\
1025     "movq  (%0, "#index", 2), %%mm0     \n\t" /*buf0[eax]*/\
1026     "movq  (%1, "#index", 2), %%mm1     \n\t" /*buf1[eax]*/\
1027     "movq 8(%0, "#index", 2), %%mm6     \n\t" /*buf0[eax]*/\
1028     "movq 8(%1, "#index", 2), %%mm7     \n\t" /*buf1[eax]*/\
1029     "psubw             %%mm1, %%mm0     \n\t" /* buf0[eax] - buf1[eax]*/\
1030     "psubw             %%mm7, %%mm6     \n\t" /* buf0[eax] - buf1[eax]*/\
1031     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
1032     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
1033     "psraw                $7, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1034     "psraw                $7, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1035     "paddw             %%mm0, %%mm1     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
1036     "paddw             %%mm6, %%mm7     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
1037
1038 #define YSCALEYUV2PACKED(index, c)  REAL_YSCALEYUV2PACKED(index, c)
1039
1040 static void RENAME(yuv2yuyv422_2)(SwsContext *c, const int16_t *buf[2],
1041                                   const int16_t *ubuf[2], const int16_t *vbuf[2],
1042                                   const int16_t *abuf[2], uint8_t *dest,
1043                                   int dstW, int yalpha, int uvalpha, int y)
1044 {
1045     const int16_t *buf0  = buf[0],  *buf1  = buf[1],
1046                   *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1047
1048     __asm__ volatile(
1049         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1050         "mov %4, %%"REG_b"                        \n\t"
1051         "push %%"REG_BP"                        \n\t"
1052         YSCALEYUV2PACKED(%%REGBP, %5)
1053         WRITEYUY2(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1054         "pop %%"REG_BP"                         \n\t"
1055         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1056         :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1057            "a" (&c->redDither)
1058     );
1059 }
1060
1061 #define REAL_YSCALEYUV2RGB1(index, c) \
1062     "xor            "#index", "#index"  \n\t"\
1063     ".p2align              4            \n\t"\
1064     "1:                                 \n\t"\
1065     "movq     (%2, "#index"), %%mm3     \n\t" /* uvbuf0[eax]*/\
1066     "add "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1067     "movq     (%2, "#index"), %%mm4     \n\t" /* uvbuf0[eax+2048]*/\
1068     "sub "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1069     "psraw                $4, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
1070     "psraw                $4, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
1071     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\
1072     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\
1073     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\
1074     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\
1075     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\
1076     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\
1077     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
1078     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
1079     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
1080     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1081     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1082     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\
1083     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\
1084     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\
1085     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\
1086     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\
1087     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\
1088     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
1089     "paddw             %%mm3, %%mm4     \n\t"\
1090     "movq              %%mm2, %%mm0     \n\t"\
1091     "movq              %%mm5, %%mm6     \n\t"\
1092     "movq              %%mm4, %%mm3     \n\t"\
1093     "punpcklwd         %%mm2, %%mm2     \n\t"\
1094     "punpcklwd         %%mm5, %%mm5     \n\t"\
1095     "punpcklwd         %%mm4, %%mm4     \n\t"\
1096     "paddw             %%mm1, %%mm2     \n\t"\
1097     "paddw             %%mm1, %%mm5     \n\t"\
1098     "paddw             %%mm1, %%mm4     \n\t"\
1099     "punpckhwd         %%mm0, %%mm0     \n\t"\
1100     "punpckhwd         %%mm6, %%mm6     \n\t"\
1101     "punpckhwd         %%mm3, %%mm3     \n\t"\
1102     "paddw             %%mm7, %%mm0     \n\t"\
1103     "paddw             %%mm7, %%mm6     \n\t"\
1104     "paddw             %%mm7, %%mm3     \n\t"\
1105     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
1106     "packuswb          %%mm0, %%mm2     \n\t"\
1107     "packuswb          %%mm6, %%mm5     \n\t"\
1108     "packuswb          %%mm3, %%mm4     \n\t"\
1109
1110 #define YSCALEYUV2RGB1(index, c)  REAL_YSCALEYUV2RGB1(index, c)
1111
1112 // do vertical chrominance interpolation
1113 #define REAL_YSCALEYUV2RGB1b(index, c) \
1114     "xor            "#index", "#index"  \n\t"\
1115     ".p2align              4            \n\t"\
1116     "1:                                 \n\t"\
1117     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
1118     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
1119     "add "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1120     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
1121     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
1122     "sub "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1123     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
1124     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
1125     "psrlw                $5, %%mm3     \n\t" /*FIXME might overflow*/\
1126     "psrlw                $5, %%mm4     \n\t" /*FIXME might overflow*/\
1127     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\
1128     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\
1129     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\
1130     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\
1131     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\
1132     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\
1133     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
1134     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
1135     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
1136     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1137     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1138     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\
1139     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\
1140     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\
1141     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\
1142     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\
1143     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\
1144     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
1145     "paddw             %%mm3, %%mm4     \n\t"\
1146     "movq              %%mm2, %%mm0     \n\t"\
1147     "movq              %%mm5, %%mm6     \n\t"\
1148     "movq              %%mm4, %%mm3     \n\t"\
1149     "punpcklwd         %%mm2, %%mm2     \n\t"\
1150     "punpcklwd         %%mm5, %%mm5     \n\t"\
1151     "punpcklwd         %%mm4, %%mm4     \n\t"\
1152     "paddw             %%mm1, %%mm2     \n\t"\
1153     "paddw             %%mm1, %%mm5     \n\t"\
1154     "paddw             %%mm1, %%mm4     \n\t"\
1155     "punpckhwd         %%mm0, %%mm0     \n\t"\
1156     "punpckhwd         %%mm6, %%mm6     \n\t"\
1157     "punpckhwd         %%mm3, %%mm3     \n\t"\
1158     "paddw             %%mm7, %%mm0     \n\t"\
1159     "paddw             %%mm7, %%mm6     \n\t"\
1160     "paddw             %%mm7, %%mm3     \n\t"\
1161     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
1162     "packuswb          %%mm0, %%mm2     \n\t"\
1163     "packuswb          %%mm6, %%mm5     \n\t"\
1164     "packuswb          %%mm3, %%mm4     \n\t"\
1165
1166 #define YSCALEYUV2RGB1b(index, c)  REAL_YSCALEYUV2RGB1b(index, c)
1167
1168 #define REAL_YSCALEYUV2RGB1_ALPHA(index) \
1169     "movq  (%1, "#index", 2), %%mm7     \n\t" /* abuf0[index  ]     */\
1170     "movq 8(%1, "#index", 2), %%mm1     \n\t" /* abuf0[index+4]     */\
1171     "psraw                $7, %%mm7     \n\t" /* abuf0[index  ] >>7 */\
1172     "psraw                $7, %%mm1     \n\t" /* abuf0[index+4] >>7 */\
1173     "packuswb          %%mm1, %%mm7     \n\t"
1174 #define YSCALEYUV2RGB1_ALPHA(index) REAL_YSCALEYUV2RGB1_ALPHA(index)
1175
1176 /**
1177  * YV12 to RGB without scaling or interpolating
1178  */
1179 static void RENAME(yuv2rgb32_1)(SwsContext *c, const int16_t *buf0,
1180                                 const int16_t *ubuf[2], const int16_t *vbuf[2],
1181                                 const int16_t *abuf0, uint8_t *dest,
1182                                 int dstW, int uvalpha, int y)
1183 {
1184     const int16_t *ubuf0 = ubuf[0];
1185     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1186
1187     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1188         const int16_t *ubuf1 = ubuf[0];
1189         if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1190             __asm__ volatile(
1191                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1192                 "mov        %4, %%"REG_b"               \n\t"
1193                 "push %%"REG_BP"                        \n\t"
1194                 YSCALEYUV2RGB1(%%REGBP, %5)
1195                 YSCALEYUV2RGB1_ALPHA(%%REGBP)
1196                 WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1197                 "pop %%"REG_BP"                         \n\t"
1198                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1199                 :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1200                    "a" (&c->redDither)
1201             );
1202         } else {
1203             __asm__ volatile(
1204                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1205                 "mov        %4, %%"REG_b"               \n\t"
1206                 "push %%"REG_BP"                        \n\t"
1207                 YSCALEYUV2RGB1(%%REGBP, %5)
1208                 "pcmpeqd %%mm7, %%mm7                   \n\t"
1209                 WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1210                 "pop %%"REG_BP"                         \n\t"
1211                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1212                 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1213                    "a" (&c->redDither)
1214             );
1215         }
1216     } else {
1217         const int16_t *ubuf1 = ubuf[1];
1218         if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1219             __asm__ volatile(
1220                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1221                 "mov        %4, %%"REG_b"               \n\t"
1222                 "push %%"REG_BP"                        \n\t"
1223                 YSCALEYUV2RGB1b(%%REGBP, %5)
1224                 YSCALEYUV2RGB1_ALPHA(%%REGBP)
1225                 WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1226                 "pop %%"REG_BP"                         \n\t"
1227                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1228                 :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1229                    "a" (&c->redDither)
1230             );
1231         } else {
1232             __asm__ volatile(
1233                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1234                 "mov        %4, %%"REG_b"               \n\t"
1235                 "push %%"REG_BP"                        \n\t"
1236                 YSCALEYUV2RGB1b(%%REGBP, %5)
1237                 "pcmpeqd %%mm7, %%mm7                   \n\t"
1238                 WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1239                 "pop %%"REG_BP"                         \n\t"
1240                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1241                 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1242                    "a" (&c->redDither)
1243             );
1244         }
1245     }
1246 }
1247
1248 static void RENAME(yuv2bgr24_1)(SwsContext *c, const int16_t *buf0,
1249                                 const int16_t *ubuf[2], const int16_t *vbuf[2],
1250                                 const int16_t *abuf0, uint8_t *dest,
1251                                 int dstW, int uvalpha, int y)
1252 {
1253     const int16_t *ubuf0 = ubuf[0];
1254     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1255
1256     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1257         const int16_t *ubuf1 = ubuf[0];
1258         __asm__ volatile(
1259             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1260             "mov        %4, %%"REG_b"               \n\t"
1261             "push %%"REG_BP"                        \n\t"
1262             YSCALEYUV2RGB1(%%REGBP, %5)
1263             "pxor    %%mm7, %%mm7                   \n\t"
1264             WRITEBGR24(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1265             "pop %%"REG_BP"                         \n\t"
1266             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1267             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1268                "a" (&c->redDither)
1269                NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B)
1270         );
1271     } else {
1272         const int16_t *ubuf1 = ubuf[1];
1273         __asm__ volatile(
1274             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1275             "mov        %4, %%"REG_b"               \n\t"
1276             "push %%"REG_BP"                        \n\t"
1277             YSCALEYUV2RGB1b(%%REGBP, %5)
1278             "pxor    %%mm7, %%mm7                   \n\t"
1279             WRITEBGR24(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1280             "pop %%"REG_BP"                         \n\t"
1281             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1282             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1283                "a" (&c->redDither)
1284                NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B)
1285         );
1286     }
1287 }
1288
1289 static void RENAME(yuv2rgb555_1)(SwsContext *c, const int16_t *buf0,
1290                                  const int16_t *ubuf[2], const int16_t *vbuf[2],
1291                                  const int16_t *abuf0, uint8_t *dest,
1292                                  int dstW, int uvalpha, int y)
1293 {
1294     const int16_t *ubuf0 = ubuf[0];
1295     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1296
1297     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1298         const int16_t *ubuf1 = ubuf[0];
1299         __asm__ volatile(
1300             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1301             "mov        %4, %%"REG_b"               \n\t"
1302             "push %%"REG_BP"                        \n\t"
1303             YSCALEYUV2RGB1(%%REGBP, %5)
1304             "pxor    %%mm7, %%mm7                   \n\t"
1305             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1306 #ifdef DITHER1XBPP
1307             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1308             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1309             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1310 #endif
1311             WRITERGB15(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1312             "pop %%"REG_BP"                         \n\t"
1313             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1314             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1315                "a" (&c->redDither)
1316                NAMED_CONSTRAINTS_ADD(bF8)
1317         );
1318     } else {
1319         const int16_t *ubuf1 = ubuf[1];
1320         __asm__ volatile(
1321             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1322             "mov        %4, %%"REG_b"               \n\t"
1323             "push %%"REG_BP"                        \n\t"
1324             YSCALEYUV2RGB1b(%%REGBP, %5)
1325             "pxor    %%mm7, %%mm7                   \n\t"
1326             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1327 #ifdef DITHER1XBPP
1328             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1329             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1330             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1331 #endif
1332             WRITERGB15(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1333             "pop %%"REG_BP"                         \n\t"
1334             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1335             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1336                "a" (&c->redDither)
1337                NAMED_CONSTRAINTS_ADD(bF8)
1338         );
1339     }
1340 }
1341
1342 static void RENAME(yuv2rgb565_1)(SwsContext *c, const int16_t *buf0,
1343                                  const int16_t *ubuf[2], const int16_t *vbuf[2],
1344                                  const int16_t *abuf0, uint8_t *dest,
1345                                  int dstW, int uvalpha, int y)
1346 {
1347     const int16_t *ubuf0 = ubuf[0];
1348     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1349
1350     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1351         const int16_t *ubuf1 = ubuf[0];
1352         __asm__ volatile(
1353             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1354             "mov        %4, %%"REG_b"               \n\t"
1355             "push %%"REG_BP"                        \n\t"
1356             YSCALEYUV2RGB1(%%REGBP, %5)
1357             "pxor    %%mm7, %%mm7                   \n\t"
1358             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1359 #ifdef DITHER1XBPP
1360             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1361             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1362             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1363 #endif
1364             WRITERGB16(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1365             "pop %%"REG_BP"                         \n\t"
1366             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1367             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1368                "a" (&c->redDither)
1369                NAMED_CONSTRAINTS_ADD(bF8,bFC)
1370         );
1371     } else {
1372         const int16_t *ubuf1 = ubuf[1];
1373         __asm__ volatile(
1374             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1375             "mov        %4, %%"REG_b"               \n\t"
1376             "push %%"REG_BP"                        \n\t"
1377             YSCALEYUV2RGB1b(%%REGBP, %5)
1378             "pxor    %%mm7, %%mm7                   \n\t"
1379             /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1380 #ifdef DITHER1XBPP
1381             "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1382             "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1383             "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1384 #endif
1385             WRITERGB16(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1386             "pop %%"REG_BP"                         \n\t"
1387             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1388             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1389                "a" (&c->redDither)
1390                NAMED_CONSTRAINTS_ADD(bF8,bFC)
1391         );
1392     }
1393 }
1394
1395 #define REAL_YSCALEYUV2PACKED1(index, c) \
1396     "xor            "#index", "#index"  \n\t"\
1397     ".p2align              4            \n\t"\
1398     "1:                                 \n\t"\
1399     "movq     (%2, "#index"), %%mm3     \n\t" /* uvbuf0[eax]*/\
1400     "add "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1401     "movq     (%2, "#index"), %%mm4     \n\t" /* uvbuf0[eax+2048]*/\
1402     "sub "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1403     "psraw                $7, %%mm3     \n\t" \
1404     "psraw                $7, %%mm4     \n\t" \
1405     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
1406     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
1407     "psraw                $7, %%mm1     \n\t" \
1408     "psraw                $7, %%mm7     \n\t" \
1409
1410 #define YSCALEYUV2PACKED1(index, c)  REAL_YSCALEYUV2PACKED1(index, c)
1411
1412 #define REAL_YSCALEYUV2PACKED1b(index, c) \
1413     "xor "#index", "#index"             \n\t"\
1414     ".p2align              4            \n\t"\
1415     "1:                                 \n\t"\
1416     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
1417     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
1418     "add "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1419     "movq     (%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
1420     "movq     (%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
1421     "sub "UV_OFF_BYTE"("#c"), "#index"  \n\t" \
1422     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
1423     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
1424     "psrlw                $8, %%mm3     \n\t" \
1425     "psrlw                $8, %%mm4     \n\t" \
1426     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
1427     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
1428     "psraw                $7, %%mm1     \n\t" \
1429     "psraw                $7, %%mm7     \n\t"
1430 #define YSCALEYUV2PACKED1b(index, c)  REAL_YSCALEYUV2PACKED1b(index, c)
1431
1432 static void RENAME(yuv2yuyv422_1)(SwsContext *c, const int16_t *buf0,
1433                                   const int16_t *ubuf[2], const int16_t *vbuf[2],
1434                                   const int16_t *abuf0, uint8_t *dest,
1435                                   int dstW, int uvalpha, int y)
1436 {
1437     const int16_t *ubuf0 = ubuf[0];
1438     const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1439
1440     if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1441         const int16_t *ubuf1 = ubuf[0];
1442         __asm__ volatile(
1443             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1444             "mov        %4, %%"REG_b"               \n\t"
1445             "push %%"REG_BP"                        \n\t"
1446             YSCALEYUV2PACKED1(%%REGBP, %5)
1447             WRITEYUY2(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1448             "pop %%"REG_BP"                         \n\t"
1449             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1450             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1451                "a" (&c->redDither)
1452         );
1453     } else {
1454         const int16_t *ubuf1 = ubuf[1];
1455         __asm__ volatile(
1456             "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1457             "mov        %4, %%"REG_b"               \n\t"
1458             "push %%"REG_BP"                        \n\t"
1459             YSCALEYUV2PACKED1b(%%REGBP, %5)
1460             WRITEYUY2(%%REGb, DSTW_OFFSET"(%5)", %%REGBP)
1461             "pop %%"REG_BP"                         \n\t"
1462             "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1463             :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1464                "a" (&c->redDither)
1465         );
1466     }
1467 }
1468 static av_cold void RENAME(sws_init_swscale)(SwsContext *c)
1469 {
1470     enum AVPixelFormat dstFormat = c->dstFormat;
1471
1472     c->use_mmx_vfilter= 0;
1473     if (!is16BPS(dstFormat) && !is9_OR_10BPS(dstFormat) && dstFormat != AV_PIX_FMT_NV12
1474         && dstFormat != AV_PIX_FMT_NV21 && !(c->flags & SWS_BITEXACT)) {
1475             if (c->flags & SWS_ACCURATE_RND) {
1476                 if (!(c->flags & SWS_FULL_CHR_H_INT)) {
1477                     switch (c->dstFormat) {
1478                     case AV_PIX_FMT_RGB32:   c->yuv2packedX = RENAME(yuv2rgb32_X_ar);   break;
1479 #if HAVE_6REGS
1480                     case AV_PIX_FMT_BGR24:   c->yuv2packedX = RENAME(yuv2bgr24_X_ar);   break;
1481 #endif
1482                     case AV_PIX_FMT_RGB555:  c->yuv2packedX = RENAME(yuv2rgb555_X_ar);  break;
1483                     case AV_PIX_FMT_RGB565:  c->yuv2packedX = RENAME(yuv2rgb565_X_ar);  break;
1484                     case AV_PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X_ar); break;
1485                     default: break;
1486                     }
1487                 }
1488             } else {
1489                 c->use_mmx_vfilter= 1;
1490                 c->yuv2planeX = RENAME(yuv2yuvX    );
1491                 if (!(c->flags & SWS_FULL_CHR_H_INT)) {
1492                     switch (c->dstFormat) {
1493                     case AV_PIX_FMT_RGB32:   c->yuv2packedX = RENAME(yuv2rgb32_X);   break;
1494 #if HAVE_6REGS
1495                     case AV_PIX_FMT_BGR24:   c->yuv2packedX = RENAME(yuv2bgr24_X);   break;
1496 #endif
1497                     case AV_PIX_FMT_RGB555:  c->yuv2packedX = RENAME(yuv2rgb555_X);  break;
1498                     case AV_PIX_FMT_RGB565:  c->yuv2packedX = RENAME(yuv2rgb565_X);  break;
1499                     case AV_PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X); break;
1500                     default: break;
1501                     }
1502                 }
1503             }
1504         if (!(c->flags & SWS_FULL_CHR_H_INT)) {
1505             switch (c->dstFormat) {
1506             case AV_PIX_FMT_RGB32:
1507                 c->yuv2packed1 = RENAME(yuv2rgb32_1);
1508                 c->yuv2packed2 = RENAME(yuv2rgb32_2);
1509                 break;
1510             case AV_PIX_FMT_BGR24:
1511                 c->yuv2packed1 = RENAME(yuv2bgr24_1);
1512                 c->yuv2packed2 = RENAME(yuv2bgr24_2);
1513                 break;
1514             case AV_PIX_FMT_RGB555:
1515                 c->yuv2packed1 = RENAME(yuv2rgb555_1);
1516                 c->yuv2packed2 = RENAME(yuv2rgb555_2);
1517                 break;
1518             case AV_PIX_FMT_RGB565:
1519                 c->yuv2packed1 = RENAME(yuv2rgb565_1);
1520                 c->yuv2packed2 = RENAME(yuv2rgb565_2);
1521                 break;
1522             case AV_PIX_FMT_YUYV422:
1523                 c->yuv2packed1 = RENAME(yuv2yuyv422_1);
1524                 c->yuv2packed2 = RENAME(yuv2yuyv422_2);
1525                 break;
1526             default:
1527                 break;
1528             }
1529         }
1530     }
1531
1532     if (c->srcBpc == 8 && c->dstBpc <= 14) {
1533     // Use the new MMX scaler if the MMXEXT one can't be used (it is faster than the x86 ASM one).
1534 #if COMPILE_TEMPLATE_MMXEXT
1535     if (c->flags & SWS_FAST_BILINEAR && c->canMMXEXTBeUsed) {
1536         c->hyscale_fast = ff_hyscale_fast_mmxext;
1537         c->hcscale_fast = ff_hcscale_fast_mmxext;
1538     } else {
1539 #endif /* COMPILE_TEMPLATE_MMXEXT */
1540         c->hyscale_fast = NULL;
1541         c->hcscale_fast = NULL;
1542 #if COMPILE_TEMPLATE_MMXEXT
1543     }
1544 #endif /* COMPILE_TEMPLATE_MMXEXT */
1545     }
1546 }