src/video/SDL_yuv_mmx.c
changeset 887 b4b64bb88f2f
parent 769 b8d311d90021
child 946 8520712f8ef0
equal deleted inserted replaced
886:05c551e5bc64 887:b4b64bb88f2f
   118 		 "movd (%%ebx), %%mm1\n"   //         0  0  0  0  v3 v2 v1 v0
   118 		 "movd (%%ebx), %%mm1\n"   //         0  0  0  0  v3 v2 v1 v0
   119 		 "pxor %%mm7,%%mm7\n"      //         00 00 00 00 00 00 00 00
   119 		 "pxor %%mm7,%%mm7\n"      //         00 00 00 00 00 00 00 00
   120 		 "movd (%2), %%mm2\n"           //    0  0  0  0 l3 l2 l1 l0
   120 		 "movd (%2), %%mm2\n"           //    0  0  0  0 l3 l2 l1 l0
   121 		 "punpcklbw %%mm7,%%mm1\n" //         0  v3 0  v2 00 v1 00 v0
   121 		 "punpcklbw %%mm7,%%mm1\n" //         0  v3 0  v2 00 v1 00 v0
   122 		 "punpckldq %%mm1,%%mm1\n" //         00 v1 00 v0 00 v1 00 v0
   122 		 "punpckldq %%mm1,%%mm1\n" //         00 v1 00 v0 00 v1 00 v0
   123 		 "psubw _MMX_0080w,%%mm1\n"  // mm1-128:r1 r1 r0 r0 r1 r1 r0 r0 
   123 		 "psubw %[_MMX_0080w],%%mm1\n"  // mm1-128:r1 r1 r0 r0 r1 r1 r0 r0 
   124 
   124 
   125 		 // create Cr_g (result in mm0)
   125 		 // create Cr_g (result in mm0)
   126 		 "movq %%mm1,%%mm0\n"           // r1 r1 r0 r0 r1 r1 r0 r0
   126 		 "movq %%mm1,%%mm0\n"           // r1 r1 r0 r0 r1 r1 r0 r0
   127 		 "pmullw _MMX_VgrnRGB,%%mm0\n"// red*-46dec=0.7136*64
   127 		 "pmullw %[_MMX_VgrnRGB],%%mm0\n"// red*-46dec=0.7136*64
   128 		 "pmullw _MMX_VredRGB,%%mm1\n"// red*89dec=1.4013*64
   128 		 "pmullw %[_MMX_VredRGB],%%mm1\n"// red*89dec=1.4013*64
   129 		 "psraw  $6, %%mm0\n"           // red=red/64
   129 		 "psraw  $6, %%mm0\n"           // red=red/64
   130 		 "psraw  $6, %%mm1\n"           // red=red/64
   130 		 "psraw  $6, %%mm1\n"           // red=red/64
   131 		 
   131 		 
   132 		 // create L1 L2 (result in mm2,mm4)
   132 		 // create L1 L2 (result in mm2,mm4)
   133 		 // L2=lum+cols
   133 		 // L2=lum+cols
   134 		 "movq (%2,%4),%%mm3\n"         //    0  0  0  0 L3 L2 L1 L0
   134 		 "movq (%2,%4),%%mm3\n"         //    0  0  0  0 L3 L2 L1 L0
   135 		 "punpckldq %%mm3,%%mm2\n"      //   L3 L2 L1 L0 l3 l2 l1 l0
   135 		 "punpckldq %%mm3,%%mm2\n"      //   L3 L2 L1 L0 l3 l2 l1 l0
   136 		 "movq %%mm2,%%mm4\n"           //   L3 L2 L1 L0 l3 l2 l1 l0
   136 		 "movq %%mm2,%%mm4\n"           //   L3 L2 L1 L0 l3 l2 l1 l0
   137 		 "pand _MMX_FF00w,%%mm2\n"      //   L3 0  L1  0 l3  0 l1  0
   137 		 "pand %[_MMX_FF00w],%%mm2\n"      //   L3 0  L1  0 l3  0 l1  0
   138 		 "pand _MMX_00FFw,%%mm4\n"      //   0  L2  0 L0  0 l2  0 l0
   138 		 "pand %[_MMX_00FFw],%%mm4\n"      //   0  L2  0 L0  0 l2  0 l0
   139 		 "psrlw $8,%%mm2\n"             //   0  L3  0 L1  0 l3  0 l1
   139 		 "psrlw $8,%%mm2\n"             //   0  L3  0 L1  0 l3  0 l1
   140 
   140 
   141 		 // create R (result in mm6)
   141 		 // create R (result in mm6)
   142 		 "movq %%mm2,%%mm5\n"           //   0 L3  0 L1  0 l3  0 l1
   142 		 "movq %%mm2,%%mm5\n"           //   0 L3  0 L1  0 l3  0 l1
   143 		 "movq %%mm4,%%mm6\n"           //   0 L2  0 L0  0 l2  0 l0
   143 		 "movq %%mm4,%%mm6\n"           //   0 L2  0 L0  0 l2  0 l0
   150 
   150 
   151 		 // create Cb (result in mm1)
   151 		 // create Cb (result in mm1)
   152 		 "movd (%1), %%mm1\n"      //         0  0  0  0  u3 u2 u1 u0
   152 		 "movd (%1), %%mm1\n"      //         0  0  0  0  u3 u2 u1 u0
   153 		 "punpcklbw %%mm7,%%mm1\n" //         0  u3 0  u2 00 u1 00 u0
   153 		 "punpcklbw %%mm7,%%mm1\n" //         0  u3 0  u2 00 u1 00 u0
   154 		 "punpckldq %%mm1,%%mm1\n" //         00 u1 00 u0 00 u1 00 u0
   154 		 "punpckldq %%mm1,%%mm1\n" //         00 u1 00 u0 00 u1 00 u0
   155 		 "psubw _MMX_0080w,%%mm1\n"  // mm1-128:u1 u1 u0 u0 u1 u1 u0 u0 
   155 		 "psubw %[_MMX_0080w],%%mm1\n"  // mm1-128:u1 u1 u0 u0 u1 u1 u0 u0 
   156 		 // create Cb_g (result in mm5)
   156 		 // create Cb_g (result in mm5)
   157 		 "movq %%mm1,%%mm5\n"            // u1 u1 u0 u0 u1 u1 u0 u0
   157 		 "movq %%mm1,%%mm5\n"            // u1 u1 u0 u0 u1 u1 u0 u0
   158 		 "pmullw _MMX_UgrnRGB,%%mm5\n"    // blue*-109dec=1.7129*64
   158 		 "pmullw %[_MMX_UgrnRGB],%%mm5\n"    // blue*-109dec=1.7129*64
   159 		 "pmullw _MMX_UbluRGB,%%mm1\n"    // blue*114dec=1.78125*64
   159 		 "pmullw %[_MMX_UbluRGB],%%mm1\n"    // blue*114dec=1.78125*64
   160 		 "psraw  $6, %%mm5\n"            // blue=red/64
   160 		 "psraw  $6, %%mm5\n"            // blue=red/64
   161 		 "psraw  $6, %%mm1\n"            // blue=blue/64
   161 		 "psraw  $6, %%mm1\n"            // blue=blue/64
   162 
   162 
   163 		 // create G (result in mm7)
   163 		 // create G (result in mm7)
   164 		 "movq %%mm2,%%mm3\n"      //   0  L3  0 L1  0 l3  0 l1
   164 		 "movq %%mm2,%%mm3\n"      //   0  L3  0 L1  0 l3  0 l1
   236 		 "jl             1b\n"
   236 		 "jl             1b\n"
   237 		 "emms\n"
   237 		 "emms\n"
   238 		 "popl %%ebx\n"
   238 		 "popl %%ebx\n"
   239 		 :
   239 		 :
   240 		 : "m" (cr), "r"(cb),"r"(lum),
   240 		 : "m" (cr), "r"(cb),"r"(lum),
   241 		 "r"(row1),"r"(cols),"r"(row2),"m"(x),"m"(y),"m"(mod)
   241 		 "r"(row1),"r"(cols),"r"(row2),"m"(x),"m"(y),"m"(mod),
   242 		 : "%ebx"
   242          [_MMX_0080w] "m" (*_MMX_0080w),
       
   243          [_MMX_00FFw] "m" (*_MMX_00FFw),
       
   244          [_MMX_FF00w] "m" (*_MMX_FF00w),
       
   245          [_MMX_VgrnRGB] "m" (*_MMX_VgrnRGB),
       
   246          [_MMX_VredRGB] "m" (*_MMX_VredRGB),
       
   247          [_MMX_UgrnRGB] "m" (*_MMX_UgrnRGB),
       
   248          [_MMX_UbluRGB] "m" (*_MMX_UbluRGB)
   243 		 );
   249 		 );
   244 }
   250 }
   245 
   251 
   246 void Color565DitherYV12MMX1X( int *colortab, Uint32 *rgb_2_pix,
   252 void Color565DitherYV12MMX1X( int *colortab, Uint32 *rgb_2_pix,
   247                              unsigned char *lum, unsigned char *cr,
   253                              unsigned char *lum, unsigned char *cr,
   411 	 "jl             1b\n"
   417 	 "jl             1b\n"
   412          "emms\n"
   418          "emms\n"
   413 	 "popl %%ebx\n"
   419 	 "popl %%ebx\n"
   414          :
   420          :
   415          :"m" (cr), "r"(cb),"r"(lum),
   421          :"m" (cr), "r"(cb),"r"(lum),
   416 	 "r"(row1),"r"(cols),"r"(row2),"m"(x),"m"(y),"m"(mod)
   422 	 "r"(row1),"r"(cols),"r"(row2),"m"(x),"m"(y),"m"(mod),
   417 	 : "%ebx"
   423      [_MMX_0080w] "m" (*_MMX_0080w),
       
   424  [_MMX_Ugrn565] "m" (*_MMX_Ugrn565),
       
   425  [_MMX_Ublu5x5] "m" (*_MMX_Ublu5x5),
       
   426  [_MMX_00FFw] "m" (*_MMX_00FFw),
       
   427  [_MMX_Vgrn565] "m" (*_MMX_Vgrn565),
       
   428  [_MMX_Vred5x5] "m" (*_MMX_Vred5x5),
       
   429  [_MMX_Ycoeff] "m" (*_MMX_Ycoeff),
       
   430  [_MMX_red565] "m" (*_MMX_red565),
       
   431  [_MMX_grn565] "m" (*_MMX_grn565)
   418          );
   432          );
   419 }
   433 }
   420 
   434 
   421 #endif /* GCC i386 inline assembly */
   435 #endif /* GCC i386 inline assembly */