mmx yuy2 output
authorMichael Niedermayer <michaelni@gmx.at>
Sun, 13 Oct 2002 17:23:02 +0000 (17:23 +0000)
committerMichael Niedermayer <michaelni@gmx.at>
Sun, 13 Oct 2002 17:23:02 +0000 (17:23 +0000)
Originally committed as revision 7724 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc

postproc/swscale.c
postproc/swscale_template.c

index dbb4a8be4320a9e06c98f213f04e11f5309a88b6..3b2b8d36eb01828180c068554379122c63b32c9c 100644 (file)
@@ -421,7 +421,7 @@ static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilt
 }
 
 
-#define YSCALE_YUV_2_X_C(type) \
+#define YSCALE_YUV_2_PACKEDX_C(type) \
                for(i=0; i<(dstW>>1); i++){\
                        int j;\
                        int Y1=0;\
@@ -458,12 +458,12 @@ static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilt
                        }
                         
 #define YSCALE_YUV_2_RGBX_C(type) \
-                       YSCALE_YUV_2_X_C(type)\
+                       YSCALE_YUV_2_PACKEDX_C(type)\
                        r = c->table_rV[V];\
                        g = c->table_gU[U] + c->table_gV[V];\
                        b = c->table_bU[U];\
 
-#define YSCALE_YUV_2_2_C \
+#define YSCALE_YUV_2_PACKED2_C \
                for(i=0; i<(dstW>>1); i++){\
                        const int i2= 2*i;\
                        int Y1= (buf0[i2  ]*yalpha1+buf1[i2  ]*yalpha)>>19;\
@@ -472,13 +472,13 @@ static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilt
                        int V= (uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19;\
 
 #define YSCALE_YUV_2_RGB2_C(type) \
-                       YSCALE_YUV_2_2_C\
+                       YSCALE_YUV_2_PACKED2_C\
                        type *r, *b, *g;\
                        r = c->table_rV[V];\
                        g = c->table_gU[U] + c->table_gV[V];\
                        b = c->table_bU[U];\
 
-#define YSCALE_YUV_2_1_C \
+#define YSCALE_YUV_2_PACKED1_C \
                for(i=0; i<(dstW>>1); i++){\
                        const int i2= 2*i;\
                        int Y1= buf0[i2  ]>>7;\
@@ -487,13 +487,13 @@ static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilt
                        int V= (uvbuf1[i+2048])>>7;\
 
 #define YSCALE_YUV_2_RGB1_C(type) \
-                       YSCALE_YUV_2_1_C\
+                       YSCALE_YUV_2_PACKED1_C\
                        type *r, *b, *g;\
                        r = c->table_rV[V];\
                        g = c->table_gU[U] + c->table_gV[V];\
                        b = c->table_bU[U];\
 
-#define YSCALE_YUV_2_1B_C \
+#define YSCALE_YUV_2_PACKED1B_C \
                for(i=0; i<(dstW>>1); i++){\
                        const int i2= 2*i;\
                        int Y1= buf0[i2  ]>>7;\
@@ -502,7 +502,7 @@ static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilt
                        int V= (uvbuf0[i+2048] + uvbuf1[i+2048])>>8;\
 
 #define YSCALE_YUV_2_RGB1B_C(type) \
-                       YSCALE_YUV_2_1B_C\
+                       YSCALE_YUV_2_PACKED1B_C\
                        type *r, *b, *g;\
                        r = c->table_rV[V];\
                        g = c->table_gU[U] + c->table_gV[V];\
@@ -668,7 +668,7 @@ static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilt
        }\
 
 
-static inline void yuv2rgbXinC(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
+static inline void yuv2packedXinC(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
                                    int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
                                    uint8_t *dest, int dstW, int y)
 {
@@ -791,7 +791,7 @@ static inline void yuv2rgbXinC(SwsContext *c, int16_t *lumFilter, int16_t **lumS
                }
                break;
        case IMGFMT_YUY2:
-               YSCALE_YUV_2_X_C(void)
+               YSCALE_YUV_2_PACKEDX_C(void)
                        ((uint8_t*)dest)[2*i2+0]= Y1;
                        ((uint8_t*)dest)[2*i2+1]= U;
                        ((uint8_t*)dest)[2*i2+2]= Y2;
index 507e8826766202502ee84380b8d4df3ce8140aec..0bc0457717238f2a06f07568fcc2d8c3f783c164 100644 (file)
                           "m" (lumSrc+lumFilterSize), "m" (chrSrc+chrFilterSize)
                        : "%eax", "%ebx", "%ecx", "%edx", "%esi"
 */
-#define YSCALEYUV2RGBX \
+#define YSCALEYUV2PACKEDX \
                "xorl %%eax, %%eax              \n\t"\
                ".balign 16                     \n\t"\
                "1:                             \n\t"\
                "paddw %%mm5, %%mm7             \n\t"\
                "addl $1, %%edx                 \n\t"\
                " jnz 2b                        \n\t"\
-\
+
+
+#define YSCALEYUV2RGBX \
+               YSCALEYUV2PACKEDX\
                "psubw "MANGLE(w400)", %%mm3    \n\t" /* (U-128)8*/\
                "psubw "MANGLE(w400)", %%mm4    \n\t" /* (V-128)8*/\
                "movq %%mm3, %%mm2              \n\t" /* (U-128)8*/\
 \
                "packuswb %%mm1, %%mm1          \n\t"
 
+#define YSCALEYUV2PACKED \
+               "movd %6, %%mm6                 \n\t" /*yalpha1*/\
+               "punpcklwd %%mm6, %%mm6         \n\t"\
+               "punpcklwd %%mm6, %%mm6         \n\t"\
+                "psraw $3, %%mm6               \n\t"\
+               "movq %%mm6, 3968(%2)           \n\t"\
+               "movd %7, %%mm5                 \n\t" /*uvalpha1*/\
+               "punpcklwd %%mm5, %%mm5         \n\t"\
+               "punpcklwd %%mm5, %%mm5         \n\t"\
+                "psraw $3, %%mm5               \n\t"\
+               "movq %%mm5, 3976(%2)           \n\t"\
+               "xorl %%eax, %%eax              \n\t"\
+               ".balign 16                     \n\t"\
+               "1:                             \n\t"\
+               "movq (%2, %%eax), %%mm2        \n\t" /* uvbuf0[eax]*/\
+               "movq (%3, %%eax), %%mm3        \n\t" /* uvbuf1[eax]*/\
+               "movq 4096(%2, %%eax), %%mm5    \n\t" /* uvbuf0[eax+2048]*/\
+               "movq 4096(%3, %%eax), %%mm4    \n\t" /* uvbuf1[eax+2048]*/\
+               "psubw %%mm3, %%mm2             \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
+               "psubw %%mm4, %%mm5             \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
+               "movq 3976(%2), %%mm0           \n\t"\
+               "pmulhw %%mm0, %%mm2            \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
+               "pmulhw %%mm0, %%mm5            \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\
+               "psraw $7, %%mm3                \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
+               "psraw $7, %%mm4                \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
+               "paddw %%mm2, %%mm3             \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
+               "paddw %%mm5, %%mm4             \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\
+               "movq (%0, %%eax, 2), %%mm0     \n\t" /*buf0[eax]*/\
+               "movq (%1, %%eax, 2), %%mm1     \n\t" /*buf1[eax]*/\
+               "movq 8(%0, %%eax, 2), %%mm6    \n\t" /*buf0[eax]*/\
+               "movq 8(%1, %%eax, 2), %%mm7    \n\t" /*buf1[eax]*/\
+               "psubw %%mm1, %%mm0             \n\t" /* buf0[eax] - buf1[eax]*/\
+               "psubw %%mm7, %%mm6             \n\t" /* buf0[eax] - buf1[eax]*/\
+               "pmulhw 3968(%2), %%mm0         \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
+               "pmulhw 3968(%2), %%mm6         \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
+               "psraw $7, %%mm1                \n\t" /* buf0[eax] - buf1[eax] >>4*/\
+               "psraw $7, %%mm7                \n\t" /* buf0[eax] - buf1[eax] >>4*/\
+               "paddw %%mm0, %%mm1             \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
+               "paddw %%mm6, %%mm7             \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
+                
 #define YSCALEYUV2RGB \
                "movd %6, %%mm6                 \n\t" /*yalpha1*/\
                "punpcklwd %%mm6, %%mm6         \n\t"\
                "packuswb %%mm6, %%mm5          \n\t"\
                "packuswb %%mm3, %%mm4          \n\t"\
                "pxor %%mm7, %%mm7              \n\t"
-
+                
+#define YSCALEYUV2PACKED1 \
+               "xorl %%eax, %%eax              \n\t"\
+               ".balign 16                     \n\t"\
+               "1:                             \n\t"\
+               "movq (%2, %%eax), %%mm3        \n\t" /* uvbuf0[eax]*/\
+               "movq 4096(%2, %%eax), %%mm4    \n\t" /* uvbuf0[eax+2048]*/\
+               "psraw $7, %%mm3                \n\t" \
+               "psraw $7, %%mm4                \n\t" \
+               "movq (%0, %%eax, 2), %%mm1     \n\t" /*buf0[eax]*/\
+               "movq 8(%0, %%eax, 2), %%mm7    \n\t" /*buf0[eax]*/\
+               "psraw $7, %%mm1                \n\t" \
+               "psraw $7, %%mm7                \n\t" \
+                
 #define YSCALEYUV2RGB1 \
                "xorl %%eax, %%eax              \n\t"\
                ".balign 16                     \n\t"\
                "packuswb %%mm3, %%mm4          \n\t"\
                "pxor %%mm7, %%mm7              \n\t"
 
+#define YSCALEYUV2PACKED1b \
+               "xorl %%eax, %%eax              \n\t"\
+               ".balign 16                     \n\t"\
+               "1:                             \n\t"\
+               "movq (%2, %%eax), %%mm2        \n\t" /* uvbuf0[eax]*/\
+               "movq (%3, %%eax), %%mm3        \n\t" /* uvbuf1[eax]*/\
+               "movq 4096(%2, %%eax), %%mm5    \n\t" /* uvbuf0[eax+2048]*/\
+               "movq 4096(%3, %%eax), %%mm4    \n\t" /* uvbuf1[eax+2048]*/\
+               "paddw %%mm2, %%mm3             \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
+               "paddw %%mm5, %%mm4             \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
+               "psrlw $8, %%mm3                \n\t" \
+               "psrlw $8, %%mm4                \n\t" \
+               "movq (%0, %%eax, 2), %%mm1     \n\t" /*buf0[eax]*/\
+               "movq 8(%0, %%eax, 2), %%mm7    \n\t" /*buf0[eax]*/\
+               "psraw $7, %%mm1                \n\t" \
+               "psraw $7, %%mm7                \n\t" 
+                
 // do vertical chrominance interpolation
 #define YSCALEYUV2RGB1b \
                "xorl %%eax, %%eax              \n\t"\
 #define WRITEBGR24 WRITEBGR24MMX
 #endif
 
+#define WRITEYUY2 \
+                       "packuswb %%mm3, %%mm3          \n\t"\
+                       "packuswb %%mm4, %%mm4          \n\t"\
+                       "packuswb %%mm7, %%mm1          \n\t"\
+                       "punpcklbw %%mm4, %%mm3         \n\t"\
+                       "movq %%mm1, %%mm7              \n\t"\
+                       "punpcklbw %%mm3, %%mm1         \n\t"\
+                       "punpckhbw %%mm3, %%mm7         \n\t"\
+\
+                       MOVNTQ(%%mm1, (%4, %%eax, 2))\
+                       MOVNTQ(%%mm7, 8(%4, %%eax, 2))\
+\
+                       "addl $8, %%eax                 \n\t"\
+                       "cmpl %5, %%eax                 \n\t"\
+                       " jb 1b                         \n\t"
+
+
 static inline void RENAME(yuv2yuvX)(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
                                    int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
                                    uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW,
@@ -752,7 +842,7 @@ static inline void RENAME(yuv2yuv1)(int16_t *lumSrc, int16_t *chrSrc,
 /**
  * vertical scale YV12 to RGB
  */
-static inline void RENAME(yuv2rgbX)(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
+static inline void RENAME(yuv2packedX)(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
                                    int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
                            uint8_t *dest, int dstW, int16_t * lumMmxFilter, int16_t * chrMmxFilter, int dstY)
 {
@@ -831,9 +921,29 @@ static inline void RENAME(yuv2rgbX)(SwsContext *c, int16_t *lumFilter, int16_t *
                        );
                }
                break;
+       case IMGFMT_YUY2:
+               {
+                       asm volatile(
+                               YSCALEYUV2PACKEDX
+               /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
+
+                               "psraw $3, %%mm3                \n\t"
+                               "psraw $3, %%mm4                \n\t"
+                               "psraw $3, %%mm1                \n\t"
+                               "psraw $3, %%mm7                \n\t"
+                               WRITEYUY2
+
+                       :: "m" (-lumFilterSize), "m" (-chrFilterSize),
+                          "m" (lumMmxFilter+lumFilterSize*4), "m" (chrMmxFilter+chrFilterSize*4),
+                          "r" (dest), "m" (dstW),
+                          "m" (lumSrc+lumFilterSize), "m" (chrSrc+chrFilterSize)
+                       : "%eax", "%ebx", "%ecx", "%edx", "%esi"
+                       );
+               }
+               break;
 #endif
        default:
-               yuv2rgbXinC(c, lumFilter, lumSrc, lumFilterSize,
+               yuv2packedXinC(c, lumFilter, lumSrc, lumFilterSize,
                            chrFilter, chrSrc, chrFilterSize,
                            dest, dstW, dstY);
                break;
@@ -843,7 +953,7 @@ static inline void RENAME(yuv2rgbX)(SwsContext *c, int16_t *lumFilter, int16_t *
 /**
  * vertical bilinear scale YV12 to RGB
  */
-static inline void RENAME(yuv2rgb2)(SwsContext *c, uint16_t *buf0, uint16_t *buf1, uint16_t *uvbuf0, uint16_t *uvbuf1,
+static inline void RENAME(yuv2packed2)(SwsContext *c, uint16_t *buf0, uint16_t *buf1, uint16_t *uvbuf0, uint16_t *uvbuf1,
                            uint8_t *dest, int dstW, int yalpha, int uvalpha, int y)
 {
        int yalpha1=yalpha^4095;
@@ -1119,6 +1229,16 @@ FULL_YSCALEYUV2RGB
 
                                WRITEBGR16
 
+                       :: "r" (buf0), "r" (buf1), "r" (uvbuf0), "r" (uvbuf1), "r" (dest), "m" (dstW),
+                       "m" (yalpha1), "m" (uvalpha1)
+                       : "%eax"
+                       );
+                       return;
+       case IMGFMT_YUY2:
+                       asm volatile(
+                               YSCALEYUV2PACKED
+                               WRITEYUY2
+
                        :: "r" (buf0), "r" (buf1), "r" (uvbuf0), "r" (uvbuf1), "r" (dest), "m" (dstW),
                        "m" (yalpha1), "m" (uvalpha1)
                        : "%eax"
@@ -1127,13 +1247,13 @@ FULL_YSCALEYUV2RGB
        default: break;
        }
 #endif //HAVE_MMX
-YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB2_C, YSCALE_YUV_2_2_C)
+YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB2_C, YSCALE_YUV_2_PACKED2_C)
 }
 
 /**
  * YV12 to RGB without scaling or interpolating
  */
-static inline void RENAME(yuv2rgb1)(SwsContext *c, uint16_t *buf0, uint16_t *uvbuf0, uint16_t *uvbuf1,
+static inline void RENAME(yuv2packed1)(SwsContext *c, uint16_t *buf0, uint16_t *uvbuf0, uint16_t *uvbuf1,
                            uint8_t *dest, int dstW, int uvalpha, int dstFormat, int flags, int y)
 {
        int uvalpha1=uvalpha^4095;
@@ -1145,7 +1265,7 @@ static inline void RENAME(yuv2rgb1)(SwsContext *c, uint16_t *buf0, uint16_t *uvb
 
        if(flags&SWS_FULL_CHR_H_INT)
        {
-               RENAME(yuv2rgb2)(c, buf0, buf0, uvbuf0, uvbuf1, dest, dstW, 0, uvalpha, y);
+               RENAME(yuv2packed2)(c, buf0, buf0, uvbuf0, uvbuf1, dest, dstW, 0, uvalpha, y);
                return;
        }
 
@@ -1204,6 +1324,15 @@ static inline void RENAME(yuv2rgb1)(SwsContext *c, uint16_t *buf0, uint16_t *uvb
                        : "%eax"
                        );
                        return;
+               case IMGFMT_YUY2:
+                       asm volatile(
+                               YSCALEYUV2PACKED1
+                               WRITEYUY2
+                       :: "r" (buf0), "r" (buf0), "r" (uvbuf0), "r" (uvbuf1), "r" (dest), "m" (dstW),
+                       "m" (yalpha1), "m" (uvalpha1)
+                       : "%eax"
+                       );
+                       return;
                }
        }
        else
@@ -1260,14 +1389,23 @@ static inline void RENAME(yuv2rgb1)(SwsContext *c, uint16_t *buf0, uint16_t *uvb
                        : "%eax"
                        );
                        return;
+               case IMGFMT_YUY2:
+                       asm volatile(
+                               YSCALEYUV2PACKED1b
+                               WRITEYUY2
+                       :: "r" (buf0), "r" (buf0), "r" (uvbuf0), "r" (uvbuf1), "r" (dest), "m" (dstW),
+                       "m" (yalpha1), "m" (uvalpha1)
+                       : "%eax"
+                       );
+                       return;
                }
        }
 #endif
        if( uvalpha < 2048 )
        {
-               YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1_C, YSCALE_YUV_2_1_C)
+               YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1_C, YSCALE_YUV_2_PACKED1_C)
        }else{
-               YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1B_C, YSCALE_YUV_2_1B_C)
+               YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1B_C, YSCALE_YUV_2_PACKED1B_C)
        }
 }
 
@@ -2533,7 +2671,7 @@ i--;
                        {
                                int chrAlpha= vChrFilter[2*dstY+1];
 
-                               RENAME(yuv2rgb1)(c, *lumSrcPtr, *chrSrcPtr, *(chrSrcPtr+1),
+                               RENAME(yuv2packed1)(c, *lumSrcPtr, *chrSrcPtr, *(chrSrcPtr+1),
                                                 dest, dstW, chrAlpha, dstFormat, flags, dstY);
                        }
                        else if(vLumFilterSize == 2 && vChrFilterSize == 2) //BiLinear Upscale RGB
@@ -2541,12 +2679,12 @@ i--;
                                int lumAlpha= vLumFilter[2*dstY+1];
                                int chrAlpha= vChrFilter[2*dstY+1];
 
-                               RENAME(yuv2rgb2)(c, *lumSrcPtr, *(lumSrcPtr+1), *chrSrcPtr, *(chrSrcPtr+1),
+                               RENAME(yuv2packed2)(c, *lumSrcPtr, *(lumSrcPtr+1), *chrSrcPtr, *(chrSrcPtr+1),
                                                 dest, dstW, lumAlpha, chrAlpha, dstY);
                        }
                        else //General RGB
                        {
-                               RENAME(yuv2rgbX)(c,
+                               RENAME(yuv2packedX)(c,
                                        vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
                                        vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
                                        dest, dstW,
@@ -2571,7 +2709,7 @@ i--;
                {
                        ASSERT(lumSrcPtr + vLumFilterSize - 1 < lumPixBuf + vLumBufSize*2);
                        ASSERT(chrSrcPtr + vChrFilterSize - 1 < chrPixBuf + vChrBufSize*2);
-                       yuv2rgbXinC(c, 
+                       yuv2packedXinC(c, 
                                vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
                                vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
                                dest, dstW, dstY);