Originally committed as revision 12302 to svn://svn.mplayerhq.hu/mplayer/trunk/postproctags/v0.5
| @@ -71,18 +71,18 @@ void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *sr | |||||
| #ifdef ARCH_X86 | #ifdef ARCH_X86 | ||||
| static const uint64_t mmx_null __attribute__((aligned(8))) = 0x0000000000000000ULL; | static const uint64_t mmx_null __attribute__((aligned(8))) = 0x0000000000000000ULL; | ||||
| static const uint64_t mmx_one __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL; | static const uint64_t mmx_one __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL; | ||||
| static const uint64_t mask32b __attribute__((aligned(8))) = 0x000000FF000000FFULL; | |||||
| static const uint64_t mask32g __attribute__((aligned(8))) = 0x0000FF000000FF00ULL; | |||||
| static const uint64_t mask32r __attribute__((aligned(8))) = 0x00FF000000FF0000ULL; | |||||
| static const uint64_t mask32b attribute_used __attribute__((aligned(8))) = 0x000000FF000000FFULL; | |||||
| static const uint64_t mask32g attribute_used __attribute__((aligned(8))) = 0x0000FF000000FF00ULL; | |||||
| static const uint64_t mask32r attribute_used __attribute__((aligned(8))) = 0x00FF000000FF0000ULL; | |||||
| static const uint64_t mask32 __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL; | static const uint64_t mask32 __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL; | ||||
| static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL; | static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL; | ||||
| static const uint64_t mask3216g __attribute__((aligned(8)))=0x0000FC000000FC00ULL; | static const uint64_t mask3216g __attribute__((aligned(8)))=0x0000FC000000FC00ULL; | ||||
| static const uint64_t mask3215g __attribute__((aligned(8)))=0x0000F8000000F800ULL; | static const uint64_t mask3215g __attribute__((aligned(8)))=0x0000F8000000F800ULL; | ||||
| static const uint64_t mul3216 __attribute__((aligned(8))) = 0x2000000420000004ULL; | static const uint64_t mul3216 __attribute__((aligned(8))) = 0x2000000420000004ULL; | ||||
| static const uint64_t mul3215 __attribute__((aligned(8))) = 0x2000000820000008ULL; | static const uint64_t mul3215 __attribute__((aligned(8))) = 0x2000000820000008ULL; | ||||
| static const uint64_t mask24b __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL; | |||||
| static const uint64_t mask24g __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL; | |||||
| static const uint64_t mask24r __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL; | |||||
| static const uint64_t mask24b attribute_used __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL; | |||||
| static const uint64_t mask24g attribute_used __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL; | |||||
| static const uint64_t mask24r attribute_used __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL; | |||||
| static const uint64_t mask24l __attribute__((aligned(8))) = 0x0000000000FFFFFFULL; | static const uint64_t mask24l __attribute__((aligned(8))) = 0x0000000000FFFFFFULL; | ||||
| static const uint64_t mask24h __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL; | static const uint64_t mask24h __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL; | ||||
| static const uint64_t mask24hh __attribute__((aligned(8))) = 0xffff000000000000ULL; | static const uint64_t mask24hh __attribute__((aligned(8))) = 0xffff000000000000ULL; | ||||
| @@ -104,17 +104,17 @@ static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000007 | |||||
| static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL; | static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL; | ||||
| #ifdef FAST_BGR2YV12 | #ifdef FAST_BGR2YV12 | ||||
| static const uint64_t bgr2YCoeff __attribute__((aligned(8))) = 0x000000210041000DULL; | |||||
| static const uint64_t bgr2UCoeff __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL; | |||||
| static const uint64_t bgr2VCoeff __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL; | |||||
| static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL; | |||||
| static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL; | |||||
| static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL; | |||||
| #else | #else | ||||
| static const uint64_t bgr2YCoeff __attribute__((aligned(8))) = 0x000020E540830C8BULL; | |||||
| static const uint64_t bgr2UCoeff __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL; | |||||
| static const uint64_t bgr2VCoeff __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL; | |||||
| static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL; | |||||
| static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL; | |||||
| static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL; | |||||
| #endif | #endif | ||||
| static const uint64_t bgr2YOffset __attribute__((aligned(8))) = 0x1010101010101010ULL; | |||||
| static const uint64_t bgr2UVOffset __attribute__((aligned(8)))= 0x8080808080808080ULL; | |||||
| static const uint64_t w1111 __attribute__((aligned(8))) = 0x0001000100010001ULL; | |||||
| static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL; | |||||
| static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8)))= 0x8080808080808080ULL; | |||||
| static const uint64_t w1111 attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL; | |||||
| #if 0 | #if 0 | ||||
| static volatile uint64_t __attribute__((aligned(8))) b5Dither; | static volatile uint64_t __attribute__((aligned(8))) b5Dither; | ||||
| @@ -149,19 +149,19 @@ write special BGR->BGR scaler | |||||
| #define MAX(a,b) ((a) < (b) ? (b) : (a)) | #define MAX(a,b) ((a) < (b) ? (b) : (a)) | ||||
| #ifdef ARCH_X86 | #ifdef ARCH_X86 | ||||
| static uint64_t __attribute__((aligned(8))) bF8= 0xF8F8F8F8F8F8F8F8LL; | |||||
| static uint64_t __attribute__((aligned(8))) bFC= 0xFCFCFCFCFCFCFCFCLL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) bF8= 0xF8F8F8F8F8F8F8F8LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) bFC= 0xFCFCFCFCFCFCFCFCLL; | |||||
| static uint64_t __attribute__((aligned(8))) w10= 0x0010001000100010LL; | static uint64_t __attribute__((aligned(8))) w10= 0x0010001000100010LL; | ||||
| static uint64_t __attribute__((aligned(8))) w02= 0x0002000200020002LL; | |||||
| static uint64_t __attribute__((aligned(8))) bm00001111=0x00000000FFFFFFFFLL; | |||||
| static uint64_t __attribute__((aligned(8))) bm00000111=0x0000000000FFFFFFLL; | |||||
| static uint64_t __attribute__((aligned(8))) bm11111000=0xFFFFFFFFFF000000LL; | |||||
| static uint64_t __attribute__((aligned(8))) bm01010101=0x00FF00FF00FF00FFLL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) w02= 0x0002000200020002LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) bm00001111=0x00000000FFFFFFFFLL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) bm00000111=0x0000000000FFFFFFLL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) bm11111000=0xFFFFFFFFFF000000LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) bm01010101=0x00FF00FF00FF00FFLL; | |||||
| static volatile uint64_t __attribute__((aligned(8))) b5Dither; | |||||
| static volatile uint64_t __attribute__((aligned(8))) g5Dither; | |||||
| static volatile uint64_t __attribute__((aligned(8))) g6Dither; | |||||
| static volatile uint64_t __attribute__((aligned(8))) r5Dither; | |||||
| static volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither; | |||||
| static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither; | |||||
| static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither; | |||||
| static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither; | |||||
| static uint64_t __attribute__((aligned(8))) dither4[2]={ | static uint64_t __attribute__((aligned(8))) dither4[2]={ | ||||
| 0x0103010301030103LL, | 0x0103010301030103LL, | ||||
| @@ -172,28 +172,28 @@ static uint64_t __attribute__((aligned(8))) dither8[2]={ | |||||
| 0x0004000400040004LL,}; | 0x0004000400040004LL,}; | ||||
| static uint64_t __attribute__((aligned(8))) b16Mask= 0x001F001F001F001FLL; | static uint64_t __attribute__((aligned(8))) b16Mask= 0x001F001F001F001FLL; | ||||
| static uint64_t __attribute__((aligned(8))) g16Mask= 0x07E007E007E007E0LL; | |||||
| static uint64_t __attribute__((aligned(8))) r16Mask= 0xF800F800F800F800LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) g16Mask= 0x07E007E007E007E0LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) r16Mask= 0xF800F800F800F800LL; | |||||
| static uint64_t __attribute__((aligned(8))) b15Mask= 0x001F001F001F001FLL; | static uint64_t __attribute__((aligned(8))) b15Mask= 0x001F001F001F001FLL; | ||||
| static uint64_t __attribute__((aligned(8))) g15Mask= 0x03E003E003E003E0LL; | |||||
| static uint64_t __attribute__((aligned(8))) r15Mask= 0x7C007C007C007C00LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) g15Mask= 0x03E003E003E003E0LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) r15Mask= 0x7C007C007C007C00LL; | |||||
| static uint64_t __attribute__((aligned(8))) M24A= 0x00FF0000FF0000FFLL; | |||||
| static uint64_t __attribute__((aligned(8))) M24B= 0xFF0000FF0000FF00LL; | |||||
| static uint64_t __attribute__((aligned(8))) M24C= 0x0000FF0000FF0000LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) M24A= 0x00FF0000FF0000FFLL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) M24B= 0xFF0000FF0000FF00LL; | |||||
| static uint64_t attribute_used __attribute__((aligned(8))) M24C= 0x0000FF0000FF0000LL; | |||||
| #ifdef FAST_BGR2YV12 | #ifdef FAST_BGR2YV12 | ||||
| static const uint64_t bgr2YCoeff __attribute__((aligned(8))) = 0x000000210041000DULL; | |||||
| static const uint64_t bgr2UCoeff __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL; | |||||
| static const uint64_t bgr2VCoeff __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL; | |||||
| static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL; | |||||
| static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL; | |||||
| static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL; | |||||
| #else | #else | ||||
| static const uint64_t bgr2YCoeff __attribute__((aligned(8))) = 0x000020E540830C8BULL; | |||||
| static const uint64_t bgr2UCoeff __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL; | |||||
| static const uint64_t bgr2VCoeff __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL; | |||||
| static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL; | |||||
| static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL; | |||||
| static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL; | |||||
| #endif | #endif | ||||
| static const uint64_t bgr2YOffset __attribute__((aligned(8))) = 0x1010101010101010ULL; | |||||
| static const uint64_t bgr2UVOffset __attribute__((aligned(8)))= 0x8080808080808080ULL; | |||||
| static const uint64_t w1111 __attribute__((aligned(8))) = 0x0001000100010001ULL; | |||||
| static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL; | |||||
| static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8)))= 0x8080808080808080ULL; | |||||
| static const uint64_t w1111 attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL; | |||||
| #endif | #endif | ||||
| // clipping helper table for C implementations: | // clipping helper table for C implementations: | ||||
| @@ -159,20 +159,20 @@ const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={ | |||||
| #ifdef ARCH_X86 | #ifdef ARCH_X86 | ||||
| /* hope these constant values are cache line aligned */ | /* hope these constant values are cache line aligned */ | ||||
| uint64_t __attribute__((aligned(8))) mmx_00ffw = 0x00ff00ff00ff00ffULL; | |||||
| uint64_t __attribute__((aligned(8))) mmx_redmask = 0xf8f8f8f8f8f8f8f8ULL; | |||||
| uint64_t __attribute__((aligned(8))) mmx_grnmask = 0xfcfcfcfcfcfcfcfcULL; | |||||
| uint64_t attribute_used __attribute__((aligned(8))) mmx_00ffw = 0x00ff00ff00ff00ffULL; | |||||
| uint64_t attribute_used __attribute__((aligned(8))) mmx_redmask = 0xf8f8f8f8f8f8f8f8ULL; | |||||
| uint64_t attribute_used __attribute__((aligned(8))) mmx_grnmask = 0xfcfcfcfcfcfcfcfcULL; | |||||
| uint64_t __attribute__((aligned(8))) M24A= 0x00FF0000FF0000FFULL; | |||||
| uint64_t __attribute__((aligned(8))) M24B= 0xFF0000FF0000FF00ULL; | |||||
| uint64_t __attribute__((aligned(8))) M24C= 0x0000FF0000FF0000ULL; | |||||
| uint64_t attribute_used __attribute__((aligned(8))) M24A= 0x00FF0000FF0000FFULL; | |||||
| uint64_t attribute_used __attribute__((aligned(8))) M24B= 0xFF0000FF0000FF00ULL; | |||||
| uint64_t attribute_used __attribute__((aligned(8))) M24C= 0x0000FF0000FF0000ULL; | |||||
| // the volatile is required because gcc otherwise optimizes some writes away not knowing that these | // the volatile is required because gcc otherwise optimizes some writes away not knowing that these | ||||
| // are read in the asm block | // are read in the asm block | ||||
| volatile uint64_t __attribute__((aligned(8))) b5Dither; | |||||
| volatile uint64_t __attribute__((aligned(8))) g5Dither; | |||||
| volatile uint64_t __attribute__((aligned(8))) g6Dither; | |||||
| volatile uint64_t __attribute__((aligned(8))) r5Dither; | |||||
| volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither; | |||||
| volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither; | |||||
| volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither; | |||||
| volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither; | |||||
| uint64_t __attribute__((aligned(8))) dither4[2]={ | uint64_t __attribute__((aligned(8))) dither4[2]={ | ||||
| 0x0103010301030103LL, | 0x0103010301030103LL, | ||||