| -rw-r--r-- | core/multimedia/opieplayer/libmpeg3/video/output.c | 217 |
1 files changed, 107 insertions, 110 deletions
diff --git a/core/multimedia/opieplayer/libmpeg3/video/output.c b/core/multimedia/opieplayer/libmpeg3/video/output.c index 919a0ff..bf0d6ed 100644 --- a/core/multimedia/opieplayer/libmpeg3/video/output.c +++ b/core/multimedia/opieplayer/libmpeg3/video/output.c @@ -218,39 +218,39 @@ inline void mpeg3_bgra32_mmx(unsigned long y, unsigned long v, unsigned long *output) { -asm(" + +asm( /* Output will be 0x00rrggbb with the 00 trailing so this can also be used */ /* for bgr24. */ - movd (%0), %%mm0; /* Load y 0x00000000000000yy */ - movd (%1), %%mm1; /* Load u 0x00000000000000cr */ - movq %%mm0, %%mm3; /* Copy y to temp */ - psllq $16, %%mm1; /* Shift u 0x0000000000cr0000 */ - movd (%2), %%mm2; /* Load v 0x00000000000000cb */ - psllq $16, %%mm3; /* Shift y */ - movq %%mm1, %%mm4; /* Copy u to temp */ - por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */ - psllq $16, %%mm4; /* Shift u */ - movq %%mm2, %%mm5; /* Copy v to temp */ - psllq $16, %%mm3; /* Shift y */ - por %%mm4, %%mm1; /* Overlay new u byte 0x000000cr00cr0000 */ - psllq $16, %%mm5; /* Shift v */ - por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */ - por %%mm5, %%mm2; /* Overlay new v byte 0x0000000000cb00cb */ + "movd (%0), %%mm0;" /* Load y 0x00000000000000yy */ + "movd (%1), %%mm1;" /* Load u 0x00000000000000cr */ + "movq %%mm0, %%mm3;" /* Copy y to temp */ + "psllq $16, %%mm1;" /* Shift u 0x0000000000cr0000 */ + "movd (%2), %%mm2;" /* Load v 0x00000000000000cb */ + "psllq $16, %%mm3;" /* Shift y */ + "movq %%mm1, %%mm4;" /* Copy u to temp */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x0000000000yy00yy */ + "psllq $16, %%mm4;" /* Shift u */ + "movq %%mm2, %%mm5;" /* Copy v to temp */ + "psllq $16, %%mm3;" /* Shift y */ + "por %%mm4, %%mm1;" /* Overlay new u byte 0x000000cr00cr0000 */ + "psllq $16, %%mm5;" /* Shift v */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x000000yy00yy00yy */ + "por %%mm5, %%mm2;" /* Overlay new v byte 0x0000000000cb00cb */ /* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */ - psubw mpeg3_MMX_U_80, %%mm1; /* Subtract 128 from u 0x000000uu00uu0000 */ - pmullw mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */ - psllw $6, %%mm0; /* Shift y coeffs 0x0000yyy0yyy0yyy0 */ - psubw mpeg3_MMX_V_80, %%mm2; /* Subtract 128 from v 0x0000000000cb00cb */ - pmullw mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */ + "psubw mpeg3_MMX_U_80, %%mm1;" /* Subtract 128 from u 0x000000uu00uu0000 */ + "pmullw mpeg3_MMX_U_COEF, %%mm1;" /* Multiply u coeffs 0x0000uuuuuuuu0000 */ + "psllw $6, %%mm0;" /* Shift y coeffs 0x0000yyy0yyy0yyy0 */ + "psubw mpeg3_MMX_V_80, %%mm2;" /* Subtract 128 from v 0x0000000000cb00cb */ + "pmullw mpeg3_MMX_V_COEF, %%mm2;" /* Multiply v coeffs 0x0000crcrcrcrcrcr */ /* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */ - paddsw %%mm1, %%mm0; /* Add u to result */ - paddsw %%mm2, %%mm0; /* Add v to result 0x0000rrrrggggbbbb */ - psraw $6, %%mm0; /* Demote precision */ - packuswb %%mm0, %%mm0; /* Pack into ARGB 0x0000000000rrggbb */ - movd %%mm0, (%3); /* Store output */ - " + "paddsw %%mm1, %%mm0;" /* Add u to result */ + "paddsw %%mm2, %%mm0;" /* Add v to result 0x0000rrrrggggbbbb */ + "psraw $6, %%mm0;" /* Demote precision */ + "packuswb %%mm0, %%mm0;" /* Pack into ARGB 0x0000000000rrggbb */ + "movd %%mm0, (%3);" /* Store output */ : : "r" (&y), "r" (&u), "r" (&v), "r" (output)); } @@ -260,40 +260,39 @@ inline void mpeg3_601_bgra32_mmx(unsigned long y, unsigned long v, unsigned long *output) { -asm(" +asm( /* Output will be 0x00rrggbb with the 00 trailing so this can also be used */ /* for bgr24. */ - movd (%0), %%mm0; /* Load y 0x00000000000000yy */ - psubsw mpeg3_MMX_601_Y_DIFF, %%mm0; /* Subtract 16 from y */ - movd (%1), %%mm1; /* Load u 0x00000000000000cr */ - movq %%mm0, %%mm3; /* Copy y to temp */ - psllq $16, %%mm1; /* Shift u 0x0000000000cr0000 */ - movd (%2), %%mm2; /* Load v 0x00000000000000cb */ - psllq $16, %%mm3; /* Shift y */ - movq %%mm1, %%mm4; /* Copy u to temp */ - por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */ - psllq $16, %%mm4; /* Shift u */ - movq %%mm2, %%mm5; /* Copy v to temp */ - psllq $16, %%mm3; /* Shift y */ - por %%mm4, %%mm1; /* Overlay new u byte 0x000000cr00cr0000 */ - psllq $16, %%mm5; /* Shift v */ - por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */ - por %%mm5, %%mm2; /* Overlay new v byte 0x0000000000cb00cb */ + "movd (%0), %%mm0;" /* Load y 0x00000000000000yy */ + "psubsw mpeg3_MMX_601_Y_DIFF, %%mm0;" /* Subtract 16 from y */ + "movd (%1), %%mm1;" /* Load u 0x00000000000000cr */ + "movq %%mm0, %%mm3;" /* Copy y to temp */ + "psllq $16, %%mm1;" /* Shift u 0x0000000000cr0000 */ + "movd (%2), %%mm2;" /* Load v 0x00000000000000cb */ + "psllq $16, %%mm3;" /* Shift y */ + "movq %%mm1, %%mm4;" /* Copy u to temp */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x0000000000yy00yy */ + "psllq $16, %%mm4;" /* Shift u */ + "movq %%mm2, %%mm5;" /* Copy v to temp */ + "psllq $16, %%mm3;" /* Shift y */ + "por %%mm4, %%mm1;" /* Overlay new u byte 0x000000cr00cr0000 */ + "psllq $16, %%mm5;" /* Shift v */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x000000yy00yy00yy */ + "por %%mm5, %%mm2;" /* Overlay new v byte 0x0000000000cb00cb */ /* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */ - pmullw mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale and shift y coeffs */ - psubw mpeg3_MMX_U_80, %%mm1; /* Subtract 128 from u 0x000000uu00uu0000 */ - pmullw mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */ - psubw mpeg3_MMX_V_80, %%mm2; /* Subtract 128 from v 0x0000000000cb00cb */ - pmullw mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */ + "pmullw mpeg3_MMX_601_Y_COEF, %%mm0;" /* Scale and shift y coeffs */ + "psubw mpeg3_MMX_U_80, %%mm1;" /* Subtract 128 from u 0x000000uu00uu0000 */ + "pmullw mpeg3_MMX_U_COEF, %%mm1;" /* Multiply u coeffs 0x0000uuuuuuuu0000 */ + "psubw mpeg3_MMX_V_80, %%mm2;" /* Subtract 128 from v 0x0000000000cb00cb */ + "pmullw mpeg3_MMX_V_COEF, %%mm2;" /* Multiply v coeffs 0x0000crcrcrcrcrcr */ /* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */ - paddsw %%mm1, %%mm0; /* Add u to result */ - paddsw %%mm2, %%mm0; /* Add v to result 0x0000rrrrggggbbbb */ - psraw $6, %%mm0; /* Demote precision */ - packuswb %%mm0, %%mm0; /* Pack into ARGB 0x0000000000rrggbb */ - movd %%mm0, (%3); /* Store output */ - " + "paddsw %%mm1, %%mm0;" /* Add u to result */ + "paddsw %%mm2, %%mm0;" /* Add v to result 0x0000rrrrggggbbbb */ + "psraw $6, %%mm0;" /* Demote precision */ + "packuswb %%mm0, %%mm0;" /* Pack into ARGB 0x0000000000rrggbb */ + "movd %%mm0, (%3);" /* Store output */ : : "r" (&y), "r" (&u), "r" (&v), "r" (output)); } @@ -308,39 +307,38 @@ inline void mpeg3_rgba32_mmx(unsigned long y, unsigned long v, unsigned long *output) { -asm(" +asm( /* Output will be 0x00bbggrr with the 00 trailing so this can also be used */ /* for rgb24. */ - movd (%0), %%mm0; /* Load y 0x00000000000000yy */ - movd (%1), %%mm1; /* Load v 0x00000000000000vv */ - movq %%mm0, %%mm3; /* Copy y to temp */ - psllq $16, %%mm1; /* Shift v 0x0000000000vv0000 */ - movd (%2), %%mm2; /* Load u 0x00000000000000uu */ - psllq $16, %%mm3; /* Shift y */ - movq %%mm1, %%mm4; /* Copy v to temp */ - por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */ - psllq $16, %%mm4; /* Shift v */ - movq %%mm2, %%mm5; /* Copy u to temp */ - psllq $16, %%mm3; /* Shift y */ - por %%mm4, %%mm1; /* Overlay new v byte 0x000000vv00vv0000 */ - psllq $16, %%mm5; /* Shift u */ - por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */ - por %%mm5, %%mm2; /* Overlay new u byte 0x0000000000uu00uu */ + "movd (%0), %%mm0;" /* Load y 0x00000000000000yy */ + "movd (%1), %%mm1;" /* Load v 0x00000000000000vv */ + "movq %%mm0, %%mm3;" /* Copy y to temp */ + "psllq $16, %%mm1;" /* Shift v 0x0000000000vv0000 */ + "movd (%2), %%mm2;" /* Load u 0x00000000000000uu */ + "psllq $16, %%mm3;" /* Shift y */ + "movq %%mm1, %%mm4;" /* Copy v to temp */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x0000000000yy00yy */ + "psllq $16, %%mm4;" /* Shift v */ + "movq %%mm2, %%mm5;" /* Copy u to temp */ + "psllq $16, %%mm3;" /* Shift y */ + "por %%mm4, %%mm1;" /* Overlay new v byte 0x000000vv00vv0000 */ + "psllq $16, %%mm5;" /* Shift u */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x000000yy00yy00yy */ + "por %%mm5, %%mm2;" /* Overlay new u byte 0x0000000000uu00uu */ /* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */ - psubw mpeg3_MMX_V_80_RGB, %%mm1; /* Subtract 128 from v 0x000000vv00vv0000 */ - pmullw mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */ - psllw $6, %%mm0; /* Shift y coeffs 0x0000yyy0yyy0yyy0 */ - psubw mpeg3_MMX_U_80_RGB, %%mm2; /* Subtract 128 from u 0x0000000000uu00uu */ - pmullw mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */ + "psubw mpeg3_MMX_V_80_RGB, %%mm1;" /* Subtract 128 from v 0x000000vv00vv0000 */ + "pmullw mpeg3_MMX_V_COEF_RGB, %%mm1;" /* Multiply v coeffs 0x0000vvvvvvvv0000 */ + "psllw $6, %%mm0;" /* Shift y coeffs 0x0000yyy0yyy0yyy0 */ + "psubw mpeg3_MMX_U_80_RGB, %%mm2;" /* Subtract 128 from u 0x0000000000uu00uu */ + "pmullw mpeg3_MMX_U_COEF_RGB, %%mm2;" /* Multiply u coeffs 0x0000uuuuuuuuuuuu */ /* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */ - paddsw %%mm1, %%mm0; /* Add v to result */ - paddsw %%mm2, %%mm0; /* Add u to result 0x0000bbbbggggrrrr */ - psraw $6, %%mm0; /* Demote precision */ - packuswb %%mm0, %%mm0; /* Pack into RGBA 0x0000000000bbggrr */ - movd %%mm0, (%3); /* Store output */ - " + "paddsw %%mm1, %%mm0;" /* Add v to result */ + "paddsw %%mm2, %%mm0;" /* Add u to result 0x0000bbbbggggrrrr */ + "psraw $6, %%mm0;" /* Demote precision */ + "packuswb %%mm0, %%mm0;" /* Pack into RGBA 0x0000000000bbggrr */ + "movd %%mm0, (%3);" /* Store output */ : : "r" (&y), "r" (&v), "r" (&u), "r" (output)); } @@ -350,40 +348,39 @@ inline void mpeg3_601_rgba32_mmx(unsigned long y, unsigned long v, unsigned long *output) { -asm(" +asm( /* Output will be 0x00bbggrr with the 00 trailing so this can also be used */ /* for rgb24. */ - movd (%0), %%mm0; /* Load y 0x00000000000000yy */ - psubsw mpeg3_MMX_601_Y_DIFF, %%mm0; /* Subtract 16 from y */ - movd (%1), %%mm1; /* Load v 0x00000000000000vv */ - movq %%mm0, %%mm3; /* Copy y to temp */ - psllq $16, %%mm1; /* Shift v 0x0000000000vv0000 */ - movd (%2), %%mm2; /* Load u 0x00000000000000uu */ - psllq $16, %%mm3; /* Shift y */ - movq %%mm1, %%mm4; /* Copy v to temp */ - por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */ - psllq $16, %%mm4; /* Shift v */ - movq %%mm2, %%mm5; /* Copy u to temp */ - psllq $16, %%mm3; /* Shift y */ - por %%mm4, %%mm1; /* Overlay new v byte 0x000000vv00vv0000 */ - psllq $16, %%mm5; /* Shift u */ - por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */ - por %%mm5, %%mm2; /* Overlay new u byte 0x0000000000uu00uu */ + "movd (%0), %%mm0;" /* Load y 0x00000000000000yy */ + "psubsw mpeg3_MMX_601_Y_DIFF, %%mm0;" /* Subtract 16 from y */ + "movd (%1), %%mm1;" /* Load v 0x00000000000000vv */ + "movq %%mm0, %%mm3;" /* Copy y to temp */ + "psllq $16, %%mm1;" /* Shift v 0x0000000000vv0000 */ + "movd (%2), %%mm2;" /* Load u 0x00000000000000uu */ + "psllq $16, %%mm3;" /* Shift y */ + "movq %%mm1, %%mm4;" /* Copy v to temp */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x0000000000yy00yy */ + "psllq $16, %%mm4;" /* Shift v */ + "movq %%mm2, %%mm5;" /* Copy u to temp */ + "psllq $16, %%mm3;" /* Shift y */ + "por %%mm4, %%mm1;" /* Overlay new v byte 0x000000vv00vv0000 */ + "psllq $16, %%mm5;" /* Shift u */ + "por %%mm3, %%mm0;" /* Overlay new y byte 0x000000yy00yy00yy */ + "por %%mm5, %%mm2;" /* Overlay new u byte 0x0000000000uu00uu */ /* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */ - pmullw mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale y coeffs */ - psubw mpeg3_MMX_V_80_RGB, %%mm1; /* Subtract 128 from v 0x000000vv00vv0000 */ - pmullw mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */ - psubw mpeg3_MMX_U_80_RGB, %%mm2; /* Subtract 128 from u 0x0000000000uu00uu */ - pmullw mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */ + "pmullw mpeg3_MMX_601_Y_COEF, %%mm0;" /* Scale y coeffs */ + "psubw mpeg3_MMX_V_80_RGB, %%mm1;" /* Subtract 128 from v 0x000000vv00vv0000 */ + "pmullw mpeg3_MMX_V_COEF_RGB, %%mm1;" /* Multiply v coeffs 0x0000vvvvvvvv0000 */ + "psubw mpeg3_MMX_U_80_RGB, %%mm2;" /* Subtract 128 from u 0x0000000000uu00uu */ + "pmullw mpeg3_MMX_U_COEF_RGB, %%mm2;" /* Multiply u coeffs 0x0000uuuuuuuuuuuu */ /* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */ - paddsw %%mm1, %%mm0; /* Add v to result */ - paddsw %%mm2, %%mm0; /* Add u to result 0x0000bbbbggggrrrr */ - psraw $6, %%mm0; /* Demote precision */ - packuswb %%mm0, %%mm0; /* Pack into RGBA 0x0000000000bbggrr */ - movd %%mm0, (%3); /* Store output */ - " + "paddsw %%mm1, %%mm0;" /* Add v to result */ + "paddsw %%mm2, %%mm0;" /* Add u to result 0x0000bbbbggggrrrr */ + "psraw $6, %%mm0;" /* Demote precision */ + "packuswb %%mm0, %%mm0;" /* Pack into RGBA 0x0000000000bbggrr */ + "movd %%mm0, (%3);" /* Store output */ : : "r" (&y), "r" (&v), "r" (&u), "r" (output)); } |