Attachment #19944 for bug #32122

View | Details | Raw Unified | Return to bug 32122
Collapse All | Expand All




		unsigned long v, 
		unsigned long *output)
{
asm("\n\
/* Output will be 0x00rrggbb with the 00 trailing so this can also be used */\n\
/* for bgr24. */\n\
	movd (%0), %%mm0;          /* Load y   0x00000000000000yy */\n\
	movd (%1), %%mm1;          /* Load u    0x00000000000000cr */\n\
	movq %%mm0, %%mm3;         /* Copy y to temp */\n\
	psllq $16, %%mm1;          /* Shift u   0x0000000000cr0000 */\n\
	movd (%2), %%mm2;          /* Load v    0x00000000000000cb */\n\
	psllq $16, %%mm3;          /* Shift y */\n\
	movq %%mm1, %%mm4;         /* Copy u to temp */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */\n\
	psllq $16, %%mm4;          /* Shift u */\n\
	movq %%mm2, %%mm5;         /* Copy v to temp */\n\
	psllq $16, %%mm3;          /* Shift y  */\n\
	por %%mm4, %%mm1;          /* Overlay new u byte 0x000000cr00cr0000 */\n\
	psllq $16, %%mm5;          /* Shift v  */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */\n\
	por %%mm5, %%mm2;          /* Overlay new v byte 0x0000000000cb00cb */\n\
\n\
/* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */\n\
 	psubw _mpeg3_MMX_U_80, %%mm1;    /* Subtract 128 from u 0x000000uu00uu0000 */\n\
 	pmullw _mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */\n\
 	psllw $6, %%mm0;                /* Shift y coeffs 0x0000yyy0yyy0yyy0 */\n\
 	psubw _mpeg3_MMX_V_80, %%mm2;    /* Subtract 128 from v 0x0000000000cb00cb */\n\
 	pmullw _mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */\n\
\n\
/* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */\n\
	paddsw %%mm1, %%mm0;        /* Add u to result */\n\
	paddsw %%mm2, %%mm0;        /* Add v to result 0x0000rrrrggggbbbb */\n\
	psraw $6, %%mm0;           /* Demote precision */\n\
	packuswb %%mm0, %%mm0;     /* Pack into ARGB 0x0000000000rrggbb */\n\
	movd %%mm0, (%3);          /* Store output */\n\
	"
:
: "r" (&y), "r" (&u), "r" (&v), "r" (output));

		unsigned long v, 
		unsigned long *output)
{
asm("\n\
/* Output will be 0x00rrggbb with the 00 trailing so this can also be used */\n\
/* for bgr24. */\n\
	movd (%0), %%mm0;          /* Load y   0x00000000000000yy */\n\
	psubsw _mpeg3_MMX_601_Y_DIFF, %%mm0;      /* Subtract 16 from y */\n\
	movd (%1), %%mm1;          /* Load u    0x00000000000000cr */\n\
	movq %%mm0, %%mm3;         /* Copy y to temp */\n\
	psllq $16, %%mm1;          /* Shift u   0x0000000000cr0000 */\n\
	movd (%2), %%mm2;          /* Load v    0x00000000000000cb */\n\
	psllq $16, %%mm3;          /* Shift y */\n\
	movq %%mm1, %%mm4;         /* Copy u to temp */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */\n\
	psllq $16, %%mm4;          /* Shift u */\n\
	movq %%mm2, %%mm5;         /* Copy v to temp */\n\
	psllq $16, %%mm3;          /* Shift y  */\n\
	por %%mm4, %%mm1;          /* Overlay new u byte 0x000000cr00cr0000 */\n\
	psllq $16, %%mm5;          /* Shift v  */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */\n\
	por %%mm5, %%mm2;          /* Overlay new v byte 0x0000000000cb00cb */\n\
\n\
/* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */\n\
	pmullw _mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale and shift y coeffs */\n\
	psubw _mpeg3_MMX_U_80, %%mm1;     /* Subtract 128 from u 0x000000uu00uu0000 */\n\
 	pmullw _mpeg3_MMX_U_COEF, %%mm1;  /* Multiply u coeffs 0x0000uuuuuuuu0000 */\n\
	psubw _mpeg3_MMX_V_80, %%mm2;     /* Subtract 128 from v 0x0000000000cb00cb */\n\
 	pmullw _mpeg3_MMX_V_COEF, %%mm2;  /* Multiply v coeffs 0x0000crcrcrcrcrcr */\n\
\n\
/* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */\n\
	paddsw %%mm1, %%mm0;        /* Add u to result */\n\
	paddsw %%mm2, %%mm0;        /* Add v to result 0x0000rrrrggggbbbb */\n\
	psraw $6, %%mm0;           /* Demote precision */\n\
	packuswb %%mm0, %%mm0;     /* Pack into ARGB 0x0000000000rrggbb */\n\
	movd %%mm0, (%3);          /* Store output */\n\
	"
:
: "r" (&y), "r" (&u), "r" (&v), "r" (output));

		unsigned long v, 
		unsigned long *output)
{
asm("\n\
/* Output will be 0x00bbggrr with the 00 trailing so this can also be used */\n\
/* for rgb24. */\n\
	movd (%0), %%mm0;          /* Load y   0x00000000000000yy */\n\
	movd (%1), %%mm1;          /* Load v    0x00000000000000vv */\n\
	movq %%mm0, %%mm3;         /* Copy y to temp */\n\
	psllq $16, %%mm1;          /* Shift v   0x0000000000vv0000 */\n\
	movd (%2), %%mm2;          /* Load u    0x00000000000000uu */\n\
	psllq $16, %%mm3;          /* Shift y */\n\
	movq %%mm1, %%mm4;         /* Copy v to temp */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */\n\
	psllq $16, %%mm4;          /* Shift v */\n\
	movq %%mm2, %%mm5;         /* Copy u to temp */\n\
	psllq $16, %%mm3;          /* Shift y  */\n\
	por %%mm4, %%mm1;          /* Overlay new v byte 0x000000vv00vv0000 */\n\
	psllq $16, %%mm5;          /* Shift u  */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */\n\
	por %%mm5, %%mm2;          /* Overlay new u byte 0x0000000000uu00uu */\n\
\n\
/* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */\n\
 	psubw _mpeg3_MMX_V_80_RGB, %%mm1;    /* Subtract 128 from v 0x000000vv00vv0000 */\n\
 	pmullw _mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */\n\
 	psllw $6, %%mm0;                /* Shift y coeffs 0x0000yyy0yyy0yyy0 */\n\
 	psubw _mpeg3_MMX_U_80_RGB, %%mm2;    /* Subtract 128 from u 0x0000000000uu00uu */\n\
 	pmullw _mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */\n\
\n\
/* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */\n\
	paddsw %%mm1, %%mm0;        /* Add v to result */\n\
	paddsw %%mm2, %%mm0;        /* Add u to result 0x0000bbbbggggrrrr */\n\
	psraw $6, %%mm0;           /* Demote precision */\n\
	packuswb %%mm0, %%mm0;     /* Pack into RGBA 0x0000000000bbggrr */\n\
	movd %%mm0, (%3);          /* Store output */\n\
	"
:
: "r" (&y), "r" (&v), "r" (&u), "r" (output));

		unsigned long v, 
		unsigned long *output)
{
asm("\n\
/* Output will be 0x00bbggrr with the 00 trailing so this can also be used */\n\
/* for rgb24. */\n\
	movd (%0), %%mm0;          /* Load y   0x00000000000000yy */\n\
	psubsw _mpeg3_MMX_601_Y_DIFF, %%mm0;      /* Subtract 16 from y */\n\
	movd (%1), %%mm1;          /* Load v    0x00000000000000vv */\n\
	movq %%mm0, %%mm3;         /* Copy y to temp */\n\
	psllq $16, %%mm1;          /* Shift v   0x0000000000vv0000 */\n\
	movd (%2), %%mm2;          /* Load u    0x00000000000000uu */\n\
	psllq $16, %%mm3;          /* Shift y */\n\
	movq %%mm1, %%mm4;         /* Copy v to temp */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */\n\
	psllq $16, %%mm4;          /* Shift v */\n\
	movq %%mm2, %%mm5;         /* Copy u to temp */\n\
	psllq $16, %%mm3;          /* Shift y  */\n\
	por %%mm4, %%mm1;          /* Overlay new v byte 0x000000vv00vv0000 */\n\
	psllq $16, %%mm5;          /* Shift u  */\n\
	por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */\n\
	por %%mm5, %%mm2;          /* Overlay new u byte 0x0000000000uu00uu */\n\
\n\
/* mm0: 0x000000yy00yy00yy     mm1: 0x000000vv00vv0000     mm2: 0x0000000000uu00uu */\n\
	pmullw _mpeg3_MMX_601_Y_COEF, %%mm0;     /* Scale y coeffs */\n\
 	psubw _mpeg3_MMX_V_80_RGB, %%mm1;    /* Subtract 128 from v 0x000000vv00vv0000 */\n\
 	pmullw _mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */\n\
 	psubw _mpeg3_MMX_U_80_RGB, %%mm2;    /* Subtract 128 from u 0x0000000000uu00uu */\n\
 	pmullw _mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */\n\
\n\
/* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */\n\
	paddsw %%mm1, %%mm0;        /* Add v to result */\n\
	paddsw %%mm2, %%mm0;        /* Add u to result 0x0000bbbbggggrrrr */\n\
	psraw $6, %%mm0;           /* Demote precision */\n\
	packuswb %%mm0, %%mm0;     /* Pack into RGBA 0x0000000000bbggrr */\n\
	movd %%mm0, (%3);          /* Store output */\n\
	"
:
: "r" (&y), "r" (&v), "r" (&u), "r" (output));

Return to bug 32122

Lines 261-298 inline void mpeg3_bgra32_mmx(unsigned lo Link Here

(-)platforms/Cross/plugins/Mpeg3Plugin/libmpeg/video/output.c.org (-130 / +130 lines)
261	unsigned long v,	261	unsigned long v,
262	unsigned long *output)	262	unsigned long *output)
263	{	263	{
264	asm("	264	asm("\n\
265	/* Output will be 0x00rrggbb with the 00 trailing so this can also be used */	265	/* Output will be 0x00rrggbb with the 00 trailing so this can also be used */\n\
266	/* for bgr24. */	266	/* for bgr24. */\n\
267	movd (%0), %%mm0; /* Load y 0x00000000000000yy */	267	movd (%0), %%mm0; /* Load y 0x00000000000000yy */\n\
268	movd (%1), %%mm1; /* Load u 0x00000000000000cr */	268	movd (%1), %%mm1; /* Load u 0x00000000000000cr */\n\
269	movq %%mm0, %%mm3; /* Copy y to temp */	269	movq %%mm0, %%mm3; /* Copy y to temp */\n\
270	psllq $16, %%mm1; /* Shift u 0x0000000000cr0000 */	270	psllq $16, %%mm1; /* Shift u 0x0000000000cr0000 */\n\
271	movd (%2), %%mm2; /* Load v 0x00000000000000cb */	271	movd (%2), %%mm2; /* Load v 0x00000000000000cb */\n\
272	psllq $16, %%mm3; /* Shift y */	272	psllq $16, %%mm3; /* Shift y */\n\
273	movq %%mm1, %%mm4; /* Copy u to temp */	273	movq %%mm1, %%mm4; /* Copy u to temp */\n\
274	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */	274	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */\n\
275	psllq $16, %%mm4; /* Shift u */	275	psllq $16, %%mm4; /* Shift u */\n\
276	movq %%mm2, %%mm5; /* Copy v to temp */	276	movq %%mm2, %%mm5; /* Copy v to temp */\n\
277	psllq $16, %%mm3; /* Shift y */	277	psllq $16, %%mm3; /* Shift y */\n\
278	por %%mm4, %%mm1; /* Overlay new u byte 0x000000cr00cr0000 */	278	por %%mm4, %%mm1; /* Overlay new u byte 0x000000cr00cr0000 */\n\
279	psllq $16, %%mm5; /* Shift v */	279	psllq $16, %%mm5; /* Shift v */\n\
280	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */	280	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */\n\
281	por %%mm5, %%mm2; /* Overlay new v byte 0x0000000000cb00cb */	281	por %%mm5, %%mm2; /* Overlay new v byte 0x0000000000cb00cb */\n\
282		282	\n\
283	/* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */	283	/* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */\n\
284	psubw _mpeg3_MMX_U_80, %%mm1; /* Subtract 128 from u 0x000000uu00uu0000 */	284	psubw _mpeg3_MMX_U_80, %%mm1; /* Subtract 128 from u 0x000000uu00uu0000 */\n\
285	pmullw _mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */	285	pmullw _mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */\n\
286	psllw $6, %%mm0; /* Shift y coeffs 0x0000yyy0yyy0yyy0 */	286	psllw $6, %%mm0; /* Shift y coeffs 0x0000yyy0yyy0yyy0 */\n\
287	psubw _mpeg3_MMX_V_80, %%mm2; /* Subtract 128 from v 0x0000000000cb00cb */	287	psubw _mpeg3_MMX_V_80, %%mm2; /* Subtract 128 from v 0x0000000000cb00cb */\n\
288	pmullw _mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */	288	pmullw _mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */\n\
289		289	\n\
290	/* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */	290	/* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */\n\
291	paddsw %%mm1, %%mm0; /* Add u to result */	291	paddsw %%mm1, %%mm0; /* Add u to result */\n\
292	paddsw %%mm2, %%mm0; /* Add v to result 0x0000rrrrggggbbbb */	292	paddsw %%mm2, %%mm0; /* Add v to result 0x0000rrrrggggbbbb */\n\
293	psraw $6, %%mm0; /* Demote precision */	293	psraw $6, %%mm0; /* Demote precision */\n\
294	packuswb %%mm0, %%mm0; /* Pack into ARGB 0x0000000000rrggbb */	294	packuswb %%mm0, %%mm0; /* Pack into ARGB 0x0000000000rrggbb */\n\
295	movd %%mm0, (%3); /* Store output */	295	movd %%mm0, (%3); /* Store output */\n\
296	"	296	"
297	:	297	:
298	: "r" (&y), "r" (&u), "r" (&v), "r" (output));	298	: "r" (&y), "r" (&u), "r" (&v), "r" (output));
Lines 303-341 inline void mpeg3_601_bgra32_mmx(unsigne Link Here
303	unsigned long v,	303	unsigned long v,
304	unsigned long *output)	304	unsigned long *output)
305	{	305	{
306	asm("	306	asm("\n\
307	/* Output will be 0x00rrggbb with the 00 trailing so this can also be used */	307	/* Output will be 0x00rrggbb with the 00 trailing so this can also be used */\n\
308	/* for bgr24. */	308	/* for bgr24. */\n\
309	movd (%0), %%mm0; /* Load y 0x00000000000000yy */	309	movd (%0), %%mm0; /* Load y 0x00000000000000yy */\n\
310	psubsw _mpeg3_MMX_601_Y_DIFF, %%mm0; /* Subtract 16 from y */	310	psubsw _mpeg3_MMX_601_Y_DIFF, %%mm0; /* Subtract 16 from y */\n\
311	movd (%1), %%mm1; /* Load u 0x00000000000000cr */	311	movd (%1), %%mm1; /* Load u 0x00000000000000cr */\n\
312	movq %%mm0, %%mm3; /* Copy y to temp */	312	movq %%mm0, %%mm3; /* Copy y to temp */\n\
313	psllq $16, %%mm1; /* Shift u 0x0000000000cr0000 */	313	psllq $16, %%mm1; /* Shift u 0x0000000000cr0000 */\n\
314	movd (%2), %%mm2; /* Load v 0x00000000000000cb */	314	movd (%2), %%mm2; /* Load v 0x00000000000000cb */\n\
315	psllq $16, %%mm3; /* Shift y */	315	psllq $16, %%mm3; /* Shift y */\n\
316	movq %%mm1, %%mm4; /* Copy u to temp */	316	movq %%mm1, %%mm4; /* Copy u to temp */\n\
317	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */	317	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */\n\
318	psllq $16, %%mm4; /* Shift u */	318	psllq $16, %%mm4; /* Shift u */\n\
319	movq %%mm2, %%mm5; /* Copy v to temp */	319	movq %%mm2, %%mm5; /* Copy v to temp */\n\
320	psllq $16, %%mm3; /* Shift y */	320	psllq $16, %%mm3; /* Shift y */\n\
321	por %%mm4, %%mm1; /* Overlay new u byte 0x000000cr00cr0000 */	321	por %%mm4, %%mm1; /* Overlay new u byte 0x000000cr00cr0000 */\n\
322	psllq $16, %%mm5; /* Shift v */	322	psllq $16, %%mm5; /* Shift v */\n\
323	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */	323	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */\n\
324	por %%mm5, %%mm2; /* Overlay new v byte 0x0000000000cb00cb */	324	por %%mm5, %%mm2; /* Overlay new v byte 0x0000000000cb00cb */\n\
325		325	\n\
326	/* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */	326	/* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */\n\
327	pmullw _mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale and shift y coeffs */	327	pmullw _mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale and shift y coeffs */\n\
328	psubw _mpeg3_MMX_U_80, %%mm1; /* Subtract 128 from u 0x000000uu00uu0000 */	328	psubw _mpeg3_MMX_U_80, %%mm1; /* Subtract 128 from u 0x000000uu00uu0000 */\n\
329	pmullw _mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */	329	pmullw _mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */\n\
330	psubw _mpeg3_MMX_V_80, %%mm2; /* Subtract 128 from v 0x0000000000cb00cb */	330	psubw _mpeg3_MMX_V_80, %%mm2; /* Subtract 128 from v 0x0000000000cb00cb */\n\
331	pmullw _mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */	331	pmullw _mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */\n\
332		332	\n\
333	/* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */	333	/* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */\n\
334	paddsw %%mm1, %%mm0; /* Add u to result */	334	paddsw %%mm1, %%mm0; /* Add u to result */\n\
335	paddsw %%mm2, %%mm0; /* Add v to result 0x0000rrrrggggbbbb */	335	paddsw %%mm2, %%mm0; /* Add v to result 0x0000rrrrggggbbbb */\n\
336	psraw $6, %%mm0; /* Demote precision */	336	psraw $6, %%mm0; /* Demote precision */\n\
337	packuswb %%mm0, %%mm0; /* Pack into ARGB 0x0000000000rrggbb */	337	packuswb %%mm0, %%mm0; /* Pack into ARGB 0x0000000000rrggbb */\n\
338	movd %%mm0, (%3); /* Store output */	338	movd %%mm0, (%3); /* Store output */\n\
339	"	339	"
340	:	340	:
341	: "r" (&y), "r" (&u), "r" (&v), "r" (output));	341	: "r" (&y), "r" (&u), "r" (&v), "r" (output));
Lines 351-388 inline void mpeg3_rgba32_mmx(unsigned lo Link Here
351	unsigned long v,	351	unsigned long v,
352	unsigned long *output)	352	unsigned long *output)
353	{	353	{
354	asm("	354	asm("\n\
355	/* Output will be 0x00bbggrr with the 00 trailing so this can also be used */	355	/* Output will be 0x00bbggrr with the 00 trailing so this can also be used */\n\
356	/* for rgb24. */	356	/* for rgb24. */\n\
357	movd (%0), %%mm0; /* Load y 0x00000000000000yy */	357	movd (%0), %%mm0; /* Load y 0x00000000000000yy */\n\
358	movd (%1), %%mm1; /* Load v 0x00000000000000vv */	358	movd (%1), %%mm1; /* Load v 0x00000000000000vv */\n\
359	movq %%mm0, %%mm3; /* Copy y to temp */	359	movq %%mm0, %%mm3; /* Copy y to temp */\n\
360	psllq $16, %%mm1; /* Shift v 0x0000000000vv0000 */	360	psllq $16, %%mm1; /* Shift v 0x0000000000vv0000 */\n\
361	movd (%2), %%mm2; /* Load u 0x00000000000000uu */	361	movd (%2), %%mm2; /* Load u 0x00000000000000uu */\n\
362	psllq $16, %%mm3; /* Shift y */	362	psllq $16, %%mm3; /* Shift y */\n\
363	movq %%mm1, %%mm4; /* Copy v to temp */	363	movq %%mm1, %%mm4; /* Copy v to temp */\n\
364	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */	364	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */\n\
365	psllq $16, %%mm4; /* Shift v */	365	psllq $16, %%mm4; /* Shift v */\n\
366	movq %%mm2, %%mm5; /* Copy u to temp */	366	movq %%mm2, %%mm5; /* Copy u to temp */\n\
367	psllq $16, %%mm3; /* Shift y */	367	psllq $16, %%mm3; /* Shift y */\n\
368	por %%mm4, %%mm1; /* Overlay new v byte 0x000000vv00vv0000 */	368	por %%mm4, %%mm1; /* Overlay new v byte 0x000000vv00vv0000 */\n\
369	psllq $16, %%mm5; /* Shift u */	369	psllq $16, %%mm5; /* Shift u */\n\
370	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */	370	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */\n\
371	por %%mm5, %%mm2; /* Overlay new u byte 0x0000000000uu00uu */	371	por %%mm5, %%mm2; /* Overlay new u byte 0x0000000000uu00uu */\n\
372		372	\n\
373	/* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */	373	/* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */\n\
374	psubw _mpeg3_MMX_V_80_RGB, %%mm1; /* Subtract 128 from v 0x000000vv00vv0000 */	374	psubw _mpeg3_MMX_V_80_RGB, %%mm1; /* Subtract 128 from v 0x000000vv00vv0000 */\n\
375	pmullw _mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */	375	pmullw _mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */\n\
376	psllw $6, %%mm0; /* Shift y coeffs 0x0000yyy0yyy0yyy0 */	376	psllw $6, %%mm0; /* Shift y coeffs 0x0000yyy0yyy0yyy0 */\n\
377	psubw _mpeg3_MMX_U_80_RGB, %%mm2; /* Subtract 128 from u 0x0000000000uu00uu */	377	psubw _mpeg3_MMX_U_80_RGB, %%mm2; /* Subtract 128 from u 0x0000000000uu00uu */\n\
378	pmullw _mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */	378	pmullw _mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */\n\
379		379	\n\
380	/* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */	380	/* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */\n\
381	paddsw %%mm1, %%mm0; /* Add v to result */	381	paddsw %%mm1, %%mm0; /* Add v to result */\n\
382	paddsw %%mm2, %%mm0; /* Add u to result 0x0000bbbbggggrrrr */	382	paddsw %%mm2, %%mm0; /* Add u to result 0x0000bbbbggggrrrr */\n\
383	psraw $6, %%mm0; /* Demote precision */	383	psraw $6, %%mm0; /* Demote precision */\n\
384	packuswb %%mm0, %%mm0; /* Pack into RGBA 0x0000000000bbggrr */	384	packuswb %%mm0, %%mm0; /* Pack into RGBA 0x0000000000bbggrr */\n\
385	movd %%mm0, (%3); /* Store output */	385	movd %%mm0, (%3); /* Store output */\n\
386	"	386	"
387	:	387	:
388	: "r" (&y), "r" (&v), "r" (&u), "r" (output));	388	: "r" (&y), "r" (&v), "r" (&u), "r" (output));
Lines 393-431 inline void mpeg3_601_rgba32_mmx(unsigne Link Here
393	unsigned long v,	393	unsigned long v,
394	unsigned long *output)	394	unsigned long *output)
395	{	395	{
396	asm("	396	asm("\n\
397	/* Output will be 0x00bbggrr with the 00 trailing so this can also be used */	397	/* Output will be 0x00bbggrr with the 00 trailing so this can also be used */\n\
398	/* for rgb24. */	398	/* for rgb24. */\n\
399	movd (%0), %%mm0; /* Load y 0x00000000000000yy */	399	movd (%0), %%mm0; /* Load y 0x00000000000000yy */\n\
400	psubsw _mpeg3_MMX_601_Y_DIFF, %%mm0; /* Subtract 16 from y */	400	psubsw _mpeg3_MMX_601_Y_DIFF, %%mm0; /* Subtract 16 from y */\n\
401	movd (%1), %%mm1; /* Load v 0x00000000000000vv */	401	movd (%1), %%mm1; /* Load v 0x00000000000000vv */\n\
402	movq %%mm0, %%mm3; /* Copy y to temp */	402	movq %%mm0, %%mm3; /* Copy y to temp */\n\
403	psllq $16, %%mm1; /* Shift v 0x0000000000vv0000 */	403	psllq $16, %%mm1; /* Shift v 0x0000000000vv0000 */\n\
404	movd (%2), %%mm2; /* Load u 0x00000000000000uu */	404	movd (%2), %%mm2; /* Load u 0x00000000000000uu */\n\
405	psllq $16, %%mm3; /* Shift y */	405	psllq $16, %%mm3; /* Shift y */\n\
406	movq %%mm1, %%mm4; /* Copy v to temp */	406	movq %%mm1, %%mm4; /* Copy v to temp */\n\
407	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */	407	por %%mm3, %%mm0; /* Overlay new y byte 0x0000000000yy00yy */\n\
408	psllq $16, %%mm4; /* Shift v */	408	psllq $16, %%mm4; /* Shift v */\n\
409	movq %%mm2, %%mm5; /* Copy u to temp */	409	movq %%mm2, %%mm5; /* Copy u to temp */\n\
410	psllq $16, %%mm3; /* Shift y */	410	psllq $16, %%mm3; /* Shift y */\n\
411	por %%mm4, %%mm1; /* Overlay new v byte 0x000000vv00vv0000 */	411	por %%mm4, %%mm1; /* Overlay new v byte 0x000000vv00vv0000 */\n\
412	psllq $16, %%mm5; /* Shift u */	412	psllq $16, %%mm5; /* Shift u */\n\
413	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */	413	por %%mm3, %%mm0; /* Overlay new y byte 0x000000yy00yy00yy */\n\
414	por %%mm5, %%mm2; /* Overlay new u byte 0x0000000000uu00uu */	414	por %%mm5, %%mm2; /* Overlay new u byte 0x0000000000uu00uu */\n\
415		415	\n\
416	/* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */	416	/* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */\n\
417	pmullw _mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale y coeffs */	417	pmullw _mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale y coeffs */\n\
418	psubw _mpeg3_MMX_V_80_RGB, %%mm1; /* Subtract 128 from v 0x000000vv00vv0000 */	418	psubw _mpeg3_MMX_V_80_RGB, %%mm1; /* Subtract 128 from v 0x000000vv00vv0000 */\n\
419	pmullw _mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */	419	pmullw _mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */\n\
420	psubw _mpeg3_MMX_U_80_RGB, %%mm2; /* Subtract 128 from u 0x0000000000uu00uu */	420	psubw _mpeg3_MMX_U_80_RGB, %%mm2; /* Subtract 128 from u 0x0000000000uu00uu */\n\
421	pmullw _mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */	421	pmullw _mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */\n\
422		422	\n\
423	/* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */	423	/* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */\n\
424	paddsw %%mm1, %%mm0; /* Add v to result */	424	paddsw %%mm1, %%mm0; /* Add v to result */\n\
425	paddsw %%mm2, %%mm0; /* Add u to result 0x0000bbbbggggrrrr */	425	paddsw %%mm2, %%mm0; /* Add u to result 0x0000bbbbggggrrrr */\n\
426	psraw $6, %%mm0; /* Demote precision */	426	psraw $6, %%mm0; /* Demote precision */\n\
427	packuswb %%mm0, %%mm0; /* Pack into RGBA 0x0000000000bbggrr */	427	packuswb %%mm0, %%mm0; /* Pack into RGBA 0x0000000000bbggrr */\n\
428	movd %%mm0, (%3); /* Store output */	428	movd %%mm0, (%3); /* Store output */\n\
429	"	429	"
430	:	430	:
431	: "r" (&y), "r" (&v), "r" (&u), "r" (output));	431	: "r" (&y), "r" (&v), "r" (&u), "r" (output));