Skip to content

L
libswscale
Commit 82551fd3 authored by nplourde's avatar nplourde
Browse files
Options

add support for intel mac. mp3lib is not fixed yet. 


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18105 b3059339-0415-0410-9bf9-f77b7e298cf2
parent 2cbe899d
Show whitespace changes
Inline Side-by-side
Showing with 44 additions and 31 deletions
rgb2rgb_template.c
View file @ 82551fd3
...@@ -12,6 +12,8 @@ ...@@ -12,6 +12,8 @@
#include <stddef.h> #include <stddef.h>
#include <inttypes.h> /* for __WORDSIZE */ #include <inttypes.h> /* for __WORDSIZE */
#include "asmalign.h"
#ifndef __WORDSIZE #ifndef __WORDSIZE
// #warning You have misconfigured system and probably will lose performance! // #warning You have misconfigured system and probably will lose performance!
#define __WORDSIZE MP_WORDSIZE #define __WORDSIZE MP_WORDSIZE
...@@ -40,9 +42,14 @@ ...@@ -40,9 +42,14 @@
#define PREFETCHW "prefetcht0" #define PREFETCHW "prefetcht0"
#define PAVGB "pavgb" #define PAVGB "pavgb"
#else #else
#ifdef __APPLE__
#define PREFETCH "#"
#define PREFETCHW "#"
#elif
#define PREFETCH "/nop" #define PREFETCH "/nop"
#define PREFETCHW "/nop" #define PREFETCHW "/nop"
#endif #endif
#endif
#ifdef HAVE_3DNOW #ifdef HAVE_3DNOW
/* On K6 femms is faster of emms. On K7 femms is directly mapped on emms. */ /* On K6 femms is faster of emms. On K7 femms is directly mapped on emms. */
...@@ -56,8 +63,12 @@ ...@@ -56,8 +63,12 @@
#define SFENCE "sfence" #define SFENCE "sfence"
#else #else
#define MOVNTQ "movq" #define MOVNTQ "movq"
#ifdef __APPLE__
#define SFENCE "#"
#elif
#define SFENCE "/nop" #define SFENCE "/nop"
#endif #endif
#endif
static inline void RENAME(rgb24to32)(const uint8_t *src,uint8_t *dst,long src_size) static inline void RENAME(rgb24to32)(const uint8_t *src,uint8_t *dst,long src_size)
{ {
...@@ -332,7 +343,7 @@ static inline void RENAME(rgb32to16)(const uint8_t *src, uint8_t *dst, long src_ ...@@ -332,7 +343,7 @@ static inline void RENAME(rgb32to16)(const uint8_t *src, uint8_t *dst, long src_
"movq %3, %%mm5 \n\t" "movq %3, %%mm5 \n\t"
"movq %4, %%mm6 \n\t" "movq %4, %%mm6 \n\t"
"movq %5, %%mm7 \n\t" "movq %5, %%mm7 \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 32(%1) \n\t" PREFETCH" 32(%1) \n\t"
"movd (%1), %%mm0 \n\t" "movd (%1), %%mm0 \n\t"
...@@ -489,7 +500,7 @@ static inline void RENAME(rgb32to15)(const uint8_t *src, uint8_t *dst, long src_ ...@@ -489,7 +500,7 @@ static inline void RENAME(rgb32to15)(const uint8_t *src, uint8_t *dst, long src_
"movq %3, %%mm5 \n\t" "movq %3, %%mm5 \n\t"
"movq %4, %%mm6 \n\t" "movq %4, %%mm6 \n\t"
"movq %5, %%mm7 \n\t" "movq %5, %%mm7 \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 32(%1) \n\t" PREFETCH" 32(%1) \n\t"
"movd (%1), %%mm0 \n\t" "movd (%1), %%mm0 \n\t"
...@@ -1344,7 +1355,7 @@ static inline void RENAME(rgb32tobgr32)(const uint8_t *src, uint8_t *dst, long s ...@@ -1344,7 +1355,7 @@ static inline void RENAME(rgb32tobgr32)(const uint8_t *src, uint8_t *dst, long s
/* TODO: unroll this loop */ /* TODO: unroll this loop */
asm volatile ( asm volatile (
"xor %%"REG_a", %%"REG_a" \n\t" "xor %%"REG_a", %%"REG_a" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 32(%0, %%"REG_a") \n\t" PREFETCH" 32(%0, %%"REG_a") \n\t"
"movq (%0, %%"REG_a"), %%mm0 \n\t" "movq (%0, %%"REG_a"), %%mm0 \n\t"
...@@ -1394,7 +1405,7 @@ static inline void RENAME(rgb24tobgr24)(const uint8_t *src, uint8_t *dst, long s ...@@ -1394,7 +1405,7 @@ static inline void RENAME(rgb24tobgr24)(const uint8_t *src, uint8_t *dst, long s
"movq "MANGLE(mask24r)", %%mm5 \n\t" "movq "MANGLE(mask24r)", %%mm5 \n\t"
"movq "MANGLE(mask24g)", %%mm6 \n\t" "movq "MANGLE(mask24g)", %%mm6 \n\t"
"movq "MANGLE(mask24b)", %%mm7 \n\t" "movq "MANGLE(mask24b)", %%mm7 \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 32(%1, %%"REG_a") \n\t" PREFETCH" 32(%1, %%"REG_a") \n\t"
"movq (%1, %%"REG_a"), %%mm0 \n\t" // BGR BGR BG "movq (%1, %%"REG_a"), %%mm0 \n\t" // BGR BGR BG
...@@ -1464,7 +1475,7 @@ static inline void RENAME(yuvPlanartoyuy2)(const uint8_t *ysrc, const uint8_t *u ...@@ -1464,7 +1475,7 @@ static inline void RENAME(yuvPlanartoyuy2)(const uint8_t *ysrc, const uint8_t *u
//FIXME handle 2 lines a once (fewer prefetch, reuse some chrom, but very likely limited by mem anyway) //FIXME handle 2 lines a once (fewer prefetch, reuse some chrom, but very likely limited by mem anyway)
asm volatile( asm volatile(
"xor %%"REG_a", %%"REG_a" \n\t" "xor %%"REG_a", %%"REG_a" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 32(%1, %%"REG_a", 2) \n\t" PREFETCH" 32(%1, %%"REG_a", 2) \n\t"
PREFETCH" 32(%2, %%"REG_a") \n\t" PREFETCH" 32(%2, %%"REG_a") \n\t"
...@@ -1617,7 +1628,7 @@ static inline void RENAME(yuvPlanartouyvy)(const uint8_t *ysrc, const uint8_t *u ...@@ -1617,7 +1628,7 @@ static inline void RENAME(yuvPlanartouyvy)(const uint8_t *ysrc, const uint8_t *u
//FIXME handle 2 lines a once (fewer prefetch, reuse some chrom, but very likely limited by mem anyway) //FIXME handle 2 lines a once (fewer prefetch, reuse some chrom, but very likely limited by mem anyway)
asm volatile( asm volatile(
"xor %%"REG_a", %%"REG_a" \n\t" "xor %%"REG_a", %%"REG_a" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 32(%1, %%"REG_a", 2) \n\t" PREFETCH" 32(%1, %%"REG_a", 2) \n\t"
PREFETCH" 32(%2, %%"REG_a") \n\t" PREFETCH" 32(%2, %%"REG_a") \n\t"
...@@ -1741,7 +1752,7 @@ static inline void RENAME(yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t ...@@ -1741,7 +1752,7 @@ static inline void RENAME(yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t
"xor %%"REG_a", %%"REG_a" \n\t" "xor %%"REG_a", %%"REG_a" \n\t"
"pcmpeqw %%mm7, %%mm7 \n\t" "pcmpeqw %%mm7, %%mm7 \n\t"
"psrlw $8, %%mm7 \n\t" // FF,00,FF,00... "psrlw $8, %%mm7 \n\t" // FF,00,FF,00...
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%"REG_a", 4) \n\t" PREFETCH" 64(%0, %%"REG_a", 4) \n\t"
"movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0) "movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0)
...@@ -1794,7 +1805,7 @@ static inline void RENAME(yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t ...@@ -1794,7 +1805,7 @@ static inline void RENAME(yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t
asm volatile( asm volatile(
"xor %%"REG_a", %%"REG_a" \n\t" "xor %%"REG_a", %%"REG_a" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%"REG_a", 4) \n\t" PREFETCH" 64(%0, %%"REG_a", 4) \n\t"
"movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0) "movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0)
...@@ -1979,7 +1990,7 @@ static inline void RENAME(uyvytoyv12)(const uint8_t *src, uint8_t *ydst, uint8_t ...@@ -1979,7 +1990,7 @@ static inline void RENAME(uyvytoyv12)(const uint8_t *src, uint8_t *ydst, uint8_t
"xorl %%eax, %%eax \n\t" "xorl %%eax, %%eax \n\t"
"pcmpeqw %%mm7, %%mm7 \n\t" "pcmpeqw %%mm7, %%mm7 \n\t"
"psrlw $8, %%mm7 \n\t" // FF,00,FF,00... "psrlw $8, %%mm7 \n\t" // FF,00,FF,00...
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%eax, 4) \n\t" PREFETCH" 64(%0, %%eax, 4) \n\t"
"movq (%0, %%eax, 4), %%mm0 \n\t" // UYVY UYVY(0) "movq (%0, %%eax, 4), %%mm0 \n\t" // UYVY UYVY(0)
...@@ -2032,7 +2043,7 @@ static inline void RENAME(uyvytoyv12)(const uint8_t *src, uint8_t *ydst, uint8_t ...@@ -2032,7 +2043,7 @@ static inline void RENAME(uyvytoyv12)(const uint8_t *src, uint8_t *ydst, uint8_t
asm volatile( asm volatile(
"xorl %%eax, %%eax \n\t" "xorl %%eax, %%eax \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%eax, 4) \n\t" PREFETCH" 64(%0, %%eax, 4) \n\t"
"movq (%0, %%eax, 4), %%mm0 \n\t" // YUYV YUYV(0) "movq (%0, %%eax, 4), %%mm0 \n\t" // YUYV YUYV(0)
...@@ -2110,7 +2121,7 @@ static inline void RENAME(rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_ ...@@ -2110,7 +2121,7 @@ static inline void RENAME(rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_
"movq "MANGLE(w1111)", %%mm5 \n\t" "movq "MANGLE(w1111)", %%mm5 \n\t"
"pxor %%mm7, %%mm7 \n\t" "pxor %%mm7, %%mm7 \n\t"
"lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t" "lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%"REG_b") \n\t" PREFETCH" 64(%0, %%"REG_b") \n\t"
"movd (%0, %%"REG_b"), %%mm0 \n\t" "movd (%0, %%"REG_b"), %%mm0 \n\t"
...@@ -2184,7 +2195,7 @@ static inline void RENAME(rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_ ...@@ -2184,7 +2195,7 @@ static inline void RENAME(rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_
"pxor %%mm7, %%mm7 \n\t" "pxor %%mm7, %%mm7 \n\t"
"lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t" "lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t"
"add %%"REG_b", %%"REG_b" \n\t" "add %%"REG_b", %%"REG_b" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%"REG_b") \n\t" PREFETCH" 64(%0, %%"REG_b") \n\t"
PREFETCH" 64(%1, %%"REG_b") \n\t" PREFETCH" 64(%1, %%"REG_b") \n\t"
......
swscale_template.c
View file @ 82551fd3
...@@ -16,6 +16,8 @@ ...@@ -16,6 +16,8 @@
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/ */
#include "asmalign.h"
#undef REAL_MOVNTQ #undef REAL_MOVNTQ
#undef MOVNTQ #undef MOVNTQ
#undef PAVGB #undef PAVGB
...@@ -71,7 +73,7 @@ ...@@ -71,7 +73,7 @@
"movq %%mm3, %%mm4 \n\t"\ "movq %%mm3, %%mm4 \n\t"\
"lea " offset "(%0), %%"REG_d" \n\t"\ "lea " offset "(%0), %%"REG_d" \n\t"\
"mov (%%"REG_d"), %%"REG_S" \n\t"\ "mov (%%"REG_d"), %%"REG_S" \n\t"\
".balign 16 \n\t" /* FIXME Unroll? */\ ASMALIGN16 /* FIXME Unroll? */\
"1: \n\t"\ "1: \n\t"\
"movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\ "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\
"movq " #x "(%%"REG_S", %%"REG_a", 2), %%mm2\n\t" /* srcData */\ "movq " #x "(%%"REG_S", %%"REG_a", 2), %%mm2\n\t" /* srcData */\
...@@ -98,7 +100,7 @@ ...@@ -98,7 +100,7 @@
#define YSCALEYUV2YV121 \ #define YSCALEYUV2YV121 \
"mov %2, %%"REG_a" \n\t"\ "mov %2, %%"REG_a" \n\t"\
".balign 16 \n\t" /* FIXME Unroll? */\ ASMALIGN16 /* FIXME Unroll? */\
"1: \n\t"\ "1: \n\t"\
"movq (%0, %%"REG_a", 2), %%mm0 \n\t"\ "movq (%0, %%"REG_a", 2), %%mm0 \n\t"\
"movq 8(%0, %%"REG_a", 2), %%mm1\n\t"\ "movq 8(%0, %%"REG_a", 2), %%mm1\n\t"\
...@@ -118,14 +120,14 @@ ...@@ -118,14 +120,14 @@
*/ */
#define YSCALEYUV2PACKEDX \ #define YSCALEYUV2PACKEDX \
"xor %%"REG_a", %%"REG_a" \n\t"\ "xor %%"REG_a", %%"REG_a" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"nop \n\t"\ "nop \n\t"\
"1: \n\t"\ "1: \n\t"\
"lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"\n\t"\ "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"\n\t"\
"mov (%%"REG_d"), %%"REG_S" \n\t"\ "mov (%%"REG_d"), %%"REG_S" \n\t"\
"movq "VROUNDER_OFFSET"(%0), %%mm3\n\t"\ "movq "VROUNDER_OFFSET"(%0), %%mm3\n\t"\
"movq %%mm3, %%mm4 \n\t"\ "movq %%mm3, %%mm4 \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"2: \n\t"\ "2: \n\t"\
"movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\ "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\
"movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* UsrcData */\ "movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* UsrcData */\
...@@ -143,7 +145,7 @@ ...@@ -143,7 +145,7 @@
"mov (%%"REG_d"), %%"REG_S" \n\t"\ "mov (%%"REG_d"), %%"REG_S" \n\t"\
"movq "VROUNDER_OFFSET"(%0), %%mm1\n\t"\ "movq "VROUNDER_OFFSET"(%0), %%mm1\n\t"\
"movq %%mm1, %%mm7 \n\t"\ "movq %%mm1, %%mm7 \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"2: \n\t"\ "2: \n\t"\
"movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\ "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\
"movq (%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* Y1srcData */\ "movq (%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* Y1srcData */\
...@@ -205,7 +207,7 @@ ...@@ -205,7 +207,7 @@
"punpcklwd %%mm5, %%mm5 \n\t"\ "punpcklwd %%mm5, %%mm5 \n\t"\
"punpcklwd %%mm5, %%mm5 \n\t"\ "punpcklwd %%mm5, %%mm5 \n\t"\
"xor %%"REG_a", %%"REG_a" \n\t"\ "xor %%"REG_a", %%"REG_a" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"1: \n\t"\ "1: \n\t"\
"movq (%0, %%"REG_a", 2), %%mm0 \n\t" /*buf0[eax]*/\ "movq (%0, %%"REG_a", 2), %%mm0 \n\t" /*buf0[eax]*/\
"movq (%1, %%"REG_a", 2), %%mm1 \n\t" /*buf1[eax]*/\ "movq (%1, %%"REG_a", 2), %%mm1 \n\t" /*buf1[eax]*/\
...@@ -258,7 +260,7 @@ ...@@ -258,7 +260,7 @@
"movq %%mm0, "CHR_MMX_FILTER_OFFSET"+8("#c")\n\t"\ "movq %%mm0, "CHR_MMX_FILTER_OFFSET"+8("#c")\n\t"\
"movq %%mm1, "LUM_MMX_FILTER_OFFSET"+8("#c")\n\t"\ "movq %%mm1, "LUM_MMX_FILTER_OFFSET"+8("#c")\n\t"\
"xor "#index", "#index" \n\t"\ "xor "#index", "#index" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
...@@ -290,7 +292,7 @@ ...@@ -290,7 +292,7 @@
#define REAL_YSCALEYUV2RGB(index, c) \ #define REAL_YSCALEYUV2RGB(index, c) \
"xor "#index", "#index" \n\t"\ "xor "#index", "#index" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
...@@ -356,7 +358,7 @@ ...@@ -356,7 +358,7 @@
#define REAL_YSCALEYUV2PACKED1(index, c) \ #define REAL_YSCALEYUV2PACKED1(index, c) \
"xor "#index", "#index" \n\t"\ "xor "#index", "#index" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\
"movq 4096(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\ "movq 4096(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\
...@@ -371,7 +373,7 @@ ...@@ -371,7 +373,7 @@
#define REAL_YSCALEYUV2RGB1(index, c) \ #define REAL_YSCALEYUV2RGB1(index, c) \
"xor "#index", "#index" \n\t"\ "xor "#index", "#index" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\
"movq 4096(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\ "movq 4096(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\
...@@ -420,7 +422,7 @@ ...@@ -420,7 +422,7 @@
#define REAL_YSCALEYUV2PACKED1b(index, c) \ #define REAL_YSCALEYUV2PACKED1b(index, c) \
"xor "#index", "#index" \n\t"\ "xor "#index", "#index" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
...@@ -439,7 +441,7 @@ ...@@ -439,7 +441,7 @@
// do vertical chrominance interpolation // do vertical chrominance interpolation
#define REAL_YSCALEYUV2RGB1b(index, c) \ #define REAL_YSCALEYUV2RGB1b(index, c) \
"xor "#index", "#index" \n\t"\ "xor "#index", "#index" \n\t"\
".balign 16 \n\t"\ ASMALIGN16\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
...@@ -1662,7 +1664,7 @@ static inline void RENAME(bgr24ToY)(uint8_t *dst, uint8_t *src, long width) ...@@ -1662,7 +1664,7 @@ static inline void RENAME(bgr24ToY)(uint8_t *dst, uint8_t *src, long width)
"movq "MANGLE(w1111)", %%mm5 \n\t" "movq "MANGLE(w1111)", %%mm5 \n\t"
"pxor %%mm7, %%mm7 \n\t" "pxor %%mm7, %%mm7 \n\t"
"lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t" "lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%"REG_b") \n\t" PREFETCH" 64(%0, %%"REG_b") \n\t"
"movd (%0, %%"REG_b"), %%mm0 \n\t" "movd (%0, %%"REG_b"), %%mm0 \n\t"
...@@ -1748,7 +1750,7 @@ static inline void RENAME(bgr24ToUV)(uint8_t *dstU, uint8_t *dstV, uint8_t *src1 ...@@ -1748,7 +1750,7 @@ static inline void RENAME(bgr24ToUV)(uint8_t *dstU, uint8_t *dstV, uint8_t *src1
"pxor %%mm7, %%mm7 \n\t" "pxor %%mm7, %%mm7 \n\t"
"lea (%%"REG_a", %%"REG_a", 2), %%"REG_b" \n\t" "lea (%%"REG_a", %%"REG_a", 2), %%"REG_b" \n\t"
"add %%"REG_b", %%"REG_b" \n\t" "add %%"REG_b", %%"REG_b" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
PREFETCH" 64(%0, %%"REG_b") \n\t" PREFETCH" 64(%0, %%"REG_b") \n\t"
PREFETCH" 64(%1, %%"REG_b") \n\t" PREFETCH" 64(%1, %%"REG_b") \n\t"
...@@ -2055,7 +2057,7 @@ static inline void RENAME(hScale)(int16_t *dst, int dstW, uint8_t *src, int srcW ...@@ -2055,7 +2057,7 @@ static inline void RENAME(hScale)(int16_t *dst, int dstW, uint8_t *src, int srcW
"movq "MANGLE(w02)", %%mm6 \n\t" "movq "MANGLE(w02)", %%mm6 \n\t"
"push %%"REG_BP" \n\t" // we use 7 regs here ... "push %%"REG_BP" \n\t" // we use 7 regs here ...
"mov %%"REG_a", %%"REG_BP" \n\t" "mov %%"REG_a", %%"REG_BP" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
"movzwl (%2, %%"REG_BP"), %%eax \n\t" "movzwl (%2, %%"REG_BP"), %%eax \n\t"
"movzwl 2(%2, %%"REG_BP"), %%ebx\n\t" "movzwl 2(%2, %%"REG_BP"), %%ebx\n\t"
...@@ -2093,7 +2095,7 @@ static inline void RENAME(hScale)(int16_t *dst, int dstW, uint8_t *src, int srcW ...@@ -2093,7 +2095,7 @@ static inline void RENAME(hScale)(int16_t *dst, int dstW, uint8_t *src, int srcW
"movq "MANGLE(w02)", %%mm6 \n\t" "movq "MANGLE(w02)", %%mm6 \n\t"
"push %%"REG_BP" \n\t" // we use 7 regs here ... "push %%"REG_BP" \n\t" // we use 7 regs here ...
"mov %%"REG_a", %%"REG_BP" \n\t" "mov %%"REG_a", %%"REG_BP" \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
"movzwl (%2, %%"REG_BP"), %%eax \n\t" "movzwl (%2, %%"REG_BP"), %%eax \n\t"
"movzwl 2(%2, %%"REG_BP"), %%ebx\n\t" "movzwl 2(%2, %%"REG_BP"), %%ebx\n\t"
...@@ -2142,7 +2144,7 @@ static inline void RENAME(hScale)(int16_t *dst, int dstW, uint8_t *src, int srcW ...@@ -2142,7 +2144,7 @@ static inline void RENAME(hScale)(int16_t *dst, int dstW, uint8_t *src, int srcW
asm volatile( asm volatile(
"pxor %%mm7, %%mm7 \n\t" "pxor %%mm7, %%mm7 \n\t"
"movq "MANGLE(w02)", %%mm6 \n\t" "movq "MANGLE(w02)", %%mm6 \n\t"
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
"mov %2, %%"REG_c" \n\t" "mov %2, %%"REG_c" \n\t"
"movzwl (%%"REG_c", %0), %%eax \n\t" "movzwl (%%"REG_c", %0), %%eax \n\t"
...@@ -2326,7 +2328,7 @@ FUNNY_Y_CODE ...@@ -2326,7 +2328,7 @@ FUNNY_Y_CODE
"xor %%"REG_a", %%"REG_a" \n\t" // i "xor %%"REG_a", %%"REG_a" \n\t" // i
"xor %%"REG_b", %%"REG_b" \n\t" // xx "xor %%"REG_b", %%"REG_b" \n\t" // xx
"xorl %%ecx, %%ecx \n\t" // 2*xalpha "xorl %%ecx, %%ecx \n\t" // 2*xalpha
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
"movzbl (%0, %%"REG_b"), %%edi \n\t" //src[xx] "movzbl (%0, %%"REG_b"), %%edi \n\t" //src[xx]
"movzbl 1(%0, %%"REG_b"), %%esi \n\t" //src[xx+1] "movzbl 1(%0, %%"REG_b"), %%esi \n\t" //src[xx+1]
...@@ -2523,7 +2525,7 @@ FUNNY_UV_CODE ...@@ -2523,7 +2525,7 @@ FUNNY_UV_CODE
"xor %%"REG_a", %%"REG_a" \n\t" // i "xor %%"REG_a", %%"REG_a" \n\t" // i
"xor %%"REG_b", %%"REG_b" \n\t" // xx "xor %%"REG_b", %%"REG_b" \n\t" // xx
"xorl %%ecx, %%ecx \n\t" // 2*xalpha "xorl %%ecx, %%ecx \n\t" // 2*xalpha
".balign 16 \n\t" ASMALIGN16
"1: \n\t" "1: \n\t"
"mov %0, %%"REG_S" \n\t" "mov %0, %%"REG_S" \n\t"
"movzbl (%%"REG_S", %%"REG_b"), %%edi \n\t" //src[xx] "movzbl (%%"REG_S", %%"REG_b"), %%edi \n\t" //src[xx]
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment