Skip to content

F
ffmpeg.wasm-core
Commit 96b956cc authored by Nick Kurshev's avatar Nick Kurshev
Browse files
Options

mmx, mmx2, 3dnow optimized 24to32 

Originally committed as revision 2512 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
parent 49a0c6ee
Hide whitespace changes
Inline Side-by-side
Showing with 72 additions and 4 deletions
postproc/rgb2rgb.c
View file @ 96b956cc
...@@ -3,6 +3,26 @@ ...@@ -3,6 +3,26 @@
#include "rgb2rgb.h" #include "rgb2rgb.h"
#include "mmx.h" #include "mmx.h"
#ifdef HAVE_3DNOW
#define PREFETCH "prefetch"
#define PREFETCHW "prefetchw"
#elif HAVE_MMX2
#define PREFETCH "prefetchnta"
#define PREFETCHW "prefetcht0"
#endif
#ifdef HAVE_3DNOW
#define EMMS "femms"
#else
#define EMMS "emms"
#endif
#ifdef HAVE_MMX2
#define MOVNTQ "movntq"
#else
#define MOVNTQ "movq"
#endif
void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size) void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
{ {
uint8_t *dest = dst; uint8_t *dest = dst;
...@@ -14,10 +34,23 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size) ...@@ -14,10 +34,23 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
#endif #endif
end = s + src_size; end = s + src_size;
#ifdef HAVE_MMX #ifdef HAVE_MMX
#ifdef PREFETCH
__asm __volatile(
PREFETCH" %0\n\t"
PREFETCH" 64%0\n\t"
PREFETCHW" %1\n\t"
PREFETCHW" 64%1\n\t"::"m"(*s),"m"(*dest):"memory");
#endif
mm_end = (uint8_t*)((((unsigned long)end)/16)*16); mm_end = (uint8_t*)((((unsigned long)end)/16)*16);
__asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory"); __asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory");
while(s < mm_end) while(s < mm_end)
{ {
#ifdef PREFETCH
__asm __volatile(
PREFETCH" 128%0\n\t"
PREFETCHW" 128%1"
::"m"(*s),"m"(*dest):"memory");
#endif
__asm __volatile( __asm __volatile(
"movd %1, %%mm0\n\t" "movd %1, %%mm0\n\t"
"movd 3%1, %%mm1\n\t" "movd 3%1, %%mm1\n\t"
...@@ -27,14 +60,15 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size) ...@@ -27,14 +60,15 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
"punpckldq %%mm3, %%mm2\n\t" "punpckldq %%mm3, %%mm2\n\t"
"pand %%mm7, %%mm0\n\t" "pand %%mm7, %%mm0\n\t"
"pand %%mm7, %%mm2\n\t" "pand %%mm7, %%mm2\n\t"
"movq %%mm0, %0\n\t" MOVNTQ" %%mm0, %0\n\t"
"movq %%mm2, 8%0" MOVNTQ" %%mm2, 8%0"
:"=m"(*dest) :"=m"(*dest)
:"m"(*s) :"m"(*s)
:"memory"); :"memory");
dest += 16; dest += 16;
s += 12; s += 12;
} }
__asm __volatile(EMMS:::"memory");
#endif #endif
while(s < end) while(s < end)
{ {
......
postproc/rgb2rgb_template.c
View file @ 96b956cc
...@@ -3,6 +3,26 @@ ...@@ -3,6 +3,26 @@
#include "rgb2rgb.h" #include "rgb2rgb.h"
#include "mmx.h" #include "mmx.h"
#ifdef HAVE_3DNOW
#define PREFETCH "prefetch"
#define PREFETCHW "prefetchw"
#elif HAVE_MMX2
#define PREFETCH "prefetchnta"
#define PREFETCHW "prefetcht0"
#endif
#ifdef HAVE_3DNOW
#define EMMS "femms"
#else
#define EMMS "emms"
#endif
#ifdef HAVE_MMX2
#define MOVNTQ "movntq"
#else
#define MOVNTQ "movq"
#endif
void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size) void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
{ {
uint8_t *dest = dst; uint8_t *dest = dst;
...@@ -14,10 +34,23 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size) ...@@ -14,10 +34,23 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
#endif #endif
end = s + src_size; end = s + src_size;
#ifdef HAVE_MMX #ifdef HAVE_MMX
#ifdef PREFETCH
__asm __volatile(
PREFETCH" %0\n\t"
PREFETCH" 64%0\n\t"
PREFETCHW" %1\n\t"
PREFETCHW" 64%1\n\t"::"m"(*s),"m"(*dest):"memory");
#endif
mm_end = (uint8_t*)((((unsigned long)end)/16)*16); mm_end = (uint8_t*)((((unsigned long)end)/16)*16);
__asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory"); __asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory");
while(s < mm_end) while(s < mm_end)
{ {
#ifdef PREFETCH
__asm __volatile(
PREFETCH" 128%0\n\t"
PREFETCHW" 128%1"
::"m"(*s),"m"(*dest):"memory");
#endif
__asm __volatile( __asm __volatile(
"movd %1, %%mm0\n\t" "movd %1, %%mm0\n\t"
"movd 3%1, %%mm1\n\t" "movd 3%1, %%mm1\n\t"
...@@ -27,14 +60,15 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size) ...@@ -27,14 +60,15 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
"punpckldq %%mm3, %%mm2\n\t" "punpckldq %%mm3, %%mm2\n\t"
"pand %%mm7, %%mm0\n\t" "pand %%mm7, %%mm0\n\t"
"pand %%mm7, %%mm2\n\t" "pand %%mm7, %%mm2\n\t"
"movq %%mm0, %0\n\t" MOVNTQ" %%mm0, %0\n\t"
"movq %%mm2, 8%0" MOVNTQ" %%mm2, 8%0"
:"=m"(*dest) :"=m"(*dest)
:"m"(*s) :"m"(*s)
:"memory"); :"memory");
dest += 16; dest += 16;
s += 12; s += 12;
} }
__asm __volatile(EMMS:::"memory");
#endif #endif
while(s < end) while(s < end)
{ {
......
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