[SCM] libav/experimental: mmx optimized 24to32

[siretart at users.alioth.debian.org]

The following commit has been merged in the experimental branch:
commit 49a0c6eec83ace5faec994b8791c9eeb19778eed
Author: Nick Kurshev <nickols_k at mail.ru>
Date:   Sun Oct 28 11:48:50 2001 +0000

    mmx optimized 24to32
    
    Originally committed as revision 2511 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc

diff --git a/postproc/rgb2rgb.c b/postproc/rgb2rgb.c
index 5c85f10..dc5f062 100644
--- a/postproc/rgb2rgb.c
+++ b/postproc/rgb2rgb.c
@@ -3,14 +3,39 @@
 #include "rgb2rgb.h"
 #include "mmx.h"
 
-/* TODO: MMX optimization */
-
 void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
 {
   uint8_t *dest = dst;
   uint8_t *s = src;
   uint8_t *end;
+#ifdef HAVE_MMX
+  const uint64_t mask32 = 0x00FFFFFF00FFFFFFULL;
+  uint8_t *mm_end;
+#endif
   end = s + src_size;
+#ifdef HAVE_MMX
+  mm_end = (uint8_t*)((((unsigned long)end)/16)*16);
+  __asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory");
+  while(s < mm_end)
+  {
+    __asm __volatile(
+	"movd	%1, %%mm0\n\t"
+	"movd	3%1, %%mm1\n\t"
+	"movd	6%1, %%mm2\n\t"
+	"movd	9%1, %%mm3\n\t"
+	"punpckldq %%mm1, %%mm0\n\t"
+	"punpckldq %%mm3, %%mm2\n\t"
+	"pand	%%mm7, %%mm0\n\t"
+	"pand	%%mm7, %%mm2\n\t"
+	"movq	%%mm0, %0\n\t"
+	"movq	%%mm2, 8%0"
+	:"=m"(*dest)
+	:"m"(*s)
+	:"memory");
+    dest += 16;
+    s += 12;
+  }
+#endif
   while(s < end)
   {
     *dest++ = *s++;
@@ -20,6 +45,7 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
   }
 }
 
+/* TODO: MMX optimization */
 void rgb32to24(uint8_t *src,uint8_t *dst,uint32_t src_size)
 {
   uint8_t *dest = dst;
diff --git a/postproc/rgb2rgb_template.c b/postproc/rgb2rgb_template.c
index 5c85f10..dc5f062 100644
--- a/postproc/rgb2rgb_template.c
+++ b/postproc/rgb2rgb_template.c
@@ -3,14 +3,39 @@
 #include "rgb2rgb.h"
 #include "mmx.h"
 
-/* TODO: MMX optimization */
-
 void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
 {
   uint8_t *dest = dst;
   uint8_t *s = src;
   uint8_t *end;
+#ifdef HAVE_MMX
+  const uint64_t mask32 = 0x00FFFFFF00FFFFFFULL;
+  uint8_t *mm_end;
+#endif
   end = s + src_size;
+#ifdef HAVE_MMX
+  mm_end = (uint8_t*)((((unsigned long)end)/16)*16);
+  __asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory");
+  while(s < mm_end)
+  {
+    __asm __volatile(
+	"movd	%1, %%mm0\n\t"
+	"movd	3%1, %%mm1\n\t"
+	"movd	6%1, %%mm2\n\t"
+	"movd	9%1, %%mm3\n\t"
+	"punpckldq %%mm1, %%mm0\n\t"
+	"punpckldq %%mm3, %%mm2\n\t"
+	"pand	%%mm7, %%mm0\n\t"
+	"pand	%%mm7, %%mm2\n\t"
+	"movq	%%mm0, %0\n\t"
+	"movq	%%mm2, 8%0"
+	:"=m"(*dest)
+	:"m"(*s)
+	:"memory");
+    dest += 16;
+    s += 12;
+  }
+#endif
   while(s < end)
   {
     *dest++ = *s++;
@@ -20,6 +45,7 @@ void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
   }
 }
 
+/* TODO: MMX optimization */
 void rgb32to24(uint8_t *src,uint8_t *dst,uint32_t src_size)
 {
   uint8_t *dest = dst;

-- 
Libav/FFmpeg packaging