* baling 8 seems to have the same speed

Originally committed as revision 575 to svn://svn.ffmpeg.org/ffmpeg/trunk

diff --git a/libavcodec/i386/dsputil_mmx.c b/libavcodec/i386/dsputil_mmx.c
index d1bf967..db2f528 100644 (file)
--- a/libavcodec/i386/dsputil_mmx.c
+++ b/libavcodec/i386/dsputil_mmx.c
@@ -270,7 +270,7 @@ static void put_pixels_mmx(UINT8 *block, const UINT8 *pixels, int line_size, int
     asm volatile
        (
         "lea (%3, %3), %%eax           \n\t"
-        ".balign 16                    \n\t"
+        ".balign 8                     \n\t"
         "1:                            \n\t"
         "movq (%1), %%mm0              \n\t"
         "movq (%1, %3), %%mm1          \n\t"