Simplify and optimize the 4th vertical lifting step of the MMX code
authorMichael Niedermayer <michaelni@gmx.at>
Tue, 21 Aug 2007 00:02:19 +0000 (00:02 +0000)
committerMichael Niedermayer <michaelni@gmx.at>
Tue, 21 Aug 2007 00:02:19 +0000 (00:02 +0000)
This also reduces the needed headroom in that step by 1 bit

Originally committed as revision 10164 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavcodec/i386/snowdsp_mmx.c

index 229c721..cfdfb95 100644 (file)
@@ -604,9 +604,8 @@ void ff_snow_vertical_compose97i_mmx(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, DWTE
         snow_vertical_compose_mmx_store(REG_c,"mm0","mm2","mm4","mm6")
         snow_vertical_compose_mmx_add(REG_S,"mm0","mm2","mm4","mm6")
         snow_vertical_compose_mmx_move("mm0","mm2","mm4","mm6","mm1","mm3","mm5","mm7")
-        snow_vertical_compose_mmx_r2r_add("mm0","mm2","mm4","mm6","mm0","mm2","mm4","mm6")
-        snow_vertical_compose_mmx_r2r_add("mm1","mm3","mm5","mm7","mm0","mm2","mm4","mm6")
         snow_vertical_compose_mmx_sra("1","mm0","mm2","mm4","mm6")
+        snow_vertical_compose_mmx_r2r_add("mm1","mm3","mm5","mm7","mm0","mm2","mm4","mm6")
         snow_vertical_compose_mmx_add(REG_a,"mm0","mm2","mm4","mm6")
         snow_vertical_compose_mmx_store(REG_a,"mm0","mm2","mm4","mm6")