[x265] [PATCH Review Only] asm: assembly code for pixel_sad_8x32

[chen]

>+    PROCESS_SAD_8x4
>+    PROCESS_SAD_8x4
>+
>+    movq        m1, [r2]
>+    movq        m2, [r2 + r3]
>+    lea         r2, [r2 + 2 * r3]
>+    movq        m3, [r0]
>+    movq        m4, [r0 + r1]
>+    lea         r0, [r0 + 2 * r1]
>+    punpcklqdq  m1, m2
>+    punpcklqdq  m3, m4
>+    psadbw      m1, m3
>+    paddd       m0, m1
>+    movq        m1, [r2]
>+    movq        m2, [r2 + r3]
>+    movq        m3, [r0]
>+    movq        m4, [r0 + r1]
>+    punpcklqdq  m1, m2
>+    punpcklqdq  m3, m4
>+    psadbw      m1, m3
>+    paddd       m0, m1

You only want to reduce two of LEA instruction, don't extend code, you can pass a param to PROCESS_SAD_8x4 and control by %if, like below:
PROCESS_SAD_8x4 1
PROCESS_SAD_8x4 1
PROCESS_SAD_8x4 0
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20131030/5e7076a6/attachment-0001.html>