From: Henrik Gramner <henrik@gramner.com>
Date: Sun, 15 Jan 2017 13:52:29 +0000 (+0100)
Subject: x86util: Reduce code size of high bit-depth AVX LOAD_DIFF
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=3c7bf52c5b0a849458a45b5628ed1cc4b898da5f;p=libx264

x86util: Reduce code size of high bit-depth AVX LOAD_DIFF

AVX supports unaligned memory operands which makes the SATD code a bit denser.
---

diff --git a/common/x86/x86util.asm b/common/x86/x86util.asm
index c285f137..ea40bc87 100644
--- a/common/x86/x86util.asm
+++ b/common/x86/x86util.asm
@@ -741,25 +741,25 @@
 %if %6 ; %5 aligned?
     mova       %1, %4
     psubw      %1, %5
+%elif cpuflag(avx)
+    movu       %1, %4
+    psubw      %1, %5
 %else
     movu       %1, %4
     movu       %2, %5
     psubw      %1, %2
 %endif
 %else ; !HIGH_BIT_DEPTH
-%ifidn %3, none
     movh       %1, %4
     movh       %2, %5
+%ifidn %3, none
     punpcklbw  %1, %2
     punpcklbw  %2, %2
-    psubw      %1, %2
 %else
-    movh       %1, %4
     punpcklbw  %1, %3
-    movh       %2, %5
     punpcklbw  %2, %3
-    psubw      %1, %2
 %endif
+    psubw      %1, %2
 %endif ; HIGH_BIT_DEPTH
 %endmacro