ref: b2aa4017760c9bf9d267470ac235650e24b8adc5
parent: 6daacdb785f34efcf3c8600c65590199e79847d6
author: Yunqing Wang <yunqingwang@google.com>
date: Fri Mar 11 03:24:23 EST 2011
Align SAD output array to be 16-byte aligned Use aligned store. Change-Id: Icab4c0c53da811d0c52bb7e8134927f249ba2499
--- a/vp8/encoder/mcomp.c
+++ b/vp8/encoder/mcomp.c
@@ -1415,7 +1415,7 @@
int col_min = ref_col - distance;
int col_max = ref_col + distance;
- unsigned short sad_array8[8];
+ DECLARE_ALIGNED_ARRAY(16, unsigned short, sad_array8, 8);
unsigned int sad_array[3];
// Work out the mid point for the search
--- a/vp8/encoder/x86/sad_sse4.asm
+++ b/vp8/encoder/x86/sad_sse4.asm
@@ -186,7 +186,7 @@
PROCESS_16X2X8 0
mov rdi, arg(4) ;Results
- movdqu XMMWORD PTR [rdi], xmm1
+ movdqa XMMWORD PTR [rdi], xmm1
; begin epilog
pop rdi
@@ -224,7 +224,7 @@
PROCESS_16X2X8 0
mov rdi, arg(4) ;Results
- movdqu XMMWORD PTR [rdi], xmm1
+ movdqa XMMWORD PTR [rdi], xmm1
; begin epilog
pop rdi
@@ -262,7 +262,7 @@
PROCESS_8X2X8 0
mov rdi, arg(4) ;Results
- movdqu XMMWORD PTR [rdi], xmm1
+ movdqa XMMWORD PTR [rdi], xmm1
; begin epilog
pop rdi
@@ -303,7 +303,7 @@
PROCESS_8X2X8 0
PROCESS_8X2X8 0
mov rdi, arg(4) ;Results
- movdqu XMMWORD PTR [rdi], xmm1
+ movdqa XMMWORD PTR [rdi], xmm1
; begin epilog
pop rdi
@@ -339,7 +339,7 @@
PROCESS_4X2X8 0
mov rdi, arg(4) ;Results
- movdqu XMMWORD PTR [rdi], xmm1
+ movdqa XMMWORD PTR [rdi], xmm1
; begin epilog
pop rdi