diff options
| author | Yunqing Wang <yunqingwang@google.com> | 2011-03-11 08:24:23 -0500 |
|---|---|---|
| committer | Yunqing Wang <yunqingwang@google.com> | 2011-03-11 08:24:23 -0500 |
| commit | b2aa4017760c9bf9d267470ac235650e24b8adc5 (patch) | |
| tree | 2e0af83e9567fa226af483d8332d9b622edbd3d3 | |
| parent | 6daacdb785f34efcf3c8600c65590199e79847d6 (diff) | |
| download | libvpx-b2aa4017760c9bf9d267470ac235650e24b8adc5.tar libvpx-b2aa4017760c9bf9d267470ac235650e24b8adc5.tar.gz libvpx-b2aa4017760c9bf9d267470ac235650e24b8adc5.tar.bz2 libvpx-b2aa4017760c9bf9d267470ac235650e24b8adc5.zip | |
Align SAD output array to be 16-byte aligned
Use aligned store.
Change-Id: Icab4c0c53da811d0c52bb7e8134927f249ba2499
| -rw-r--r-- | vp8/encoder/mcomp.c | 2 | ||||
| -rw-r--r-- | vp8/encoder/x86/sad_sse4.asm | 10 |
2 files changed, 6 insertions, 6 deletions
diff --git a/vp8/encoder/mcomp.c b/vp8/encoder/mcomp.c index 33aaa2ca9..c210c1de2 100644 --- a/vp8/encoder/mcomp.c +++ b/vp8/encoder/mcomp.c @@ -1415,7 +1415,7 @@ int vp8_full_search_sadx8(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *ref_mv, int er 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 diff --git a/vp8/encoder/x86/sad_sse4.asm b/vp8/encoder/x86/sad_sse4.asm index 21e2e5007..03ecec4b3 100644 --- a/vp8/encoder/x86/sad_sse4.asm +++ b/vp8/encoder/x86/sad_sse4.asm @@ -186,7 +186,7 @@ sym(vp8_sad16x16x8_sse4): 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 @@ sym(vp8_sad16x8x8_sse4): 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 @@ sym(vp8_sad8x8x8_sse4): 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 @@ sym(vp8_sad8x16x8_sse4): 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 @@ sym(vp8_sad4x4x8_sse4): PROCESS_4X2X8 0 mov rdi, arg(4) ;Results - movdqu XMMWORD PTR [rdi], xmm1 + movdqa XMMWORD PTR [rdi], xmm1 ; begin epilog pop rdi |