[x264-devel] commit: SSSE3 high bit depth sad_aligned functions (Jason Garrett-Glaser )

[git at videolan.org]

x264 | branch: master | Jason Garrett-Glaser <darkshikari at gmail.com> | Sat Dec 18 05:40:49 2010 -0800| [e570fc2e6305cc5b5dd0422c9bdd1babe37f3d1b] | committer: Jason Garrett-Glaser 

SSSE3 high bit depth sad_aligned functions

> http://git.videolan.org/gitweb.cgi/x264.git/?a=commit;h=e570fc2e6305cc5b5dd0422c9bdd1babe37f3d1b
---

 common/pixel.c         |    1 +
 common/x86/pixel.h     |    1 +
 common/x86/sad16-a.asm |    6 ++++++
 3 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/common/pixel.c b/common/pixel.c
index 4abd6fa..ee05eab 100644
--- a/common/pixel.c
+++ b/common/pixel.c
@@ -825,6 +825,7 @@ void x264_pixel_init( int cpu, x264_pixel_function_t *pixf )
     }
     if( cpu&X264_CPU_SSSE3 )
     {
+        INIT4_NAME( sad_aligned, sad, _ssse3_aligned );
         INIT7( sad, _ssse3 );
         INIT7( sad_x3, _ssse3 );
         INIT7( sad_x4, _ssse3 );
diff --git a/common/x86/pixel.h b/common/x86/pixel.h
index bff7efb..2893244 100644
--- a/common/x86/pixel.h
+++ b/common/x86/pixel.h
@@ -50,6 +50,7 @@ DECL_X4( sad, sse2_misalign )
 DECL_X1( sad, sse3 )
 DECL_X1( sad, sse2_aligned )
 DECL_X1( sad, ssse3 )
+DECL_X1( sad, ssse3_aligned )
 DECL_X4( sad, mmxext )
 DECL_X4( sad, sse2 )
 DECL_X4( sad, sse3 )
diff --git a/common/x86/sad16-a.asm b/common/x86/sad16-a.asm
index d19d05b..bb480ce 100644
--- a/common/x86/sad16-a.asm
+++ b/common/x86/sad16-a.asm
@@ -189,6 +189,12 @@ SAD_XMM 16,  8, ssse3
 SAD_XMM  8, 16, ssse3
 SAD_XMM  8,  8, ssse3
 SAD_XMM  8,  4, ssse3
+%define movdqu movdqa
+SAD_XMM 16, 16, ssse3_aligned
+SAD_XMM 16,  8, ssse3_aligned
+SAD_XMM  8, 16, ssse3_aligned
+SAD_XMM  8,  8, ssse3_aligned
+%undef movdqu
 %undef ABS1
 %undef ABS2