x86: XOP pixel_sad_{x3, x4} high bit-depth

diff --git a/common/pixel.c b/common/pixel.c
index 78356c9d2f080b67d4e2224c092592945e5a791e..17a01428e7d163f80f85f8ea3b0fd3102f7747d1 100644 (file)
--- a/common/pixel.c
+++ b/common/pixel.c
@@ -1021,6 +1021,8 @@ void x264_pixel_init( int cpu, x264_pixel_function_t *pixf )
     }
     if( cpu&X264_CPU_XOP )
     {
+        INIT5( sad_x3, _xop );
+        INIT5( sad_x4, _xop );
         pixf->ssd_nv12_core    = x264_pixel_ssd_nv12_core_xop;
         pixf->var[PIXEL_16x16] = x264_pixel_var_16x16_xop;
         pixf->var[PIXEL_8x8]   = x264_pixel_var_8x8_xop;

diff --git a/common/x86/pixel.h b/common/x86/pixel.h
index 0ff68bab74e83d3c656c4caef04a8389210ad438..7342a9b9f0f288dda0654920ed66e8360d9f5a38 100644 (file)
--- a/common/x86/pixel.h
+++ b/common/x86/pixel.h
@@ -56,6 +56,7 @@ DECL_X4( sad, mmx2 )
 DECL_X4( sad, sse2 )
 DECL_X4( sad, sse3 )
 DECL_X4( sad, ssse3 )
+DECL_X4( sad, xop )
 DECL_X4( sad, avx )
 DECL_X4( sad, avx2 )
 DECL_X1( ssd, mmx )

diff --git a/common/x86/sad16-a.asm b/common/x86/sad16-a.asm
index ebfe9b5cf39243b2fa0552f30e0efc17f36395e2..5b5e9c8ae8104a873d5241c7a79247d86266a28a 100644 (file)
--- a/common/x86/sad16-a.asm
+++ b/common/x86/sad16-a.asm
@@ -519,6 +519,19 @@ SAD_X 4, 16,  8
 SAD_X 4,  8, 16
 SAD_X 4,  8,  8
 SAD_X 4,  8,  4
+INIT_XMM xop
+%define XMM_REGS 7
+SAD_X 3, 16, 16
+SAD_X 3, 16,  8
+SAD_X 3,  8, 16
+SAD_X 3,  8,  8
+SAD_X 3,  8,  4
+%define XMM_REGS 9
+SAD_X 4, 16, 16
+SAD_X 4, 16,  8
+SAD_X 4,  8, 16
+SAD_X 4,  8,  8
+SAD_X 4,  8,  4
 INIT_YMM avx2
 %define XMM_REGS 7
 SAD_X 3, 16, 16