/*****************************************************************************
* pixel.c: pixel metrics
*****************************************************************************
- * Copyright (C) 2003-2013 x264 project
+ * Copyright (C) 2003-2014 x264 project
*
* Authors: Loren Merritt <lorenm@u.washington.edu>
* Laurent Aimar <fenrir@via.ecp.fr>
#if HAVE_MMX
#if HIGH_BIT_DEPTH
+#define x264_predict_8x8c_v_mmx2 x264_predict_8x8c_v_mmx
+#define x264_predict_8x16c_v_mmx2 x264_predict_8x16c_v_c
#define x264_predict_8x8c_v_sse2 x264_predict_8x8c_v_sse
#define x264_predict_8x16c_v_sse2 x264_predict_8x16c_v_sse
#define x264_predict_16x16_v_sse2 x264_predict_16x16_v_sse
INTRA_MBCMP( sad, 4x4, v, h, dc, , _mmx2, _c )
-INTRA_MBCMP( sad, 8x8, dc, h, v, c, _mmx2, _c )
+INTRA_MBCMP( sad, 8x8, dc, h, v, c, _mmx2, _mmx2 )
+INTRA_MBCMP( sad, 8x16, dc, h, v, c, _mmx2, _mmx2 )
+INTRA_MBCMP(satd, 8x16, dc, h, v, c, _mmx2, _mmx2 )
INTRA_MBCMP( sad, 16x16, v, h, dc, , _mmx2, _mmx2 )
INTRA_MBCMP( sad, 8x8, dc, h, v, c, _sse2, _sse2 )
+INTRA_MBCMP( sad, 8x16, dc, h, v, c, _sse2, _sse2 )
+INTRA_MBCMP(satd, 8x16, dc, h, v, c, _sse2, _sse2 )
INTRA_MBCMP( sad, 16x16, v, h, dc, , _sse2, _sse2 )
INTRA_MBCMP( sad, 8x8, dc, h, v, c, _ssse3, _sse2 )
+INTRA_MBCMP( sad, 8x16, dc, h, v, c, _ssse3, _sse2 )
+INTRA_MBCMP(satd, 8x16, dc, h, v, c, _ssse3, _sse2 )
INTRA_MBCMP( sad, 16x16, v, h, dc, , _ssse3, _sse2 )
+INTRA_MBCMP(satd, 8x16, dc, h, v, c, _sse4, _sse2 )
+INTRA_MBCMP(satd, 8x16, dc, h, v, c, _avx, _sse2 )
#else
#define x264_predict_8x16c_v_mmx2 x264_predict_8x16c_v_mmx
INTRA_MBCMP( sad, 8x16, dc, h, v, c, _mmx2, _mmx2 )
if( cpu&X264_CPU_MMX2 )
{
INIT7( sad, _mmx2 );
+ INIT7_NAME( sad_aligned, sad, _mmx2 );
INIT7( sad_x3, _mmx2 );
INIT7( sad_x4, _mmx2 );
INIT8( satd, _mmx2 );
pixf->intra_sad_x3_8x8 = x264_intra_sad_x3_8x8_mmx2;
pixf->intra_sad_x3_8x8c = x264_intra_sad_x3_8x8c_mmx2;
pixf->intra_satd_x3_8x8c = x264_intra_satd_x3_8x8c_mmx2;
+ pixf->intra_sad_x3_8x16c = x264_intra_sad_x3_8x16c_mmx2;
+ pixf->intra_satd_x3_8x16c = x264_intra_satd_x3_8x16c_mmx2;
pixf->intra_sad_x3_16x16 = x264_intra_sad_x3_16x16_mmx2;
pixf->intra_satd_x3_16x16 = x264_intra_satd_x3_16x16_mmx2;
}
pixf->asd8 = x264_pixel_asd8_sse2;
pixf->intra_sad_x3_8x8 = x264_intra_sad_x3_8x8_sse2;
pixf->intra_sad_x3_8x8c = x264_intra_sad_x3_8x8c_sse2;
+ pixf->intra_sad_x3_8x16c = x264_intra_sad_x3_8x16c_sse2;
+ pixf->intra_satd_x3_8x16c = x264_intra_satd_x3_8x16c_sse2;
pixf->intra_sad_x3_16x16 = x264_intra_sad_x3_16x16_sse2;
}
if( cpu&X264_CPU_SSE2_IS_FAST )
if( cpu&X264_CPU_SSSE3 )
{
INIT4_NAME( sad_aligned, sad, _ssse3_aligned );
+ pixf->sad_aligned[PIXEL_4x4] = x264_pixel_sad_4x4_ssse3;
+ pixf->sad_aligned[PIXEL_4x8] = x264_pixel_sad_4x8_ssse3;
INIT7( sad, _ssse3 );
INIT7( sad_x3, _ssse3 );
INIT7( sad_x4, _ssse3 );
pixf->intra_sad_x3_4x4 = x264_intra_sad_x3_4x4_ssse3;
pixf->intra_sad_x3_8x8 = x264_intra_sad_x3_8x8_ssse3;
pixf->intra_sad_x3_8x8c = x264_intra_sad_x3_8x8c_ssse3;
+ pixf->intra_sad_x3_8x16c = x264_intra_sad_x3_8x16c_ssse3;
+ pixf->intra_satd_x3_8x16c = x264_intra_satd_x3_8x16c_ssse3;
pixf->intra_sad_x3_16x16 = x264_intra_sad_x3_16x16_ssse3;
}
if( cpu&X264_CPU_SSE4 )
#if ARCH_X86_64
pixf->sa8d_satd[PIXEL_16x16] = x264_pixel_sa8d_satd_16x16_sse4;
#endif
+ pixf->intra_satd_x3_8x16c = x264_intra_satd_x3_8x16c_sse4;
}
if( cpu&X264_CPU_AVX )
{
+ INIT5_NAME( sad_aligned, sad, _ssse3 ); /* AVX-capable CPUs doesn't benefit from an aligned version */
INIT_ADS( _avx );
INIT6( satd, _avx );
pixf->satd[PIXEL_4x16] = x264_pixel_satd_4x16_avx;
#if ARCH_X86_64
pixf->sa8d_satd[PIXEL_16x16] = x264_pixel_sa8d_satd_16x16_avx;
#endif
+ pixf->intra_satd_x3_8x16c = x264_intra_satd_x3_8x16c_avx;
}
if( cpu&X264_CPU_XOP )
{
INIT2( sad_x3, _avx2 );
INIT2( sad_x4, _avx2 );
pixf->vsad = x264_pixel_vsad_avx2;
+ pixf->ssd_nv12_core = x264_pixel_ssd_nv12_core_avx2;
+ pixf->intra_sad_x3_8x8 = x264_intra_sad_x3_8x8_avx2;
}
#endif // HAVE_MMX
#else // !HIGH_BIT_DEPTH
pixf->sad_x4[PIXEL_8x16] = x264_pixel_sad_x4_8x16_cache64_sse2;
}
}
-
- if( cpu&X264_CPU_SSE_MISALIGN )
- {
- INIT2( sad_x3, _sse2_misalign );
- INIT2( sad_x4, _sse2_misalign );
- }
}
if( cpu&X264_CPU_SSE2_IS_FAST && !(cpu&X264_CPU_CACHELINE_64) )
}
else
{
- pixf->sad_x4[PIXEL_8x4] = x264_pixel_sad_x4_8x4_ssse3;
- pixf->sad_x4[PIXEL_8x8] = x264_pixel_sad_x4_8x8_ssse3;
- pixf->sad_x4[PIXEL_8x16] = x264_pixel_sad_x4_8x16_ssse3;
+ INIT2( sad_x3, _ssse3 );
+ INIT5( sad_x4, _ssse3 );
}
if( (cpu&X264_CPU_SLOW_ATOM) || (cpu&X264_CPU_SLOW_SHUFFLE) )
{
if( cpu&X264_CPU_AVX )
{
+ INIT2_NAME( sad_aligned, sad, _sse2 ); /* AVX-capable CPUs doesn't benefit from an aligned version */
+ INIT2( sad_x3, _avx );
+ INIT2( sad_x4, _avx );
INIT8( satd, _avx );
INIT7( satd_x3, _avx );
INIT7( satd_x4, _avx );
pixf->var2[PIXEL_8x8] = x264_pixel_var2_8x8_avx2;
pixf->intra_sad_x3_16x16 = x264_intra_sad_x3_16x16_avx2;
pixf->intra_sad_x9_8x8 = x264_intra_sad_x9_8x8_avx2;
+ pixf->intra_sad_x3_8x8c = x264_intra_sad_x3_8x8c_avx2;
+ pixf->ssd_nv12_core = x264_pixel_ssd_nv12_core_avx2;
#if ARCH_X86_64
pixf->sa8d_satd[PIXEL_16x16] = x264_pixel_sa8d_satd_16x16_avx2;
#endif