diff options
| author | Scott LaVarnway <slavarnway@google.com> | 2014-07-29 16:47:34 -0700 |
|---|---|---|
| committer | Scott LaVarnway <slavarnway@google.com> | 2014-07-30 08:17:32 -0700 |
| commit | 521cf7e8795d38da8aa7b7356d97a6e561d02449 (patch) | |
| tree | d7205203e39400c194efbefa013c95a3ba7ec7e5 /vp9/common/vp9_rtcd_defs.pl | |
| parent | d6cf6eb570e8c458d89b5be728d35135dc6e2e08 (diff) | |
| download | libvpx-521cf7e8795d38da8aa7b7356d97a6e561d02449.tar libvpx-521cf7e8795d38da8aa7b7356d97a6e561d02449.tar.gz libvpx-521cf7e8795d38da8aa7b7356d97a6e561d02449.tar.bz2 libvpx-521cf7e8795d38da8aa7b7356d97a6e561d02449.zip | |
Neon version of vp9_sub_pixel_variance16x16(),
vp9_variance16x16(), and vp9_get16x16var().
On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~16.7%.
Change-Id: Ib163aa99f56e680194aabe00dacdd7f0899a4ecb
Diffstat (limited to 'vp9/common/vp9_rtcd_defs.pl')
| -rw-r--r-- | vp9/common/vp9_rtcd_defs.pl | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/vp9/common/vp9_rtcd_defs.pl b/vp9/common/vp9_rtcd_defs.pl index d58774924..d3d874dfc 100644 --- a/vp9/common/vp9_rtcd_defs.pl +++ b/vp9/common/vp9_rtcd_defs.pl @@ -420,7 +420,7 @@ add_proto qw/unsigned int vp9_variance64x64/, "const uint8_t *src_ptr, int sourc specialize qw/vp9_variance64x64 avx2/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; -specialize qw/vp9_variance16x16 mmx avx2/, "$sse2_x86inc"; +specialize qw/vp9_variance16x16 mmx avx2 neon/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance16x8 mmx/, "$sse2_x86inc"; @@ -435,7 +435,7 @@ add_proto qw/void vp9_get8x8var/, "const uint8_t *src_ptr, int source_stride, co specialize qw/vp9_get8x8var mmx/, "$sse2_x86inc"; add_proto qw/void vp9_get16x16var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum"; -specialize qw/vp9_get16x16var avx2/, "$sse2_x86inc"; +specialize qw/vp9_get16x16var avx2 neon/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance8x4/, "$sse2_x86inc"; @@ -483,7 +483,7 @@ add_proto qw/unsigned int vp9_sub_pixel_avg_variance32x32/, "const uint8_t *src_ specialize qw/vp9_sub_pixel_avg_variance32x32/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; -specialize qw/vp9_sub_pixel_variance16x16/, "$sse2_x86inc", "$ssse3_x86inc"; +specialize qw/vp9_sub_pixel_variance16x16 neon/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance16x16/, "$sse2_x86inc", "$ssse3_x86inc"; |