diff options
| author | James Almer <jamrial@gmail.com> | 2014-01-17 03:55:44 -0300 |
|---|---|---|
| committer | Clément Bœsch <clement@stupeflix.com> | 2014-01-17 14:16:38 +0100 |
| commit | d2a7314f1e0d8404eb82316ad7f12665c24d48e7 (patch) | |
| tree | 6ab77feebff979961809c70ee2dc49b0e03e9eed /libavcodec/x86/vp9dsp_init.c | |
| parent | 3ca7085ae2f7734746174762e76a9b2b1c6e8626 (diff) | |
| download | ffmpeg-d2a7314f1e0d8404eb82316ad7f12665c24d48e7.tar.gz | |
vp9/x86: add ff_vp9_loop_filter_[vh]_16_16_sse2().
Similar gains in performance as the SSSE3 version
Signed-off-by: James Almer <jamrial@gmail.com>
Diffstat (limited to 'libavcodec/x86/vp9dsp_init.c')
| -rw-r--r-- | libavcodec/x86/vp9dsp_init.c | 19 |
1 files changed, 15 insertions, 4 deletions
diff --git a/libavcodec/x86/vp9dsp_init.c b/libavcodec/x86/vp9dsp_init.c index 900efb3a0b..ab3396e098 100644 --- a/libavcodec/x86/vp9dsp_init.c +++ b/libavcodec/x86/vp9dsp_init.c @@ -177,10 +177,17 @@ itxfm_func(idct, idct, 32, avx); #undef itxfm_func #undef itxfm_funcs -void ff_vp9_loop_filter_v_16_16_ssse3(uint8_t *dst, ptrdiff_t stride, int E, int I, int H); -void ff_vp9_loop_filter_v_16_16_avx (uint8_t *dst, ptrdiff_t stride, int E, int I, int H); -void ff_vp9_loop_filter_h_16_16_ssse3(uint8_t *dst, ptrdiff_t stride, int E, int I, int H); -void ff_vp9_loop_filter_h_16_16_avx (uint8_t *dst, ptrdiff_t stride, int E, int I, int H); +#define lpf_funcs(size1, size2, opt) \ +void ff_vp9_loop_filter_v_##size1##_##size2##_##opt(uint8_t *dst, ptrdiff_t stride, \ + int E, int I, int H); \ +void ff_vp9_loop_filter_h_##size1##_##size2##_##opt(uint8_t *dst, ptrdiff_t stride, \ + int E, int I, int H) + +lpf_funcs(16, 16, sse2); +lpf_funcs(16, 16, ssse3); +lpf_funcs(16, 16, avx); + +#undef lpf_funcs #endif /* HAVE_YASM */ @@ -230,6 +237,10 @@ av_cold void ff_vp9dsp_init_x86(VP9DSPContext *dsp) init_fpel(2, 1, 16, avg, sse2); init_fpel(1, 1, 32, avg, sse2); init_fpel(0, 1, 64, avg, sse2); + if (ARCH_X86_64) { + dsp->loop_filter_16[0] = ff_vp9_loop_filter_h_16_16_sse2; + dsp->loop_filter_16[1] = ff_vp9_loop_filter_v_16_16_sse2; + } } if (EXTERNAL_SSSE3(cpu_flags)) { |