diff options
| author | Ronald S. Bultje <rsbultje@gmail.com> | 2016-07-08 17:51:56 -0400 |
|---|---|---|
| committer | Ronald S. Bultje <rsbultje@gmail.com> | 2016-07-11 10:14:58 -0400 |
| commit | f0a2b6249bb2426befa4c03247361268e45b13af (patch) | |
| tree | f86bf79a94cef01c09081900d6effe66811eaa1f /libavcodec/x86/vp9dsp_init.c | |
| parent | a2550e7d2e4c1e2f1fd262985a1d789240f7e839 (diff) | |
| download | ffmpeg-f0a2b6249bb2426befa4c03247361268e45b13af.tar.gz | |
vp9: add 16x16 idct avx2 (8-bit).
checkasm --bench, 10k runs, for *_add_${bpc}_${sub_idct}_${opt}, shows
that it's about 1.65x as fast as the AVX version for the full IDCT, and
similar speedups for the sub-IDCTs:
nop: 24.6
vp9_inv_dct_dct_16x16_add_8_1_c: 6444.8
vp9_inv_dct_dct_16x16_add_8_1_sse2: 638.6
vp9_inv_dct_dct_16x16_add_8_1_ssse3: 484.4
vp9_inv_dct_dct_16x16_add_8_1_avx: 661.2
vp9_inv_dct_dct_16x16_add_8_1_avx2: 311.5
vp9_inv_dct_dct_16x16_add_8_2_c: 6665.7
vp9_inv_dct_dct_16x16_add_8_2_sse2: 646.9
vp9_inv_dct_dct_16x16_add_8_2_ssse3: 455.2
vp9_inv_dct_dct_16x16_add_8_2_avx: 521.9
vp9_inv_dct_dct_16x16_add_8_2_avx2: 304.3
vp9_inv_dct_dct_16x16_add_8_4_c: 7022.7
vp9_inv_dct_dct_16x16_add_8_4_sse2: 647.4
vp9_inv_dct_dct_16x16_add_8_4_ssse3: 467.1
vp9_inv_dct_dct_16x16_add_8_4_avx: 446.1
vp9_inv_dct_dct_16x16_add_8_4_avx2: 297.0
vp9_inv_dct_dct_16x16_add_8_8_c: 6800.4
vp9_inv_dct_dct_16x16_add_8_8_sse2: 598.6
vp9_inv_dct_dct_16x16_add_8_8_ssse3: 465.7
vp9_inv_dct_dct_16x16_add_8_8_avx: 440.9
vp9_inv_dct_dct_16x16_add_8_8_avx2: 290.2
vp9_inv_dct_dct_16x16_add_8_16_c: 6626.6
vp9_inv_dct_dct_16x16_add_8_16_sse2: 599.5
vp9_inv_dct_dct_16x16_add_8_16_ssse3: 475.0
vp9_inv_dct_dct_16x16_add_8_16_avx: 469.9
vp9_inv_dct_dct_16x16_add_8_16_avx2: 286.4
Diffstat (limited to 'libavcodec/x86/vp9dsp_init.c')
| -rw-r--r-- | libavcodec/x86/vp9dsp_init.c | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/libavcodec/x86/vp9dsp_init.c b/libavcodec/x86/vp9dsp_init.c index 469a661717..6fa63ae0ec 100644 --- a/libavcodec/x86/vp9dsp_init.c +++ b/libavcodec/x86/vp9dsp_init.c @@ -114,6 +114,7 @@ itxfm_func(idct, idct, 32, sse2); itxfm_func(idct, idct, 32, ssse3); itxfm_func(idct, idct, 32, avx); itxfm_func(iwht, iwht, 4, mmx); +itxfm_func(idct, idct, 16, avx2); #undef itxfm_func #undef itxfm_funcs @@ -382,6 +383,7 @@ av_cold void ff_vp9dsp_init_x86(VP9DSPContext *dsp, int bpp, int bitexact) init_fpel_func(0, 1, 64, avg, _8, avx2); if (ARCH_X86_64) { #if ARCH_X86_64 && HAVE_AVX2_EXTERNAL + dsp->itxfm_add[TX_16X16][DCT_DCT] = ff_vp9_idct_idct_16x16_add_avx2; init_subpel3_32_64(0, put, 8, avx2); init_subpel3_32_64(1, avg, 8, avx2); #endif |