diff options
| author | Lynne <dev@lynne.ee> | 2022-01-29 03:08:16 +0100 |
|---|---|---|
| committer | Lynne <dev@lynne.ee> | 2022-01-29 03:08:16 +0100 |
| commit | 35080149ef5044694109037f6c6ee176382f357a (patch) | |
| tree | 19b34d312c430ac3fb2b0eebf393390c9fa5da0b /libavutil/x86 | |
| parent | 7e35e0224cb1e82bfefdedeff9db946a05363c83 (diff) | |
x86/tx_float: mark AVX2 functions as AVXSLOW
Makes Bulldozer prefer AVX functions rather than AVX2,
which are 64% slower:
AVX: 117653 decicycles in av_tx (fft), 1048535 runs, 41 skips
AVX2: 193385 decicycles in av_tx (fft), 1048561 runs, 15 skips
The only difference between both is that vgatherdpd is used in
the former. We don't want to mark them with the new SLOW_GATHER
flag however, since gathers are still faster on Haswell/Zen 2/3
than plain loads.
Diffstat (limited to 'libavutil/x86')
| -rw-r--r-- | libavutil/x86/tx_float_init.c | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/libavutil/x86/tx_float_init.c b/libavutil/x86/tx_float_init.c index 2bc0f4c52c..3a87711def 100644 --- a/libavutil/x86/tx_float_init.c +++ b/libavutil/x86/tx_float_init.c @@ -100,11 +100,11 @@ DECL_CD_DEF(fft_sr_ns_float_avx, FFT, 64, 131072, 2, 0, fft_sr_codelet_init_b8_i #if HAVE_AVX2_EXTERNAL DECL_CD_DEF(fft_sr_float_avx2, FFT, 64, 131072, 2, 0, fft_sr_codelet_init_b8_i2, - 288, AV_CPU_FLAG_AVX2, + 288, AV_CPU_FLAG_AVX2 | AV_CPU_FLAG_AVXSLOW, FF_TX_OUT_OF_PLACE) DECL_CD_DEF(fft_sr_ns_float_avx2, FFT, 64, 131072, 2, 0, fft_sr_codelet_init_b8_i2, - 352, AV_CPU_FLAG_AVX2, + 352, AV_CPU_FLAG_AVX2 | AV_CPU_FLAG_AVXSLOW, FF_TX_OUT_OF_PLACE | AV_TX_INPLACE | FF_TX_PRESHUFFLE) #endif #endif |