[FFmpeg-devel] [PATCH v2 1/5] libavfilter/x86/vf_gblur: add ff_postscale_slice_avx512()

[Wu, Jianhua]

> James Almer Wrote: 
> >>
> >> Yes, but we don't want to use AVX512 on hardware that downclocks
> heavily.
> >>
> >
> > It's okay. Should I updated the configure file to let the IceLake-AVX512 to
> be the minimum baseline or add a new macro name AVX512ICL?
> 
> Update the test in configure to look for an instruction available on Ice Lake
> (like VL), then update the check in cpu.c to make sure AVX2 is only enabled if
> AVX2-related flags Ice Lake and newer CPUs have in common are signaled.
> 

Hi there,

For this patchset, I receive two suggestions from you.

one is about to make avx512 codes only run for icelake
and newer architectures. So I submit a patchset for icelake-avx512
minimums baseline. Please check this link for more details:
http://ffmpeg.org/pipermail/ffmpeg-devel/2021-August/283973.html

The second is about giving the choice to users and developers.
Check this link for more details:
http://ffmpeg.org/pipermail/ffmpeg-devel/2021-August/283984.html

Because this patchset was developed based on the configurations
of AVX512 that FFmpeg already offered, It was totally independent
and compatible whatever the baseline of avx512 is, skylake-avx512,
icelake-avx512, or newer architectures. So whatever add a new avx512
flag for icelake or not, it would not affect the functionality of this patchset.

Could you help review the chat history and send me the modification plan?
I am glad to update and improve with whatever I am able to do. 

Best regards.
Jianhua