Re: [FFmpeg-devel] avfilter/vf_interlace and vf_tinterlace : remove avx version

2017-12-18 Thread Martin Vignali
> when running checkasm several times sse2 is mostly faster here, not always.
> But the difference is quite small.
> Since I´m not an SIMD expert I´m fine with this patch as long as no one
> with more expertise objects.
>
>
> Seems like Paul B Mahol is against.
And I don't have strong opinion on this, and i'm not an SIMD expert,
So i will drop this patch.

Martin
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


Re: [FFmpeg-devel] avfilter/vf_interlace and vf_tinterlace : remove avx version

2017-12-17 Thread Thomas Mundt
Hi,

2017-12-16 18:38 GMT+01:00 Martin Vignali :

> Hello,
>
> Following discussion "avfilter/vf_interlace : add checkasm for lowpass_line
> and AVX2 version"
> the AVX version seems to be slower than SSE
>
> Patch in attach remove it, for vf_interlace and vf_tinterlace (both use the
> same SIMD)
>

when running checkasm several times sse2 is mostly faster here, not always.
But the difference is quite small.
Since I´m not an SIMD expert I´m fine with this patch as long as no one
with more expertise objects.

Regards,
Thomas
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


[FFmpeg-devel] avfilter/vf_interlace and vf_tinterlace : remove avx version

2017-12-16 Thread Martin Vignali
Hello,

Following discussion "avfilter/vf_interlace : add checkasm for lowpass_line
and AVX2 version"
the AVX version seems to be slower than SSE

Patch in attach remove it, for vf_interlace and vf_tinterlace (both use the
same SIMD)

Martin


0004-avfilter-vf_interlace-and-vf_tinterlace.patch
Description: Binary data
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel