[FFmpeg-devel] [ping] [PATCH] mmx implementation of vc-1 inverse transformations

Ronald S. Bultje rsbultje
Sat Oct 2 15:07:58 CEST 2010


Hi,

On Sat, Oct 2, 2010 at 4:19 AM, Kostya <kostya.shishkov at gmail.com> wrote:
> On Fri, Oct 01, 2010 at 11:45:24PM +0400, Yuriy Kaminskiy wrote:
>> Yuriy Kaminskiy wrote:
>> > Yuriy Kaminskiy wrote:
>> >> Hello!
>> >>
>> >> I've noticed old and forgotten patch series by Victor Pollex
>> >> (http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2008-July/050503.html),
>> >> forward-ported it to current ffmpeg, it seems gives 12%-20% faster decoding
>> >> (benchmarked with ffmpeg [...] -f yuv4mpeg /dev/null, verified with -f framecrc).
>> [...]
>> >> be easier to drop that and just use _transposed flags.
>> >> I've also fixed problematic asm arguments: s/(0x\d\d)%0/$1(%0)/.
>> > Doh. Some of this kind somehow slipped (I was totally sure I've replaced all
>> > them long time ago :-|). Fixed version attached.
>>
>> ... and during forward porting ff_vc1dsp_init_sse2() call somehow landed in
>> wrong place [3dnow section instead of sse2] (coincidently it happen to work on
>> my cpu :-|).
>> Only difference with 40_vc1dsp_sse2-4.patch - move this call to proper place.
>
> I'm leaving this to review for somebody who knows x86 SIMD better than
> me (Michael/Jason/Loren/somebody else).
>
> Even better if Ronald would yasmify it but that's a mere wish.

Not necessarily, but the patch does miss clobber marking which I
intend to start making a big deal about. No need to keep re-breaking
Win64.

(In other words, can this hold a little until we finish the
xmm-clobber-bikeshed in the relevant thread? :-).)

Ronald



More information about the ffmpeg-devel mailing list