[FFmpeg-devel] swscale dither

Michael Niedermayer michaelni at gmx.at
Fri Jul 15 18:11:36 CEST 2011

On Fri, Jul 15, 2011 at 08:42:03AM -0700, Thomas Worth wrote:
> > Also you still use 32bit buffers all over the place for handling 10bit
> > data. ffmpegs code uses 32bit buffers only when needed to maintain
> > precission and 16bit otherwise. Which should make ffmpegs swscale
> > twice as fast on 10bit data once its all done with SIMD :)
> > I dont mind at all if it stays that way :))))
> > nor do i if you reimplement or take my code as its then less work for
> > me to maintain the 16/32bit buffer feature :)))))
> Wouldn't 32 bit be faster since you can pack a single 10 bit R,G,B
> pixel into 32 bits with padding?

thats not what ronalds code does.
ronald puts one 10bit sample in 32bit wasting 22bits and half cpu time

and its yuv not rgb (though that makes no difference to the argument)

> You're probably referring to separate
> R,G,B planes, but wouldn't it be better for performance to manipulate
> packed RGB rather than 16 bits for discrete R,G,B? Is there any
> provision for doing it this way?

packing 3 10bit samples into 32bit may be an interresting option for
some types of operations. if someone implements that and its faster
ill surely see that the code gets merged into ffmpeg.

Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

You can kill me, but you cannot change the truth.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20110715/d60c161b/attachment.asc>

More information about the ffmpeg-devel mailing list