[FFmpeg-devel] Pipeline: H.264 speed improvements

Michael Niedermayer michaelni
Tue Dec 23 13:41:41 CET 2008


On Tue, Dec 23, 2008 at 07:21:31AM -0500, Jason Garrett-Glaser wrote:
> On Tue, Dec 23, 2008 at 7:19 AM, Jason Garrett-Glaser
> <darkshikari at gmail.com> wrote:
> >> i suggest:
> >> coeff= get_vlc2();
> >> if(coeff<0)
> >>    do escape with prefix= -coeff
> >
> > But coeff can be negative or positive: the only reserved value that it
> > cannot be is zero (the idea is to make this a coeff lookup table, NOT
> > a levelcode lookup table, if possible).  So we should have:
> >
> > coeff= get_vlc2();
> > if(!coeff)
> >    do escape with prefix= -coeff
> >
> > Also, what about the fact that we will need to "roll back" X bits in
> > the case of an escape?
> 
> Oh, I see what you're trying to do here; you want to re-use coeff in
> the escape process.  

yes


> I don't think this will work with my proposal
> given the fact that "coeff" will be "the actual coefficient value
> based on the X bits read, assuming there wasn't an escape", as I don't
> think its simple to re-derive the original bits from these X bits and
> probably easier to just show_bits/skip_bits as necessary...

what about
coeff= get_vlc2();
if(coeff > BIGGEST_ALLOWED_COEFF)
    do escape with prefix= coeff - BIGGEST_ALLOWED_COEFF;

?

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Many things microsoft did are stupid, but not doing something just because
microsoft did it is even more stupid. If everything ms did were stupid they
would be bankrupt already.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20081223/1ecda672/attachment.pgp>



More information about the ffmpeg-devel mailing list