[MPlayer-users] Comment on menconder on multiprocessors.

[gabor]

On Fri, 2003-01-24 at 17:19, Jean-Marc Stévenne wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> On Friday 24 January 2003 17:14, Arpi wrote:
> > [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> > Hi,
> >
> > > > of course.
> > > > it's the main point of 3 passes: 3rd pass uses the statistics from 2nd
> > > > pass.
> > >
> > > It needs them all ?
> >
> > of course.
> >
> > otherwise it wouldn't be done in 2/3 passes but in one...
> >
> >
> > A'rpi / Astral & ESP-team
> 
> Thanks, A'rpi 
> 
> I was wondering if there was a fundamental reason for not combinning pass1 and 
> 2. Apparently there is one.
> 
> Just out of curiosity, what kind of global consolidation is done with the data 
> collected by pass=1?

a very non formal and probably not 100% correct answer:
in pass 1, for every frame mencoder saves the complexity of the frame..

in pass2 :
	-more complex frames can be less compressed => for a complicated frame
you need to use/reserve more space... 
	-so when the 2nd pass launches mencoder knows how much space does he
have, and then he distributes the free space between the frames
according to their complexity...
	-as you see at the beginning of pass2 mencoder has to see the
complexity for all the frames to be able to make the calculation

hope this helps,
gabor
> 
> Jean-Marc
> 
> 
> _______________________________________________
> RTFM!!!  http://www.MPlayerHQ.hu/DOCS
> Search:  http://www.MPlayerHQ.hu/cgi-bin/htsearch
> http://mplayerhq.hu/mailman/listinfo/mplayer-users