[MPlayer-dev-eng] New Codec Interface

[Michael Niedermayer]

Hi

On Wednesday 27 February 2002 13:34, Arpi wrote:
> Hi,
>
> > > I hope so. Michael? :)
> >
> > imho doing a row/slice of blocks is easyer and faster, as the current
> > cpus dont have enough non temporal write buffers afaik for these blocks
> > and for a
> >
> > slice the data comes from L2 cache and is written to agp memory, the
> > limitin g
> > factor will be agp mem, a per block variant would read from L1 but it
> > still writes to agp mem so i wouldnt expect much speedup especially as
> > the accesse s
> > are not sequential anymore
>
> you're right, but it won't skip skipped macroblocks...
>
> for low/middle rate divx/h263/mpg there are lots of skipped macroblocks
> (not updated at all by a frame) so don't copying them could mean less agp
> waiting/writting so overall speedup. even if for whole frame copy it will
> be a bit slower and harder to implement.
yes, but that would accelerate allready easy parts (bitstream reader, idct, 
mc have to do less too)
and would slow down hard parts, unless we support both and somehow switch 
between them ...

[...]

Michael