[FFmpeg-devel] 4xm idct computation

yann.lepetitcorps at free.fr yann.lepetitcorps at free.fr
Tue Jan 10 02:08:33 CET 2012


> > > > Note that 99% of the CPU time is used by the zlib compression stage
> > > > => without the zlib stage this can already work on real time ...
> > >
> > > replacing zlib by a static huffman coder storing zero run length +
> > > value of the following non zero value in one symbol should be quite
> > > fast and efficient.
> > > A better compressing variant would be using a context adaptive
> > > arithmetic coder
> > >
> >
> > I have found somes fast compression tools on the Net such as :
> >
> >     QuickLZ
> >     SR2
> >     FLZP
> >
> [...]
>
> LZO seem a good candidate too :)
>
> And FLZP seem to be too text file specific :(
>

The Golomb-Rice compression algorith seem to a good cadidate too :)

Ok, the compression ratio isn't as good than with huffman or others
compression/decompression algo **but** the Colomb-Rice algo seem really very
speed on the other side

And http://fr.wikipedia.org/wiki/Codage_de_Rice seem to confirm this :

"Le codage de Rice fait partie des codages entropiques les plus utilisés,
lorsque les données à compresser présentent une distribution géométrique (ou
approchante) et que la vitesse de l'algorithme est un critère important."

=> the speed is really important because the Zlib compression spend 99% of the
CPU time into the pipeline, so I have at least one order of magnitude to win
here ...

On other side, any idea about if "reordered quantized RLE wavelet blocs" have a
good geometric distribution or not ... :(



@+
Yannoo




More information about the ffmpeg-devel mailing list