On Tue, Sep 20, 2005 at 07:02:48PM -0400, Ken Snider wrote: > Matthew Toseland wrote: > >Only that it's slow (which can hopefully be significantly improved on by > >using FECFile for progressive decoding rather than doing it all at once > >at the end), and requires segments of 128*32k = 4MB. BTW is there a > >significant decoding speed gain from using less redundancy? > > I don't know if any of the original whitepapers/benchmarks and so on were > saved and available or not, but I'll ping a few of the openCOLA founders > and see if I can get you some of the details.
Tornado codes would be ideal (being around 100 times faster than vandermonde codes, and scaling linearly rather than quadratically), but are patented. :| Admittedly the webshop is patented, but the tornado codes patent is more likely to stand up in court. How they expect it to be "incorporated into network protocols" beats me... internet standard ones anyway... I suppose they were trying to do a RAMBUS ;). My understanding is that all fast codes (e.g. rateless codes) are probably covered by various patents anyway. :| So we will probably have to stick with Vandermonde codes even if there is a technically suitable alternative. -- Matthew J Toseland - toad at amphibian.dyndns.org Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20050921/89184e72/attachment.pgp>
