Re: [Vorbis-dev] Re: Flash and native support for Theora and Vorbis.
"Ian Malone" <[email protected]> Sun, 22 Jul 2007 00:53:33 +0100
| Newsgroups | gmane.comp.multimedia.ogg.advocacy,gmane.comp.multimedia.ogg.vorbis.devel |
|---|---|
| Message-ID | <[email protected]> |
On 21/07/07, Tuomo Latto <[email protected]> wrote: > Ivo Emanuel Gonçalves wrote: > > On 7/21/07, Ian Malone <[email protected]> wrote: > >> Would they [Adobe] find it > >> more acceptable to provide lower level APIs to components > >> such as the MDCT allowing codecs built in Actionscript to > >> operate more efficiently? > > > > How would this work? I reckon the performance of a scriptting > > language cannot compare with a high-level programming language. > > I believe that was his point. I believe [i]MDCT is one of the > biggest resource hogs in Vorbis {en|de}coding? > They are presumably doing their MP3/whatever support natively > (as opposed to scripting language), so if they exposed the API, > the performance of custom decode scripts, such as Tor-Einar's(?), > could be drastically improved, maybe even up to a usable level. > That's exactly what I meant, enough interface to the native routines used for other codecs to avoid having to do the heavy lifting in script. I should point out though: 1. I know nothing about Actionscript. 2. I have a rough idea how the Vorbis codec works, but not in detail. 3. I don't know what the relative times are that Tor-Einar's[1][2] decoder spends doing the various bits. Things like the Ogg layer and the vector decode may be too specific to benefit from any kind of interface unrolling. Still, probably a better chance of this happening. [1] http://lists.xiph.org/pipermail/vorbis/2007-January/026677.html [2] http://flash.j-ogg.de/ The sound /is/ choppy, but though notice a bit a hesitancy in FF, it doesn't actually appear to push this computer too hard. -- imalone _______________________________________________ Advocacy mailing list [email protected] http://lists.xiph.org/mailman/listinfo/advocacy