pavan kumar
2006-Apr-20 03:18 UTC
[Vorbis] High Latency of vorbisencoder on a hardware with no floatpoint unit
Hi All, I am porting the vorbis encoder to Symbian platform,(to support ogg recording) everything works fine for the emulator(PC), but then on the Hardware whenever I am trying to record it gives KErrOverFlow(buffer OverFlow). We get this error from the driver when the latency of the codec is too high. The main reason is that the hardware does not have the floatpoint unit and the vorbisencoder is floatpoint imple.. Fixed point tremor works properly on the hardware. Has anyone come across this kind of problem on the hardware? Can any one suggest how i can improve the performance of the vorbisencoder? Thanks for your time. Regards Pavan --------------------------------- Jiyo cricket on Yahoo! India cricket Yahoo! Messenger Mobile Stay in touch with your buddies all the time. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/vorbis/attachments/20060420/8f68982a/attachment.htm
Michael Smith
2006-Apr-20 04:26 UTC
[Vorbis] High Latency of vorbisencoder on a hardware with no floatpoint unit
On 4/20/06, pavan kumar <pavanchikkala@yahoo.co.in> wrote:> > Hi All, > > I am porting the vorbis encoder to Symbian platform,(to support ogg > recording) everything works fine for the emulator(PC), but then on the > Hardware whenever I am trying to record it gives KErrOverFlow(buffer > OverFlow). We get this error from the driver when the latency of the codec > is too high. The main reason is that the hardware does not have the > floatpoint unit and the vorbisencoder is floatpoint imple.. Fixed point > tremor works properly on the hardware. > Has anyone come across this kind of problem on the hardware? Can any one > suggest how i can improve the performance of the vorbisencoder?Really, there's not much you can do with the reference encoder on a platform with no floating point support; I don't think there's much point in trying. Instead, you should look at developing a fixed point implementation seperately. You'd probably want to do a number of things quite differently from the reference implementation for speed reasons, as well as obviously avoiding floating point everywhere. This is a pretty substantial project, of course - particularly if you want a good quality encoder. Mike