[vorbis-dev] Re: Optimizing Vorbis for a DSP

[Johannes Sandvall]

On Fri 09 Apr 2004 18:58:09 CEST, Adam D. Moss wrote:
> Adam D. Moss wrote:
> >>Here is maybe a patch of the code. Should include both optimized TI
> >>assembler and reference c code. Patch available at:
> >>http://www.sandvall.nu/patch
> >
> >Anyone been able to take a look at this patch?  I haven't had an
> >opportunity yet, myself...
> 
> Well, now I have.  :)  I used the C version.  Unfortunately
> the results were not too pleasing -- the output sounded somewhat
> heavily distorted and (for some strange reason... massive lookup
> tables?) the binary footprint of lowmem-branch Tremor on x86 with
> this patch is around 100K (!) larger.

The idea was to develope a optimized version for DSP. The C-code is
just as reference. The FFT could easily be replaced by a PC optimized
version. There should already be one in the encoder. The memory
footprint on DSP using the supplied assembler functions are less than
of original Tremor. But no effort whatsoever have been done to
optimize the reference c code.

As for sound quality I actually get a better result ( better peak and
rms deviation compared to floating point) than the original
tremor. For the c code I'm not sure but shold be close to original
Tremor. Might have forgetted something in the patch...

/ Johannes Sandvall

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.