preprocessor performance (was Re: [speex-dev] Memory leak in denoiser + a few questions)

[Jean-Marc Valin]

OK, so the problem doesn't seem to be the VAD specifically. Can you tell
me how much audio you had in the test? It may be that nothing's wrong
and the code just isn't so fast that you can do 100 channels. Or maybe
it just needs a bit of optimization...

        Jean-Marc

Le mer 31/03/2004 à 10:03, Steve Kann a écrit :
> Jean-Marc Valin wrote:
> 
> >If you set the denoiser to "on" and the VAD to "off", what difference
> >does it make in CPU time?
> >
> 
> 
> Same program, running on Athlon XP 1700+:
> 
> Test 1, using VAD, but AGC, denoise off:
> 
> stevek at canarsie:~/work/hms/app_conference $ time ./vad_test 
> /tmp/demo-instruct.sw 5
> reading from /tmp/demo-instruct.sw, repeating 5 times
> read 537760 samples
> beginning pass
> beginning pass
> beginning pass
> beginning pass
> beginning pass
> done.
>                                                                                 
> 
> real    0m4.970s
> user    0m4.628s
> sys     0m0.014s
> 
> 
> Test 2, using denoise only:
> 
> stevek at canarsie:~/work/hms/app_conference $ time ./vad_test 
> /tmp/demo-instruct.sw 5
> reading from /tmp/demo-instruct.sw, repeating 5 times
> read 537760 samples
> beginning pass
> beginning pass
> beginning pass
> beginning pass
> beginning pass
> done.
>  
> real    0m5.359s
> user    0m4.301s
> sys     0m0.024s
> 
> 
> ====================
> 
> So, it doesn't seem to make much difference.
> 
> I also ran the code, unoptimized, with oprofile.  I'll send results from 
> that to you separately.
> 
> -SteveK

-- 
Jean-Marc Valin
http://www.xiph.org/~jm/
LABORIUS
Université de Sherbrooke, Québec, Canada


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 190 bytes
Desc: Ceci est une partie de message numériquement signée.
Url : http://lists.xiph.org/pipermail/speex-dev/attachments/20040331/e07f80e5/signature.pgp