Re: [iiwusynth-devel] Bugs / optimization

[Peter Hanappe]

M. Nentwig wrote:
> Hello,

Hello Markus,

> Some profiling with parts of the code removed revealed, that the
> interpolation requires almost 50 % of the DSP time, the filter almost 20
> % and the buffer copying (pan / reverb / chorus) almost 20 %. With that
> in mind I implemented a couple of new interpolation algorithms, which
> can be selected from the command line with 'interp x' for the synth or
> 'interpc chan x' on a per-channel-basis.
> x=0: No interpolation
> x=1: linear interpolation
> x=4: old four-point interpolation algorithm, default
> x=7: sinc-interpolation, seven points
> I haven't yet figured out, whether '7' actually sounds better than '4'.
> But it needs more horsepower, that's for sure.
> Linear interpolation sounds quite OK in most cases.
> Some experiments with 3rd and 5th order methods were unsatisfying, I did
> not commit them to CVS.

If you look in the first version of iiwusynth you'll see that no
interpolation (truncation) and linear interpolation were available as
a compile option. I removed them because the four point interpolation
wasn't that much slower and the sound quality was (IMHO) better.
There's one thing that I think is important, too, for composers. Is's
that they somewhat have the guarantee that the music they make sounds
the same wherever it's being listened to and doesn't suffer from 
optimization on one configuration. So that was another argument to have
only 1 type of interpolation.

BTW, thanks for the profiling. Do you have an idea how much the DSP 
takes in respect to the parameter initialization in iiwu_voice_write? 
Also, I'm sure that we can optimize the buffer copying, too. Maybe using
SSE.

> Since the new gcc supports Intel's SSE instructions, it would be
> feasible to recode the DSP loop for that instruction set. Four float
> multiplications at the same time should really speed up interpolation...

It would be great to use SSE for the interpolation. Only to bad that you
can't use SSE to calculate the sum of four floats at once as in
sum(a0,a1,a2,a3).


Lates,
Peter

> Cheers
>
> Markus
>
> PS: The last changes in iiwu_voice.c were committed through a telnet
> login. I haven't compiled that myself yet.
>
>
>
> _______________________________________________
> iiwusynth-devel mailing list
> address@hidden
> http://mail.nongnu.org/mailman/listinfo/iiwusynth-devel