fluid-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [fluid-dev] Automatic polyphony reduction?


From: Josh Green
Subject: Re: [fluid-dev] Automatic polyphony reduction?
Date: Sat, 15 Oct 2005 12:33:49 -0700

On Sat, 2005-10-15 at 17:25 +0300, Kimmo Sundqvist wrote:
> Hello
> 
> Having the polyphony set at 256, I experienced the dropping out of sound 
> output here and there in Chris Tilton's Overture.mid.
> 
> This is different than the screeches I experienced earlier, which were 
> characterised by the volume of the noise being many times louder than the 
> music.
> 
> I set the polyphony to 64 voices, and that fixed it.  I am not sure how many 
> voices a 933MHz Pentium III can support.  I was thinking if it would benefit 
> anyone to have Fluidsynth automatically and temporarily reduce polyphony to 
> avoid problems with polyphony set too high.  It could maybe report the 
> reduction somewhere.
> 

This would be nice and it has been on my mind for a while.  Having the
ability to set a max CPU usage amount would be ideal.  Short term I
think I will probably just reduce the default polyphony to 64 until a
better solution is in place.

> I still have the Fluidsynth CPU usage at 99% all the time, so I cannot make 
> any estimates about the polyphonic capabilities of my machine.  Is there 
> anything I could do to find out why it does this?
> 

I'd recommend building FluidSynth standalone without enabling LADSPA.
It may be the LADSPA code that is causing problems.

Ohh, it just occurred to me that the PIII has problems with denormal
floating point numbers (very inefficient when processing very tiny
floating point numbers).  There were some fixes a long while back in
regards to this issue, but perhaps some have surfaced again.  The
symptoms before were that CPU usage would go to max when there was
silence, but would go back to normal usage when there was any audio.
Profiling would help to narrow down this problem, but its not a
completely trivial task to discuss over email.  A good test would be to
try FluidSynth on another machine with a different CPU (P4 I don't
believe has this issue, and I'm pretty sure AMD never did).

> -Kimmo S.
> 

Cheers.
        Josh Green

Attachment: signature.asc
Description: This is a digitally signed message part


reply via email to

[Prev in Thread] Current Thread [Next in Thread]