ZynAddSubFX was: Re: [linux-audio-dev] some new soundfiles on-line

[Jens M Andreasen]

On Fri, 2005-05-13 at 17:02 -0400, Lee Revell wrote:
> On Fri, 2005-05-13 at 15:46 -0400, Paul Davis wrote:
> > >On Fri, 2005-05-13 at 21:02 +0200, Jens M Andreasen wrote:
> > >
> > >> 
> > >> Say what? Where did the "too many notes" go? Where did the jack xrun go?
> > >> 
> > >
> > >They are back (kind of.)
> > >
> > >Now I get this everytime I hit the virtual keyboard in zyn:
> > >
> > >  **** alsa_pcm: xrun of at least 1.364 msecs
> > 
> > paul knows that zyn is not RT-safe. he had several conversations with
> > me at LAC2005 about how to redesign the internals. it works fine at
> > larger latencies.
> 
> Judging from those messages, it seems like it should certainly work at
> 128 frames, maybe 64.  But 32 is insanely low - too low for most sound
> hardware.  That setting is more appropriate for stress testing the
> kernel, than for actual music making.

The hardware in question is an Aureal Vortex running with the snd-8820
driver on a VIA chipset in a combination that previously was notoriously
buggy. Read my lips; it ain't anymore!

I remember when Benno was posting benchmarks showing that the
(Ingo-patched) kernel could reach the ~2ms barrier, and we were all very
happy. I have later mentioned here on this list that I am running
reliable with 1ms latency and +-0.3ms jitter. Nobody raised an eybrow
then, maybe you guys thought I was kidding you, but I was not.

I know that this is an unusual happy combo, and I have probably put one
or two people off posting code that can't run reliable in the general
case ...

The driving keyboard is a DX7 mk1. The trigger point is about 5mm below
idle. With the setup I mentioned, I get sound before my finger feels
that I have hit rock-bottom. This is funky!

As for insanity? My belief is that going below 0.3ms latency is
borderline.

> 
> Lee
> 
-- 

   (
    )
  c[]  //  Jens M Andreasen