Re: [Ant-phone-devel] latency

[Roland Stigge]

On Thu, 2003-11-20 at 10:21, Matthias Thomae wrote:
> I am not exactly sure what you mean: Newly generated data while 
> ant-phone is up? What I did was while ant-phone was DOWN, remove the old 
> calls file, place a call, and start ant-phone. It crashed again, with 
> the new calls file that now contained only one line (that's the file I 
> will provide to you privately).

Exactly this one. That would be useful.

> Do you (or anyone else) have a working low-delay setup of ant-phone? If 
> yes, what is the configuration?

It's the one ANT was initially been developed with: Fritz PCIv1 and
Intel 8X0 via ALSA INTEL8X0 driver. For me, it worked better than the
OSS ones. I'm debugging via a local ISDN telephone and digital telephone
switchboard. But the latter one shouldn't have any influence on delays
since ISDN is constructed as a real time network (more or less).

> Is it possible to configure (or patch) the sound driver for smaller 
> "fragment sizes"?

ALSA used to provide kernel module options to set the size of the in/out
buffers. These should be small since they somehow correlate with the
fragment size, as far as I know.

> Maybe I can do some test with my old wired ISDN card sometime in order 
> to rule out bluetooth, but it might be weeks or even months before I get 
> to do this...

That would be great. I didn't have the opportunity yet to test a
Bluetooth setup and would like to hear more about their capabilities.

bye,
  Roland

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