[asterisk-dev] Re: Asterisk timing without hardware

[Tony Mountifield]

In article <AEC6C66638C05B468B556EA548C1A77DCEBF0D at trantor>,
James Harper <james.harper at bendigoit.com.au> wrote:
> > > Very good news :) from witch version of Asterisk is the
> > > internal_timing=yes setting available ?
> > > Do you mean that, using this setting no zaptel (even ztdummy) module
> is
> > > necessary ?
> > 
> > No, that's not what he means (I hope). On the contrary,
> > internal_timing=yes
> > DOES rely on having a zaptel timing source.
> 
> Kind of a shame. It would be nice if asterisk could make use of its own
> timing source (using the 2.6 HPET would be a good thing), so no ztdummy
> would be required.

Zaptel (and without hardware, ztdummy) does more than just provide timing.
For MeetMe, for instance, the mixing engine for the conferencing is within
the zaptel module. It is this module that needs invoking at device-driver
level 1000 times every second.

> Failing that, it would be nice if ztdummy could use the HPET. HPET would
> allow ztdummy to get timing spot on rather than the 1024Hz -> 1000Hz
> approximation that it does now.

It was I who submitted the RTC-based 1024->1000 patch for ztdummy, based
on the old zaprtc from Junghanns. If you have no zaptel hardware, you must
be using VoIP, which uses 20ms frames, so the small amount of jitter
caused by the 1024->1000 conversion is not usually significant. I have not
found it to be a problem in practice.

I wasn't aware of HPET at the time, and it looks worth investigating.

Cheers
Tony
-- 
Tony Mountifield
Work: tony at softins.co.uk - http://www.softins.co.uk
Play: tony at mountifield.org - http://tony.mountifield.org