[ardour-dev] Re: [ardour-users] MTC & MMC - CORRECTION

[Geoffrey Wossum]

On Wednesday 22 September 2004 8:53 pm, Third Option wrote:
> talking with Larry Lines (also on this list - he lurks a lot) on messenger
> - this was interesting that he just said about that:
>
> (20:45:02) :p larry: but you could bypass the kernel and generate your own
> timeclock so that the kernel developers would stop fucking it up

That would be RTAI (http://www.rtai.org).  Not a particularly fun thing to
play with.  Not actually hard to write code for.  But it can be hard setting
it up and tuning it to get the exact performance you want on a given system.

I tried using RTAI with a National Semiconductor Geode SC2200 (a Cyrix Media
GX derivative <shudder>), to generate a 1200 Hz modulation for a POCSAG
paging transmitter.  The Geode had too much jitter in it's interrupt handling
to really be up to that task.

Of course, most modern hardware from Intel and AMD can do much better than
that.  I saw reports on the RTAI mailing list of getting timers at several
thousand Hz working reliably with low-latency and jitter on modern hardware.
Although, I remember when I was still subscribed to the RTAI mailing list,
that some new Intel motherboard chipset were causing huge, unpredictable
latencies.  They finally figured out a workaround for it.  It goes to show
though that even today, getting realtime performance out of commodity
hardware running commodity operating systems is neither easy nor guaranteed.

BTW, for our design we ended up letting an Atmel ATMega128 handle all the
 real time stuff instead of trying to coax the Geode into doing it.

---
Geoffrey