On Thursday 14 February 2013 13:35:26 Ralf Mardorf did opine:
On Thu, 2013-02-14 at 12:06 -0500, Gene Heskett
wrote:
5 microsecond base thread jitter in its timing
I agree that the nvidia driver could cause issues, but 5 ms jitter? This
perhaps is an issue for the RTAI, but I guess not for the audio RT
kernels.
Possibly not. What I am pointing out is that unlike all the fussing about
it, xruns and all, and invoking all sorts of black magic speels to control
it, working virtually blind, on LAU/LAD, we were forced to write test
utilities that quantified the results in an easy to understand format. The
net result being that we have a pretty good idea of the causes of the
failures when the figures do blow up.
Certain motherboards are pure junk, and a few are truly outstanding given
their relatively meager specs on the box they ship in. The intel Atom dual
core boards being a case in point. The now almost out of the pipeline
D525MW board, with a gig of ram and onboard intel video being an excellent
example, turning in results consistently under 7 microseconds for a
base_thread jitter when running the base IO thread at a 25 microsecond
repeat rate. The parent servo_thread, running every millisecond, has
similar performance on this board. Testwise while carving metal, I have
run the servo_thread at 2 kilohertz with even better results, and I intend
to raise that a kilohertz at a time until I run out of cpu because it seems
to reduce the detected noise in a quadrature encoded spindle speed control
as it goes up, effectively reducing the quantization errors in such a
software only servo control.
This motherboard, an ASUS M2N_SLI Deluxe is relatively poor despite a quad
core phenom running at 2.1 ghz with 4G of dram.
It is currently reporting the 1ms servo thread as having 15 microseconds of
timing jitter, and nearly 45 microseconds of jitter for a 25 microsecond IO
thread, and the machine is quite obviously 'pegged out'.
This base thread is probably not of interest to you audio people, but many
motherboards will have far worse servo_thread timing dependabilities simply
because the context switch takes too much time. This is of course on an
RTAI patched kernel, but we are about to move to the xenomi patchset, as
its easier to do, is all user-space, and actually working at least as well
as the RTAI patches.
Cheers, Gene
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
My web page: <http://coyoteden.dyndns-free.com:85/gene> is up!
My views
<http://www.armchairpatriot.com/What%20Has%20America%20Become.shtml>
There comes a time in the affairs of a man when he has to take the bull
by the tail and face the situation.
-- W. C. Fields
I was taught to respect my elders, but its getting
harder and harder to find any...