On 14/01/12 15:55, Paul Davis wrote:
On Sat, Jan 14, 2012 at 7:20 AM, Tvrtko
Ursulin<tvrtko(a)ursulin.net> wrote:
It may have been a hardware problem, but not the
sound card. I ended up
measuring latencies using cyclictest which was giving me shocking numbers. At
the end (cpufreq, rt kernel were in between attempts) it ended with disabling
C1E CPU state in the BIOS. For the record it is a Phenom II X2 CPU on a AMD
785G motherboard.
What BIOS do you have? I'd love to do the same thing here
but I'm not
sure I've seen anywhere in the BIOS to do this.
It is a ASUS M4A785TD-M EVO motherboard whose BIOS has C1E
enable/disable option exposed. It does not feel good to be missing on
power savings though. When I find some time I plan to dig for a better
solution.
Tvrtko
P.S. I seems that my yesterday's post with the attached screen shot
showing the distortion did not make it to the list. I basically played a
sine wave to pcm0&1 channels and captured channels 11&12 (which is
digital mixer). With that I proved that the corruption can only be in
the driver or in the PCM submission hardware (lets call it like that,
without knowing the ICE1712 architecture). Basically nature of
distortion is quite weird in that one to two samples every now and then
get slightly wrong values (not zero or max). This behaviour lasts for a
random number of seconds and goes away and back similarly.