[LTP] [PATCH v2] [RFC] pselect01: Tune thresholds

[Jan Stancek <jstancek@redhat.com>]

----- Original Message -----
> Hi!
> > > Hmm, strange, it looks like for 100us we got more or less all samples
> > > close to 160us but for 500us half of the samples are close to 560us and
> > > about 400 is more than 1000us, that does not look right. I wonder what
> > > happens there. And the same for the rest of the measurements everything
> > > but 100us sleeps has two peaks that are more or less 500us apart. So far
> > > my machines, even VMs, had one peak and some outliners.
> > > 
> > > Can you try to rerun the test with a realtime priority?
> > 
> > Ran as "chrt -f 50 ./measure", results attached.
> > 
> 
> Looks more or less the same and I still cannot reproduce the issue on
> any of the machines I have here.
> 
> Looking at errata for AMD cpus #778 says that TSC may drift under some
> circumstances:
> http://support.amd.com/TechDocs/51810_16h_00h-0Fh_Rev_Guide.pdf#unique_29
> 
> Can you try to disable P-state in BIOS or set cpufreq to performance?

Not sure if I can get easily to BIOS, it's a system in remote location.
As for cpufreq, it already was set to performance:

# cat /sys/devices/system/cpu/cpufreq/policy*/scaling_governor
performance
performance
performance
performance

Regards,
Jan