CPU Throttling

[amock]

Sometimes when my Talos II is under sustained heavy load I'll get the following messages in dmesg and then my CPUs won't throttle up and my fans are stuck at full power.  Has anyone else run into this or does anyone know how to reset this?  Currently I have to restart to get things back to normal.


[14289.353665] powernv-cpufreq: Pstate set to safe frequency
[14289.353673] powernv-cpufreq: PMSR = 6363631440000000
[14289.353677] powernv-cpufreq: CPU Frequency could be throttled

[ClassicHasClass]

Never seen that on my T2 (dual-4 or the current dual-8 with HSFs). How many cores, what type of cooling, and what were your core temperatures?

[amock]

Dual 18-core with the standard HSFs and indium pads.  I never saw temperatures above 83C, but I wasn't logging the temperatures so I only looked a few times.  I noticed that when this happens the CPU power usage is also stuck high, so that will probably keep the temperatures high and require fans.  I just don't know why it wouldn't throttle down the power once the CPUs are idle.

[ClassicHasClass]

Does messing with cpufreqctl help any? What does it say the current frequencies are? What does `sudo cpufreqctl --throttle-event` come up with (or `--throttle`)?

[amock]

Thanks for your suggestions.  I've been waiting for it to happen again, but so far it hasn't.  When it does I'll see what i can get from those.