0

This is an issue widely discussed earlier. Still, none of the suggested methods/procedures has worked for me.

I'm running Ubuntu 14.04, kernel 4.4.0-040400-generic on a fanless mini computer, Intel Celeron J1800 dual core cpu 2.41GHz. MB:

lspci 00:00.0 Host bridge: Intel Corporation Atom Processor Z36xxx/Z37xxx Series SoC Transaction Register (rev 0e) 00:02.0 VGA compatible controller: Intel Corporation Atom Processor Z36xxx/Z37xxx Series Graphics & Display (rev 0e)

cpufreqd governor policy is set to performance in all profiles, and minfreq 1330000 maxfreq 2410000 in all rules.

The system may run a few hours without other problems than short peaks of throttling. They may last milliseconds or seconds. Any time I check the cpu temps (sensors), they stay <= 40 degrees. Not even close the threshold, set at 80. As long as the cpu throttling keeps within the limits (2.14GHz - 1.33GHz), things are okay. But eventually the cpu freqs get set below the limit, something like 667MHz. Then the clock gets way too slow, skewing like 30 secs a minute. Ntp or ntpdate are to no avail. Restarting or stopping cpufreqd doesn't help. Same goes for thermald or anything else I have come up with. Removing cpufrequtils or thermald doesn't help, nor does disabling intel_pstate at boot. Same problem occurs with 4.2. kernel. What the heck in kernel is dropping the cpu frequensies?

When the cpu frequensies are dropped below the limits they will never recover to normal, even though according to sensors the temps are at, say 37 or 38 degrees. Only cure is to reboot, which of course is no solution. Any ideas?

Here's how the system after 11 hours uptime starts skewing some 40s a minute:

Apr  5 07:05:07 kernel: [41878.651689] CPU0: Core temperature above threshold, cpu clock throttled (total events = 18)
Apr  5 07:05:07 kernel: [41878.651982] mce_notify_irq: 2 callbacks suppressed
Apr  5 07:05:07 kernel: [41878.651986] mce: [Hardware Error]: Machine check events logged
Apr  5 07:05:07 kernel: [41878.652176] CPU1: Core temperature above threshold, cpu clock throttled (total events = 18)
Apr  5 07:05:07 kernel: [41878.652429] mce: [Hardware Error]: Machine check events logged
Apr  5 07:05:08 mcelog: Family 6 Model 37 CPU: only decoding architectural errors

cpufreq-info

cpufrequtils 008: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq@vger.kernel.org, please.
analyzing CPU 0:
driver: intel_pstate
CPUs which run at the same hardware frequency: 0
CPUs which need to have their frequency coordinated by software: 0
maximum transition latency: 0.97 ms.
hardware limits: 1.33 GHz - 2.58 GHz
available cpufreq governors: performance, powersave
current policy: frequency should be within 1.33 GHz and 2.41 GHz.
           The governor "performance" may decide which speed to use                  within this range.
current CPU frequency is 667 MHz.

analyzing CPU 1:
driver: intel_pstate
CPUs which run at the same hardware frequency: 1
CPUs which need to have their frequency coordinated by software: 1
maximum transition latency: 0.97 ms.
hardware limits: 1.33 GHz - 2.58 GHz
available cpufreq governors: performance, powersave
current policy: frequency should be within 1.33 GHz and 2.41 GHz.
              The governor "performance" may decide which speed to use
              within this range.
current CPU frequency is 667 MHz.

sensors

acpitz-virtual-0
Adapter: Virtual device
temp1:        +26.8°C  (crit = +90.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +36.0°C  (high = +105.0°C, crit = +105.0°C)
Core 1:       +36.0°C  (high = +105.0°C, crit = +105.0°C)
0

Okay, seems some kernel versions have issues with my setup. Pulling a newish (09-Jun-2015) trusty specific kernel (3.14) from http://kernel.ubuntu.com/~kernel-ppa/mainline/ has resolved the problem.

Cpu scaling is off, the cpus are running at a constant 2.929GHz frequency. The system has decided by itself at the kernel installation to set the cpu speed like that! (official max frequency 2.41GHz) Sensors still lying low, clock keeping time on the second.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.