0

I have spontaneous logouts, probably caused by overheating. I have the following lines in syslog:

Jul 23 13:44:19 studebaker kernel: [  491.025664] CPU8: Core temperature above threshold, cpu clock throttled (total events = 1)
Jul 23 13:44:19 studebaker kernel: [  491.025665] CPU2: Core temperature above threshold, cpu clock throttled (total events = 1)
Jul 23 13:44:19 studebaker kernel: [  491.025666] CPU4: Package temperature above threshold, cpu clock throttled (total events = 1)

How to see / change that threshold, mentioned in error message?

  • 1
    I thought that was a non-writable hardware setting – 炸鱼薯条德里克 Jul 23 '19 at 10:56
  • 1
    Then how to see it? An/or how to change reaction for it? – Dims Jul 23 '19 at 10:57
  • 2
    How about figuring out what causes overheating and make it not do that, alternatively checking fans and cooling? I would have thought overheating would be a symptom rather than a problem to fix in itself. – Kusalananda Jul 23 '19 at 11:37
  • What's your motherboard and CPU? Please do cat /proc/cpuinfo | grep "model name" && sudo lshw | grep -A5 "Mo" then click edit to add the facts to your question. Please do not click Add Comment, please use edit instead. – K7AAY Jul 23 '19 at 16:01
  • 1
    @Dims Normal work should never cause overheating. You may have cooling issues. – Kusalananda Jul 25 '19 at 7:23
6

You should be able to see the thresholds under /sys/devices/platform/coretemp.?/hwmon/hwmon?. The values replacing the question marks will depend on your system; the first is likely to be 0.

You’ll see a number of files named temp?_crit, temp?_crit_alarm, temp?_input, temp?_label, and temp?_max. The values replacing the question marks vary again, starting at 1, and increasing to cover temperatures for the package and all the (real) cores in your system. _label tells you what the set of values describes, _input shows the current temperature, _max the maximum temperature, _crit the critical temperature, and _crit_alarm indicates whether the critical temperature alarm has triggered.

You can’t change these using kernel-provided interfaces, they’re set by the CPU and/or system firmware. You shouldn’t try to change them anyway, they’re designed to protect the CPU from damage. If you’re bothered by the overheating messages you should try fixing the overheating problems instead — the ambient temperature might be too high, the CPU’s cooling could be hampered by dust, etc.

The messages in your kernel’s logs don’t reflect the kernel’s decisions; they’re the translation of machine events that the kernel is being told about. Your computer noticed that its CPU was getting too hot, so it throttled it, on its own, and told the kernel about it.

  • Each of these files contain 100000, are these millidegrees? :) I understand your common recomendations, but they are not applicable. – Dims Jul 25 '19 at 7:19

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.