Now I’ve seen that too…

While browsing my photos I suddenly got this nice little message in my Eterm:

Message from syslogd@futtelifut at Mon Oct 17 22:52:10 2005 ...
futtelifut kernel: CPU1: Temperature above threshold

Message from syslogd@futtelifut at Mon Oct 17 22:52:10 2005 ...
futtelifut kernel: CPU0: Temperature above threshold

Message from syslogd@futtelifut at Mon Oct 17 22:52:10 2005 ...
futtelifut kernel: CPU0: Running in modulated clock mode

Message from syslogd@futtelifut at Mon Oct 17 22:52:10 2005 ...
futtelifut kernel: CPU1: Running in modulated clock mode

(My Pentium IV is seen as two processors because of hyperthreading.)

Googling a bit revealed that “modulated clock mode” means that my system currently runs at 50% for up to a millisecond. After that the temperature is supposed to have dropped to a safe level again. I’m surprised if the temperature drops so fast that a millisecond is enough, but that’s what it says here :-)

6 Comments

  1. Andy:

    Me too. Also a P4 w/hyperthreading enabled (running the SMP image)
    I’m finding postings all over indicating lots of people running Redhat (Fedora)
    started getting these messages all of a sudden.

    I started seeing these messages when I updated the kernel on oct. 11 to
    2.6.12-1.1378_FC3smp.

    The messages come in bursts (to all xterms) and it does not happen constantly.
    I don’t know if the alarms are real or not, but I doubt it.

  2. Martin Geisler:

    I’m running Debian with a self-compiled kernel version 2.6.8 with SMP support, so it’s not something with Fedora. The strange thing is that I’ve never seen those errors before, and I’ve been running with this kernel for months… Oh well — so far no they have caused no harm :-)

  3. Andy:

    It turned out these temperature alarms were real in my case.
    The oddness of the report timestamps (not stable alarm pattern
    for stable conditions) might be due to the monitoring SW.

    Anyway, perform these checks and then see if the problem
    goes away.
    1) remove the PC case cover
    2) check for dust build up around the heatsink and fans
    3) clean out with PC air duster (PerfectDuster)
    4) power up the PC (w/case open) and make sure all the fans are working
    5) make sure all the PCI slot covers are installed and extra
    drive holes are installed to keep dust out of the case.

    The problem had nothing to do with the kernel update.
    Over time, the fins of the heatsink can get blocked
    enough to trigger the temperature alarm.

    Don’t ignore this alarm. Your Pentium4 doesn’t like
    running too hot :-)

  4. Martin Geisler:

    Thanks for the advice — I’ve just cleared out some dust inside the fan… I hope that it will make the alarms go away.

    In the future I want a system with no moving parts! It’s still somewhat expensive, but over the years I’ve already spend a fair amount on various “silent” coolers so I think it might be worth it.

  5. Zach:

    My question is what kind of photos where you looking at to cause the temp to rise?

  6. Martin Geisler:

    Just normal photos from my camera :-) I think the problem was that I rotated a bunch of them, and so the regeneration of the thumbnails together with me browsing along was too much.

    After cleaning my system I haven’t seen the errors since.

Leave a comment