For the last few hours, my computer has been freezing at random every minute or so. I'll type some text, and it won't appear until ~15 seconds later, all in one block. It started happening all at once for no apparent reason. I didn't make any system changes (that I can think of). I'm using about 2% CPU and 10% RAM.

The only thing out of the ordinary is that yesterday, my fan started making a bad sounding noise, and it kind of switches between on (making a grinding sound) for 30 seconds, and off for 30 seconds. But my processor is still fairly cool (the fan used to blow cold air consistently, and I don't think the issues are related.

Crunchbang/Debian Linux, kernel 3.10-3-amd64


They are probably very related.


It can only take milliseconds for the CPU to go from "OK" to "OMG I melted". They should never be used with out proper cooling. To do so, will cause damage, and could cause a fire, injury, and all manor of bad things.

That warning aside Linux and most semi-modern computers have a kind of protection built in that "throttles" the CPU when temperatures get too high. That is what is likely causing your "stuttering".

  • That didn't occur to me. I'm sending it in for a warranty repair in a few days, but in the mean time I need to back up everything on my hard drive because "If your device does not have Windows 8 installed, it will be erased and restored to factory settings". So, if I severely underclock my CPU (maybe 1.0GHz), a) will it prevent overheating, and b) will it show up in any kind of low-level logs that could void my warranty? – Lucas Phillips Nov 16 '13 at 13:03
  • a. Nope b. maybe, but I doubt it. There just going to format your drive. You have backups right. – coteyr Nov 16 '13 at 20:15
  • CPUs pretty much always have good thermal protection: if they detect overheating, they'll throttle down to a lesser speed or lock up. Other processors on the motherboard or GPU may be less resistant, and may be damaged, but a fire is almost unheard of (you'd have to mess with the power supply for that). Underclocking is likely to prevent overheating if you manage to underclock all the overheating components (which may be more than the CPU). – Gilles 'SO- stop being evil' Nov 16 '13 at 21:36

I worked as a component-level electronic technician for a decade, i.e., I wasn't expected to determine which board is defective, I was expected to determine which part on the board is defective.

What you are describing is a failing power supply fan. Mechanical parts virtually always wear out way before electrical parts.

I learned the hard way never to replace a noisy power supply fan. (I'm talking about the grinding sound that suddenly appears, not a fan that is "slightly noisy" from day one because it's cheap.)

Because it's mechanical, the fan is the weakest part of the power supply. Once you replace that, the second-weakest is almost always the voltage regulator. Months, or even more than a year later, that suddenly goes wacko. The supply doesn't die -- it starts outputting erratic voltage levels. In one instance (years after I stopped working as a tech) it fried a memory stick and crashed the hard disk -- even though the hard disk had not been writing at the time. It took me 12 hours to determine what the damage was, get the necessary parts, replace the memory stick and restore the hard disk from a tape backup. (That was when computers were a lot slower.)


Both AMD and Intel CPUs have built-in thermal protections and will slow down if they detect overheating, usually before they start making wrong computations and definitely well before they start risking causing permanent damage. The freezes you're seeing are probably from such a slow down. The OS is notified in such an event, check your kernel logs (in /var/log/kern.log).

Other components on your system may be less well-protected and may be permanently damaged by overheating (though physical risks are extremely rare as long as you don't mess with the power supply). If some chip other than the CPU has slowed down or is misbehaving, this could also explain the freezes, as the CPU may be waiting longer than it expects for this other chip to respond.

Even if the air being blown is still cool, it may be vital to cooling some component.


Your best bet at this point is moving the HDD to a different computer, and backing up there. Sending the computer in for service with a different HDD would also be prudent, just in case.

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.