How to know what CPU temperature causes machine to stop?

Question

Since some recent changes, my machine suddenly powers off at random times. I suspect it's overheating. I've installed a temperature monitoring utility to watch how hot the CPU gets. Maybe I will be lucky and happen to glance at it sometime right before the next crash, but it's much more likely I'll forget, or it'll be covered by some major app's window.

Is there a way to know, without actually having the computer overheat and crash, what the threshold temperature is? Maybe it's hardcoded in some place deep, and is the same for all brands of laptops? Can it be altered?

I'm running Ubuntu 10.10 and am watching temperatures of my Intel T5550 with acpitail. The machine is a Gateway P6860FX which has been running fine since 2008. A more general answer than for this particular situation would be appreciated.

Answer 1

There is no magic temperature for all systems. In fact, it could be random and happenstance.

That said, take a look in your BIOS. There should be an option where you can select a temperature, which when reached, triggers an alarm (with the internal speaker) and starts throttling the CPU (the system runs at a reduced speed, usually selectable as a percentage of full speed), until the temperature comes back down to normal. In the same page, there should also be an option to select another temperature, which when reached, turns the system off.

Answer 2

Why not just make sure the fan is operating properly, and that the dust is cleared? Usually that's enough to prevent overheating (and these are the two most common reasons for overheating, assuming you're not overclocking, of course).