Diggin' out this thread, because it's summer again... and I'm really at the end of my wits. I'm still having very regular overheat shutdowns, currently, when room temperature is around 28 to 30 degrees, after maybe 15 minutes a shutdown occurs, although I'm not really using the CPU a lot.
Stuff I have tried:
Got a new case
Got some Fans (two in, one out)
Tried to make the airflow as good as possible (tough, because the large graphics card is in the way, but it should be ok now. At least system temperature never rises anywhere much over 30 degrees, i.e. almost room temperature!)
Got better cooling paste (Bluefrost, conductivity >2.7W/m-k, resistance < 0.094 C-in^2/W)
Got a bigger heatsink and a better fan (not that much bigger and not that much better, it's hard to find a good AMD fan in bosnia)
Underclocked the CPU
Symptoms: CPU temp is alright in bios, idles somewhere around 35 degree C, if heated up falls reasonably fast down to 50 C, then much slower to 40, and veeeeery slow below.
However, after windows started up, it usually is at slightly over 80 C, if it was already warmed up a bit before this can go to 90. AND THE TEMPERATURE DOESN'T GO DOWN AGAIN! I.e. although the CPU cools nicely in Bios, it doesn't do so in windows. It stays, and it even rises slowly. Just now my computer shut down after starting up and doing absolutely NOTHING for about half an hour (I'm writing on my laptop, this was kind of an experiment). Interesting is that the CPU temperature doesn't go above 90C before the shutdown, else my motherboard would warn me. System temperature seems to stay constant at about 30 C all the time (and the major airflow goes right through the heatsink of my CPU).
Specs:
AMD 2.6Ghz DualCore (clocked down to 2.2 Ghz)
Gigabyte GA-M56S-S3 Motherboard
NVidia XT-1900 Graphics card
Enermax EG 365AX-VE Power Supply (358 Watts... I heard that a weak power suply can lead to overheating, that's why I'm posting this here. No Idea wheather that is a lot or not for my configuration).
---------- Post added at 12:13 PM ---------- Previous post was at 11:51 AM ----------
Try (with caution) what I call "manual detection". If the CPU is really at 90°C, then you should not even be able to touch it (or you'll get a bad burn). Also there should be a distinct smell of overheated plastic/metals. If you find that the temp seems alright, maybe the temperature probe is failing.
Else I had the same issue with my GPU, same temps and very slow going down too (what I call "thermal inertia"). The inside of the fan was full with "computer mud" (that gray, fine dust saturated with humidity), which rendered it completely inoperative. A good cleaning (by blowing air through it with my breath) solved the problem entirely. I wasn't aware there could be so much dust in a small thing like that. It's a matter of density, the fine dust agregates with humidity.
(copied over from the Power suply thread, where I wrongly posted initially. Sorry for the inconvieninace).
I can touch the cpu. It's hot alright, but I can touch it. Also, the bottom part of the sink is usually about the same heat. however, in my experiance 90 is touchable for a short time without any burning marks (we used to work at those temperatures with the firefighters...) There's no smell whatsoever.
What would I do in case of a bad temperature probe?
I clean my fans pretty regularly. The GPU doesn't get very hot (it's not really used when working on the desktop anyways). And yes, cleaning the Fan and sink of the graphics card was quite ugly, since I got it used.