My graphics card could boil an egg...

Nexus Zero

Idled. Well, hovering under at about 95C, but still, scary stuff. It's a GeForce 6800GS and I want to play Oblivion, but doing so shoots the temp up to about 120C. As far as I can gather, these are insane temperatures for a GPU to run at, but if I hadn't kept an eye on the temperature reader I'd be none-the-wiser as it's not underperforming, or showing any artifacting.

I've had to clean it once before, but without the screwdrivers needed to take the heatsink cover off for this week. I checked, cleaned what I could, and it didn't seem to be that bad. I thought it could be the power supply lacking the power for the fan; I don't know how big mine is but I know I'm pushing it with two HDDs, two DVDs and the card. I tried taking power from the two DVD drives and it stayed the same temperature.

Does anyone know what else could be wrong?

Edit: Oh, I have also tried running the PC with the case side off but that doesn't help much.

The Cat

Well, it won't on its own so much, you need moving air as well. desk-sized fan? After that, I'm lost >.>

Deusfaux

you need to take off the heatsink and reapply thermal paste (AS5)... even if that's not the issue

ViolentChemistry

Are you sure it's Celcius? Actually, are you sure of your measuring device in the first place? If the thing's running hot enough to boil water at idle, but doesn't appear to be having any performance issues, something's wierd. Particularly since that card apparently isn't supposed to get much above 65C at full-throttle.

Lodbrok

Yeah, i would not trust that reading at all. Usually you do not have to bother about overheating unless you actually experience artifacts, electronics are likely to lock up and stop functioning long before they cause any permament damage to themselves as long as there is a heatsink attached at all.

With that said, reapplying thermal paste certainly can not hurt, but it is probably a bad sensor.

ViolentChemistry

If it's really running at 120C under load any dust at all is going to turn into a fire.

Nexus Zero

Are you sure it's Celcius? Actually, are you sure of your measuring device in the first place?

Yes and no, respectively. The thermo is, I assume, whatever came onboard, but yeah I'm starting to question it with readouts like this and no actual symptoms.

you need to take off the heatsink and reapply thermal paste (AS5)... even if that's not the issue

I knew this day would come.

Edit: Good point, VC. I know there's some still in there, and there's been no funny smells recently.

Rook

My card started doing that 120degrees under load. When I opened up the top cover of the heatsink thing, there was just a solid wall of dust blocking the fan from pushing air over the heat sink. I just cleaned that out, stuck it back in and everything was fine again.

JWFokker

Nexus Zero wrote:

you need to take off the heatsink and reapply thermal paste (AS5)... even if that's not the issue

I knew this day would come.

Edit: Good point, VC. I know there's some still in there, and there's been no funny smells recently.

You have to reapply thermal paste any time you remove the heatsink and also every year or two as it will eventually dry out. There's no way around it. Without thermal paste you're asking for problems. That said, I'd recommend Artic Silver Ceramique because it's non-conductive. If you don't have much or any experience applying thermal paste, I would not recommend something like Arctic Silver 5 because if you use too much and it bridges exposed parts, you've likely just killed your GPU. And with all the exposed SMD parts around the 6800 GPU die, this is a definite possibility. Also, when applying paste, spread it evenly over the entire surface of the die, don't use the "single blob in the center" method or you'll get uneven distribution.