Pretty sure my graphics card is overheating

mere_immortal

Trying to play through Darksiders today and I'm pretty sure it's causing my GPU to overheat. It runs fine at about 45-60 FPS all the way through, but then suddenly my monitor will lose signal and act as if nothing is plugged in. The sound will continue for about 10 seconds then my PC will shut down and reboot about 5 second later.

This happened once before when playing Skyrim even on low settings, but it seems odd because the games run fine before restarting and it doesn't happen with other games.

Is it a GPU issue or something else? I've uninstalled CCC and drivers, cleaned the registries that had relevant files in and reinstalled CCC and new drivers but it's still doing the same. I've also manually set my fan speed to 100% and tried to run it but it's just the same too.

iRevert

Have you checked the event log?
What card in specific are we talking about?
What is the age of the card?

mere_immortal

Never used the event log so not sure what I'd be looking for.

XFX Radeon 4870, just under 4 years old.

AgentBryant

You could check the actual temperatures with something like GPU-Z or SpeedFan. I personally let a chart run in speedfan while playing a game or something, and periodically check it by minimizing every once in awhile. Alternatively, you can use OCCT to run a GPU stress test which will also log the temperatures.

All of these programs are extremely simple to use, and free, of course. OCCT can also test much more than the GPU, if needed.

mere_immortal

Ran GPU-Z while playing, managed to get a bit more time in than usual before the reboot, probably about an hour this time.

Temps were running about 85-90c for about 30 mins, but then went to 90-94c for the other 30 mins before the reboot.

iRevert

I think the best advice here is to bite the bullet and upgrade your graphics card.

Heck you could even use it as an excuse to do a complete new system...

But seriously though rather than futz with it I'd say if your not having issues outside of gaming (meaning the rest of the rig is stable) I'd check your wattage on your PSU and connecters and see what you could support without much of an overhaul.

mere_immortal

I have had an entire new build planned for a while, but my employers are extremely shitty about actually drawing up a new contract that was supposed to be coming several months ago.

Hopefully it will be here this month, but until then is it worth replacing the thermal paste on the card to see if that does anything?

Essee

Congratulations, I think your original guess was correct. Those don't sound like good temps, even under heavy load. Seeing as how you didn't mention anything... when was the last time you cleared the dust out of your computer? Picking up a can of compressed air and going at it can make a massive difference in temperatures. I think dust is probably the most likely cause for any overheating in a computer. It can't really hurt to clean things out and see if things improve before considering other options, at least. And it's not like compressed air cans are expensive. Plus you can use them again in the future.

iRevert

mere_immortal wrote: »

I have had an entire new build planned for a while, but my employers are extremely shitty about actually drawing up a new contract that was supposed to be coming several months ago.

Hopefully it will be here this month, but until then is it worth replacing the thermal paste on the card to see if that does anything?

Honestly if it were me I'd try to milk out the rig for another three months till Black Friday as your new build would get a wonderful budget boost with those deals.

I don't know if new thermal paste is going to fix it, its worth a try though with a good cleaning.

mere_immortal

Well I had done a clean out with compressed air a couple of weeks ago, but I took out the card this time.

The heatsink under the plastic shell looked like it was wearing the worlds most glorious dust beard. Gave it a thorough clean and now idle is no more than 52c and under load it's gone no higher than 72c.

Consider a lesson learned today.

WildEEP

Easy way to determine this - take off your case side and go get a box fan. They're like 10 bucks from Walmart.
Turn it on high and stick it right next to your case - blowing on the inside.

It'll be loud, but it'll tell you if heat is whats causing it to crap out.

corky842

WildEEP wrote: »

Easy way to determine this - take off your case side and go get a box fan. They're like 10 bucks from Walmart.
Turn it on high and stick it right next to your case - blowing on the inside.

It'll be loud, but it'll tell you if heat is whats causing it to crap out.

Not if the heatsink itself was clogged like in this particular issue. It might lower the temps a bit, but the GPU would still overheat if the heatsink wasn't getting enough air contact.

To the OP, glad you got it fixed. Just posting this here so it's not the very last thing someone sees when they have a similar issue and finds this thread.