GPU Temperature Question

Gamer Bhoy 89

At what temperature should I start to panic at? At the moment my GPU has peaked at 73 degrees Celsius when playing rather graphically-demanding games. I played Fallout New Vegas and it peaked at 69, whereas Skyrim has peaked at 72 (naturally).

My MSI Afterburner program automatically states that 100 degrees will summon a warning, but I've brought it down to 95 just incase.

But I keep uncomfortably looking up at my temperature while I'm gaming, worried about the temperature soaring up.

The reason is; my computer kept restarting because and giving me the Kernel 41 error, to which my latest finding seemed to have been my GPU being too hot (after spending a week trying to figure out the issue). I found two lines of thick dust on either side of the fan, like, carpet-thick. So I cleaned off all the dust and I downloaded Afterburner to monitor my temperature, but beforehand, I had no idea what temperature it was at when the pc restarted.

So, could someone put my mind at ease? What temperature is acceptable and what temperature should I worry?

For the record, I have a total of 5 fans not including the GPU inside the PC (2 on the side, one from CPU, one from PSU and one from rear) so I have plenty of cooling flowing through. So far I've managed to play Fallout New Vegas for a full hour and 10 mins without an issue since I cleaned out the GPU. (Before this, it (and some other games) didn't last any longer than about half an hour straight.)

Sorry for the long post.

nesf

It varies from GPU to GPU what temperature range is ok. AMD 6970s might not cut out until you hit 110 for instance (you wouldn't want it near this normally).

nesf

Actually, the easy way to see if the GPU is the problem is to run a stress test on it, get the temp up high and see if there's a crash. If there is, it could be system cooling or your PSU or the GPU or some other component near the GPU getting too hot because of the rise in surrounding temperature (unusual).

Error 41 doesn't mean much on it's own, other than something is wrong. If you *only* get it in graphically intense games then maybe it's the GPU, if it happens other times then it's probably something else.

Gamer Bhoy 89

nesf wrote: »

It varies from GPU to GPU what temperature range is ok. AMD 6970s might not cut out until you hit 110 for instance (you wouldn't want it near this normally).

Sorry, I didn't even think to call out what my GPU was ><

I'm running a GTS 250 512mb

Gamer Bhoy 89

nesf wrote: »

Actually, the easy way to see if the GPU is the problem is to run a stress test on it, get the temp up high and see if there's a crash. If there is, it could be system cooling or your PSU or the GPU or some other component near the GPU getting too hot because of the rise in surrounding temperature (unusual).

Error 41 doesn't mean much on it's own, other than something is wrong. If you *only* get it in graphically intense games then maybe it's the GPU, if it happens other times then it's probably something else.

Well this PC is brand new as of last Wednesday, custom built. It could only have been my GPU since it was the only thing I kept from my old PC. And it only cut off when I played a game that had high graphics. At no other point did it cut off, and I've been using this PC non stop since I got it.

nesf

Gamer Bhoy 89 wrote: »

Well this PC is brand new as of last Wednesday, custom built. It could only have been my GPU since it was the only thing I kept from my old PC. And it only cut off when I played a game that had high graphics. At no other point did it cut off, and I've been using this PC non stop since I got it.

Run a stress test, get the GPU hot and under heavy load and see if the system crashes. If it crashes, then, either the PSU isn't able to handle the load or there's something wrong with the GPU. Usually.

nesf

Gamer Bhoy 89 wrote: »

Sorry, I didn't even think to call out what my GPU was ><

I'm running a GTS 250 512mb

I've seen that GPU as listed as needing a 450W PSU (assuming other components are normal) from a quick google. Normally these requirements are overstated by quite a bit but if it's smaller or cheap PSU in the system it might be the cause of the issues.

Gamer Bhoy 89

nesf wrote: »

I've seen that GPU as listed as needing a 450W PSU (assuming other components are normal) from a quick google. Normally these requirements are overstated by quite a bit but if it's smaller or cheap PSU in the system it might be the cause of the issues.

Mines is a 500W

nesf

Gamer Bhoy 89 wrote: »

Mines is a 500W

That should be loads. Assuming it can actually put out 500W of clean power.

Gamer Bhoy 89

nesf wrote: »

That should be loads. Assuming it can actually put out 500W of clean power.

Well so far I've had no problems anyway. But I'm really reluctant to stress test the thing. I don't want to risk bricking it since I'd go up to a temperature that a game wouldn't normally reach

nesf

Gamer Bhoy 89 wrote: »

Well so far I've had no problems anyway. But I'm really reluctant to stress test the thing. I don't want to risk bricking it since I'd go up to a temperature that a game wouldn't normally reach

No worries, but you can't brick a PC in any reasonable meaning of that term as its used for phones or tablets.

Kirby

Those stress tests are artificial and can absolutely damage your card. I remember the millions of euro worth of cards killed by furmark a while back. :P

Run the most demanding game you have for a while and see how high the temps get. You mentioned Fallout New Vegas. That has ambient occlusion which is a resource hog so that will definitely stress your card and give you an idea about what your game temps will be during a long session.

Most new cards have excellent cooling and wont hit over 70 degrees even under heavy load. Yours is an older model with mediocre cooling but 72 degrees is nothing to worry about. At all.

I ran two cards in SLI and they would regularly hit 85 degrees for several hours at a time. They worked perfectly for years. I wouldn't worry.

danthefan

72 doesn't sound overly high for a GPU.

Gamer Bhoy 89

Kirby wrote: »

Those stress tests are artificial and can absolutely damage your card. I remember the millions of euro worth of cards killed by furmark a while back. :P

Run the most demanding game you have for a while and see how high the temps get. You mentioned Fallout New Vegas. That has ambient occlusion which is a resource hog so that will definitely stress your card and give you an idea about what your game temps will be during a long session.

Most new cards have excellent cooling and wont hit over 70 degrees even under heavy load. Yours is an older model with mediocre cooling but 72 degrees is nothing to worry about. At all.

I ran two cards in SLI and they would regularly hit 85 degrees for several hours at a time. They worked perfectly for years. I wouldn't worry.

Yeah I didn't do a stress test in the end, I did hear many disadvantages to it. No intent on dismissing any advice but I like to do mass research before I make a final decision.

I got Shogun 2 and tried it out and managed to play 3 battles without an interruption. Temperature balanced out between 64 and 69. I'm more at ease now, knowing that the temperatures I'm hitting is fine.

Thanks lads

Moon54

I'd agree, 72C is not that high, and probably nothing to worry about.

Don't forget, the recent good weather may have likely raised the ambient temperature in your room as well.

And maybe check if your PSU has one of those anti-dust mesh filter.
Mine gets clogged up after only a few weeks and when I clean it my system temps drops a couple of degrees too.

TheChrisD

I remember my old GTX275 used to idle at around 50-55 (at around 40-50% fan) and under stress would hit 75-80 before promptly giving out and just deciding "fúck it, I'ma stop working".

Since I upgraded to my 660, it's been idling around 45 and under stress barely breaks 55; with the fan still barely used.