I’m pretty sure my GPU is overheating, although I don’t know how hot it should be able to get without me worrying. Earlier today, I was playing a game (Titan Quest), when I noticed the graphics screwing up. Within seconds, I got a Blue Screen of Death. The screen had a bunch of 1" long black lines going through it in random spots. After restarting the computer, I checked the temp and it was in the mid 40s. I started up the game and after a bit of playing the temperature shot up to 66° Celsius. Now that I shut the game off, it has slowly went back down to 49°.
While I’ve read that GPUs can go up to 80° with no problem, it doesn’t really feel right for something to work at that temperature.
I just updated my drivers and when I find some dust-off, I’ll clean out the computer. So am I right to be worried about the 66° temp of my GPU? It definitely had artifacts when I Blue Screened and when I shut the game off the second time.
Oh, I’m on Win 7 64 bit, Intel i7 920 and a Nvidia GTX 260. Nothing on the computer has been overclocked.
Generally speaking, every 10 deg C above room temp cuts the expected life of a chip in half, but that’s just kind of a quick and dirty rule. In reality, the numbers don’t start going south until you get above about 45 deg C. Most computers tend to run a bit hot these days (they figure you’ll upgrade long before the reliability becomes an issue) so 50 to 55 or so isn’t uncommon. Ideally you’d like to get it as close to 45 as possible. Above 60 is definitely a problem.
“Up to 80 with no problem” is more like “up to 80 without frying instantly”. It’s still a problem if you ask me (and I design industrial computer stuff for a living).
I have a GeForce 6800 GS in this computer that tends to report itself at around 100 Celsius under load. I’m not sure if its thermostat is wholly accurate, but it hasn’t died yet, doesn’t freeze, etc. This entire computer is rather old, though; the whole works is probably getting replaced within a couple months, so I can play games with the latest gee-whiz graphics. But I’m a data point contrary to engineer_comp_geek, at least.
The 260 tends to idle at 40 and goes up to around 80 degrees. So your temps seem ok, if a bit high. Make sure there air flow in your case is adequate and flows well. Make sure wires are out of the way, and while you’re in there use a can of air to blow out any dust in there.
Make sure you’re running the latest drivers, and since you’re experiencing problems I’d suggest you do a clean install of the driver. So uninstall, boot into safe mode, run http://www.download3k.com/Install-Driver-Cleaner-Professional.html Driver cleaner (there are other apps too), then reboot and install the latest drivers.
Riva tuner also allows you to control fan speeds and monitor temps. Install that to keep an eye out on your card.
As far as I know, 66 is fine for a 260. Some GPUs can run very hot with no adverse effects… I had to use a borrowed 7950GT (single slot cooler) after my other card went in for warranty replacement and it ran at 105-110c … and this is in an antec 900 case with 7 fans and massive airflow. Totally nuts. But it worked… that was within its operating range.
Black band artifacting does tend to indicate an instability of some sort in the card, but not necesarily. Could be a driver issue or a corrupted game. I would recommend getting atitool (which isn’t limited to just working on ATI cards). It has a “scan for 3d artifacts” function that’s roughly the GPU equivelant of prime95’s torture test… it maxes out your GPU, renders stuff, and then checks for errors in the rendering. If you can pass the ATITool test for a few hours (a few minutes would probably tell you, but the longer you test, the more stable you know it is for sure) then it’s not a heat related problem with the game.
You can also get rivatuner as Kinthalis said or EVGA precision tools to adjust the fan speeds. But I would try the atitool thing first to confirm that you have a physical problem with the card.
Thanks guys. The suggestions helped a lot. I did some further research and, it seems like, the fan’s speed doesn’t fluctuate with the card’s temperature. But I bumped the fan speed to 80% and it seems to have solved the heat issue. I also tried 100% speed but the sound was unbearable.
Also, I guess atitool doesn’t really work with the newer Windows systems. GPUTool seems to be its replacement.
One last thing, I never had this issue when playing Crysis, so I wonder if it is an issue with the Titan Quest software and not really the card.
Rivatuner will allow you to use a fan response profile to the GPU temperature. See this guide. You’ll need to understand the typical operating heat range your card will experience and the minimum and maximum fan speeds you’re comfortable with running, but you can set it up so that it automatically scales up and down the fan speed based on temperature.
I set up my old card very precisely, but the new one doesn’t give me the option to - not sure if it’s the card or the new nvidia drivers or what. Not sure if this will work for your setup.
Not sure what planet you are from where those are the norm temperatures. 45 deg C is a low idle temperature for many computer components and they can easily go higher safely. CPU temp ranges vary by CPU. GPUs as well for that matter.
For the OP, per Nvidia itself, the acceptable max temp for the GTX 260 is 105 deg C.
Cooler is always better but that is what the card is engineered for so you should be fine with the temps you reported. However, there can be a big difference between the temperature in the case and the temperature of the chip itself. The chip may well be cooking well over the rated temperature warming the air in the case to the range you are talking about. Be sure you are measuring the actual GPU temperature.