Have I fried my 580?

[ArrowMk84]

Ok, I'm getting the "Display driver nvlddmkm stopped responding and has successfully recovered" error. I've gotten this with several driver versions in two different machines (I'm using the same two 580s in my new machine that were in my old one). Now, in my old computer, the GPUs routinely hit 97c. In the new machine, the GPUs hit up to 79c and 85c, and I think it's the one that's hitting 86c is the problem. Also, before the driver goes and kicks the bucket, I usually see some sort of artifacting, usually quick flashes of geometry that's rendered completely black/textureless (ranging from a small part of the screen to a significant part of the screen).

Some days it better than others (but tonight has been particularly bad - four times software, with two hard reboots). I doubt its any of the other hardware causing this, as I ran memtest86 for 12 hours, hit my CPU with various Intel Burn and Prime 95 tests all last weekend for 2 to 8 hours at a time, and stress tested my PSU using OCCT for half hour with no issues, while tonight the video driver dies after maybe five minutes.

Edit: BTW, it seems that DX11 is worse on the cards than DX9 (duh...), since I was trying to play Crysis 2 with the DX11 patch and ultra textures, and then used Uniengine Heaven for some GPU stress testing. BFBC2 and ME2 will occasionally show artifacts, but rarely go far enough the driver to stop.

[bob saget]

wow that sucks. can you RMA?

[ArrowMk84]

Yeah, they're EVGA's with life-time warranties. I'm hoping that someone has some other ideas to try first (but I'm pretty certain they're cooked - or at least the one that's currently hitting 85c is - the old case also cooked a pair of GTX285s).

What is the nominal load temperature for the 580's anyway?

[Rummy]

Do you run any kind of fan control? Afterburner? Im sure 97c is ok for the card even 85c but why would you let it get that hot? I have 2 580's in SLI on an Asus P8P67 pro the 266.58 drivers are supposed to be the best for now.Have you tried the cards separately? I would try them separately to see if there are any issues and clean install of 266.58 drivers. Did you call EVGA tech support? They are there 24/7.

[john19055]

They are supposed to be able to take 97c and be fine according to nvidia.Like was said do you use afterburner so you can set the fan speed .Running games like crysis 2 DX11 with max settings and I see you have a 30" dell and if you run at 2500 x1600 you are more then likely need the fan at 100%.Plus make sure you are not overclocking you GTX 580 unless you can overclock it without haveing to turn up the voltage.Plus try different drivers to see which one runs the best,I have use some drivers and got the "Display driver nvlddmkm stopped responding

[ArrowMk84]

With my old rig I had the nvidia drivers set up with a fan profile that sped up the fan to 100% under heavy load, and the cards will still get into the mid 90s (97c being the highest I ever saw, and I saw it frequently). I was hoping to avoid that with my new case. Anyway, I did the same thing (using EVGA Precision this time) with the fans running 70% when the temp gets to the 70c range (60% for 60c, 80% for 80c, and 100% for 90c), and the cards both stayed in the mid-70s C runnig the Uniengine Heaven demo for two hours straight with no issue. At least the fans aren't 100% any more.

I'll have to do some more testing with Crysis 2 and other games to confirm, but it looks if the cards stay below 80c then they'll function properly. That also indicates to me that I probably did some damage to the cards with my old setup (I love it: the new case has half the fans but far better airflow - CM Stackers suck).

[Rummy]

Yeah that FT02 is a beast, nice case. Glad your temps are under control. Keep us updated.

[ArrowMk84]

So far so good. After several hours of BFBC2, Crysis 2 and Mass Effect 2, I've had no further issues.

[Dazz]

At 90C the drivers throttle clock speeds typically when hot. Increase the fan speeds and see if it's any more stable.