Computer crashes on high computation kernel execution

Hi everyone,
I have GTX 690 in my machine, and whenever I run any high computation kernel such as 2DFFT (from the sample programs) the computer crashes and monitor goes blank. However for some kernel it executes normally like bilinear texture sample. Can someone please help me figure out the cause.
The day before this started to happen, I set computer to perform convolution of 1000s of images for more than 2 to 3 hours and templature of gpu was around 83~85º C. Please help me resolving this issue. Thanks

You can try to disable the Windows TDR (Timeout Detection & Recovery) and see if the problem is resolved. See:

https://devtalk.nvidia.com/default/topic/535264/cuda-programming-and-performance/kernel-runs-fine-on-osx-linux-crashes-on-windows-/post/3762516/#3762516

It’s also possible the high temperatures could have caused some intermittent hardware failures on your existing GPU, but try the disabling TDR solution first.

Thanks for your reply, I have tried this… but the problem is still there. Before this issue occured, GPU was easily handling every kernel without any problem. now, it crashes even for some simple applications. I think you are correct something must have gone wrong with the Hardware. One thing I am confused about is that if there is some problem with the hardware how come is the showing display and doing everything normally until I put it to some computational intense program.
Is it possible that its thermal paste got burnt and is not transfering heat properly?

Sounds like hardware failure in your video card. Try RMA’ing your card with the manufacturer if possible. Lesson learned, if you see temperatures that are higher than say 70 C, you should look into improving airflow in your case. I’ve had cards that fail only when at high loads/hot, yet they display video fine… it happens. What temps does it idle? If idle temps are high, it could need a re-paste. You don’t mention what temps you get when you run the code that fails, so I can’t advise you further. I’d say RMA it with the manufacturer…

If you wanted to make sure that the card is the problem, just switch it to another system and see if the behavior follows the card.

Maybe the power supply generates transients/unstable voltage at high load?
The part might have failed gradually.

Also check the 12 V cables/connectors to your video card. Maybe there is
some bad or semi-loose connection

Christian