Solved Random crashes while playing any 3D game, no blue screen, just restart

[RetroGamer]

I'm having an odd issue when I try to run and 3D game on my PC, after about 5-10 minutes the image freezes up for a few seconds (Audio still works), then the screen goes to black (About a second worth of the last bit of audio played loops), then the computer simply restarts. There is no blue screen, and when windows starts back up, it does not give me the usual; "There was a problem and windows had to shut down" dialogue. However, upon reboot, I am asked whether to start Windows in safe mode or normally. The temperatures of the hardware are all at normal, so it doesn't appear to be overheating. I feel that the power supply may be suspect, but from what I've read, 750w-800w is enough for my configuration. I tried with a 1Kw PC Power and Cooling PSU I had lying around, and the same issue occurs. This problem does not present itself when I have run 3D games in Ubuntu on this system, so I would assume it is some issue with Windows 7. My chipset and video drivers are up to date. Any help is appreciated.

 

[ShamrockRig]

I dont think its the PSU, what are your temps sitting at?

 

[RetroGamer]

Three of the GPU's are hitting a max of 48 degrees Celsius while in game, but one is hitting 65. My CPU is running around 45. In other builds I've gamed with temps as high as 90 without issue, hence why I don't think it's that. Though the one GPU at 65 may be causing it. I also have ramped up all the fan speeds to 100% to cool them better, but the problem still occurs.
EDIT: Upon disabling SLI, the problem seems to disappear, could this be the issue? And if so, what is causing it? (The temps are the same too, only one gpu at 65, the rest at 48.)

 

[ShamrockRig]

90 degrees celsius does seem quite high buddy, it could be a problem with one of the cards, try giving the computer a hoover out as dust can be a major factor in overheating

 

[Erick Aguilar]

What you are describing is a bluescreen. Odds are the system decided to change to option to display it and the option to automatically restart is ticked.

Check your dump files and see if there are any newly created ones.

 

[RetroGamer]

90 degrees celsius does seem quite high buddy, it could be a problem with one of the cards, try giving the computer a hoover out as dust can be a major factor in overheating

90 degrees is what I have run at under load on other builds, not this particular one though. I have not had any GPU get above 65 in this situation.

What you are describing is a bluescreen. Odds are the system decided to change to option to display it and the option to automatically restart is ticked.

Check your dump files and see if there are any newly created ones.

I have had other crashes for other reasons, and the blue screen has displayed. No dump file was created from crashes relating to this issue. There are only ones from other blue screens.

I believe it is SLI that is causing the problem, but I am not sure why, it shouldn't crash in this way if it were a driver issue.

 

[badger906]

when playing games try turning off AA, i know it sounds stupid as a triple SLI system can handle pretty much anything at any setting, but there was a known issue with 660ti's in SLI on certain games with AA turned to anything other than off.

 

[Viscon]

I dont think its the PSU, what are your temps sitting at?

Actually it might be PSU since it's 4x +12V rail design provides only 20 ampers each and I think this is what matters much more than wattage.
You may try some other PSU like Corsair HX1050 where single +12V rail provides 100A.
I know it's a high end and not cheap product but for your configuration there should be no hesitation in getting one like this.

 

[RetroGamer]

I dont think its the PSU, what are your temps sitting at?

Actually it might be PSU since it's 4x +12V rail design provides only 20 ampers each and I think this is what matters much more than wattage.
You may try some other PSU like Corsair HX1050 where single +12V rail provides 100A.
I know it's a high end and not cheap product but for your configuration there should be no hesitation in getting one like this.

Yeah, thought that might be the cause. Hoping it wasn't though, because single rail PSU's are expensive. And it would explain why disabling SLI fixes the issue, because the second GPU isn't drawing as much power.