Windows 7 Home Premium 64-bit SP1
Intel Core i7 920 @ 2.67GHz 40 °C
Bloomfield 45nm Technology
12.0GB Dual-Channel DDR3 @ 534MHz (8-8-8-20)
ASUSTeK Computer INC. P6X58D PREMIUM (LGA1366) 28 °C
Acer P223W (1680x1050@60Hz)
2047MB NVIDIA GeForce GTX 750 Ti (EVGA) 24 °C
596GB Western Digital WDC WD6401AALS-00J7B1 ATA Device (SATA) 43 °C
HL-DT-ST DVDRAM GH20NS15 ATA Device
I've been having some on-and-off problems with what I believe to be my GPU. I couldn't find anything about it on this site yet, though I did see some potential hits on other sites.
Basically what's been happening is at various points, almost always (though not universally) while I have a game open, one of a few things will happen:
- The screen will go black for about a second and then go a flat colour, becoming completely unresponsive with no sound. This requires a hard reboot to correct. It's NOT a BSOD. Screen is most typically white, beige, or rarely black.
- The screen will go black and the monitor will momentarily no longer receive a signal, before it recovers and the Nvidia icon shows a tooltip saying that the driver has crashed due to a kernel error (Nvidia windows kernel mode driver 378.66 stopped responding and was recovered successfully.). This was formerly the only crash, but now the others are more common.
- Similar to option one, the screen will go a blank colour, before recovering, with no error tooltip.
- Similar to option one, the screen will go a blank colour and become completely unresponsive, but will automatically reboot after about 10 seconds.
- Only once recently, a similar crash occurred an hour after Furmark had finished a two hour benchmark, while nothing major was open. The monitor went black for a moment, this time, and then came back with the pre-crash image, but screwed up with blue and red artifacts. The system was unresponsive and required a hard reboot.
These have been repeatable with most games, but irregularly, and not with all games. I have encountered this repeatedly with Total War: Attila, For Honor, and even 80 Days. Previously, during the last bout of dealing with this, Witcher 3, Company of Heroes 2, Civilization 5, and a number of other games were also affected. However, so far, I have been able to play games like Civilization 3 with impunity.
As mentioned, I haven't been able to repeat it consistently. I've been able to go an entire day without a single crash even while playing Attila or For Honour, while I've had other days where it crashes repeatedly. Typically I can get away with playing for hours at the start of the day before the first crash occurs, but occasionally it starts right off the bat when a game is opened. Also as mentioned, it occurred once with no games or major programs open, though about an hour after a two-hour Furmark test.
From what I recall, these problems only started occurring when I installed the driver update that came out around when Witcher 3 came out, that was updated for that game. It started immediately after that update. I tried searching for solutions back then, but doing a clean roll back to 353 was the only thing I could find that worked, though only with additional tweaks. That worked for a long period (3-6 months), but then it started happening again. After dusting out the computer, unseating and then reseating the GPU, it began working again (for several months again, with occasional hiccups), though lately it's started back up again heavily. Dusting and unseating/reseating had little effect. I was still using the rolled back drivers. I since tried updating them again to try to fix it, but no dice.
Things I've tried to fix/diagnose it:
1. Doing a clean install of my graphics drivers.
2. Doing a clean rollback with DDU to a suggested driver (353) from before the problems began.
3. Updating graphics driver to the newest version, and then rolling back again. Neither worked.
4. Ran GPU-Z and found no apparent issues.
5. Ran Furmark for two hours, with no problems. Crash only occurred an hour after I closed Furmark.
6. Unseated and reseated GPU.
7. Did a thorough dusting, checked cables.
There are no indications this is due to overheating unless it's the PSU, GPU-Z says the GPU gets more than sufficient power, and there are no detected conflicting driver files for graphics cards that I can find.
Any help is greatly appreciated!
Edited by Naomeister, 15 February 2017 - 06:00 AM.