The incredible spontaneously crashing graphics drivers
April 23, 2013 12:02 PM   Subscribe

A buddy of mine has been having some utterly bizarre issues with his upgraded build: Gaming consistently causes his video card to crash. This problem is not present in any other use case—other GPU-intensive tasks such as BTC mining does not trigger instability. Isolating and stress-testing individual components fails to identify the culprit. Card itself continues to run fine in the old motherboard/CPU setup. Rolling back drivers, undervolting, etc. makes no difference. Help?

Build: ASRock Z77 Extreme4 motherboard / Intel Core i5-3570K CPU / Visiontek 2GB ATI HD7850 video card. PSU is rated for 520W Antec Eco.

In his words: "Periodically (once every few hours), my graphics card driver will crash, but only if I'm playing games. Most of the time it comes back up, but sometimes it causes the entire computer to BSOD.

These issues started happening when i upgraded my old mobo/phenom4 to the new mobo/cpu combination. This led me to think there was something wrong with the software, so I formatted the main hdd, re-installed windows 8 (instead of win7), same problems. Eventually, I exchanged the mobo/cpu for a new set, and still have the same problem. I've kept my Case/RAM/HDD/Peripherals from the 'previous' computer.

I've tried every combination of booting into safe mode / using tools to un-reinstall graphics drivers/etc, and it keeps happening. I've tried overvolting, undervolting, overclocking, underclocking, nothing seems to make any effect.

A few oddities: I've tried overclocking the GPU and it will run for hours if its just doing computations, but as soon as I try to play games, it will exhibit this behavior. Temperatures are very reasonable, <50 even under high load for CPU, <80 under max load for GPU."


There's a sneaking suspicion right now that there might be a driver conflict, but we're otherwise at a loss as to what could be going on. Any leads would be much appreciated.
posted by pinsomniac to Computers & Internet (6 answers total) 1 user marked this as a favorite
 
I wonder if it could be a power issue? The 520W power supply is only slightly over the 450 required by the card, and would fit with it starting when you moved up to a more power-intensive CPU, and is still consistant with you having problems when its intensive CPU + Intensive Graphics card issues but not only intensive GPU draw. Might be worth swapping in a 750W if you can borrow one to try it out.
posted by McSwaggers at 12:09 PM on April 23, 2013


Is he using the Visiontek Drivers for his card or just the stock AMD ones?

AMD/ATI cards have always had weird flaky driver issues, that's probably it more than heat / load if he can mine without problems.
posted by Oktober at 12:10 PM on April 23, 2013


Driver conflict sounds right. There's a lot more to games than just raw GPU processing. Has he updated every single driver he could possibly think of, no matter unrelated? ASRock should have downloads for all motherboard driver components, including onboard sound.

Anything specifically called out in the BSOD?
posted by dobi at 12:31 PM on April 23, 2013


Seconding power supply issue.
posted by dosterm at 1:20 PM on April 23, 2013


Almost definitely power supply. Is this the same power supply with both the "old" and "new" setups? Probably the "old" setup still works because the CPU/MB are drawing less power (or there are fewer peripherals like hard disks, etc).

Data point: I have a very similar system -- same motherboard, i5-2500k CPU, nVidia graphics, Antec Eco 620W PSU, three hard disks -- and I'm pretty sure I'm close to the max for my PSU. If I try to overclock my GPU by more than a little bit I get driver crashes.
posted by neckro23 at 6:41 PM on April 23, 2013


Response by poster: Thanks for all the responses! I can link to the BSOD crash reports/dumps if there's interest (wasn't sure if it was allowed to post external file downloads). His old Phenom setup is supposedly rated higher in power consumption than the new i5, but he was also operating on a possible miscalculation that said 350w is sufficient.

In his words: "The power theory seems good so far. I'm saving power by disabling a core or two and it seems to stop the problems. Will buy a bigger (700W~) PSU shortly and put that in to see if it's a permanent solution."

Will report back.
posted by pinsomniac at 4:30 PM on April 25, 2013


« Older Toys and objects to deal with physical stress.   |   Why is my MBP running ML appending to the... Newer »
This thread is closed to new comments.