BSOD's and Driver Crashes

[MikeIzEpic]

A bit lengthy of a post, but I'm in desperate need of help.

So yesterday I decided to give overclocking a shot on my EVGA GTX 980 SC ACX 2.0 GPU, after seeing Paul's Hardware do it. I started with a modest overclock, and ran Firestrike. No issues. Increased the clock. No issues. Increased again up to 80Mhz, crash. Decreased to 70Mhz and worked like a charm. I ran it a few times just to make sure it worked, and I had no issues.

I then booted up AC Unity, excited to see my 2 extra FPS I'd gain. It ran stable for about 10 minutes, and then the drivers crashed. I lowered the clock to 60Mhz, ran AC Unity, it ran stable again for about 10 minutes, then BSOD (Error kernel_security_check_failure).

Frustrated, I ditched the overclock all together. The game again ran for about 10 minutes, then BSOD (I forgot the error, but I know it wasn't kernel_security_check_failure). I then proceeded to reinstall the drivers, and try again. About 10 minutes in, the drivers failed.

This was last night, and after the last driver crash I gave up for the night. Come today, I contacted EVGA about the issue, and was told there could be 3 potential issues:

1. Power Issues
2. Corrupted Drivers / Software Issues
3. Issues with the GPU itself

I then went on a spree of troubleshooting, trying to get this sorted out. Going in order of potential issues:

1. I'm not 100% sure how to check voltages, the guy from EVGA said to look for the 12 Volt in the BIOS, which I managed to find, but there was no fluctuation what so ever in the 5 minutes I stared at it. No issue here maybe? I still consider this a possible cause.

2. I uninstalled the GPU entirely from my Device Manager, downloaded DDU and ran it in Safe Mode to ensure everything was wiped, booted back into Windows, and downloaded Nvidia Driver 362.00 (The newer ones are causing issues with multiple people as I'm sure you all know). The problem still persisted after doing all of this, so it can't be the drivers.

3. I don't have another system to test the card in, so I can't be positive if it is the card. This would be the easiest solution, but is not possible for me to do.

After doing all I could, I ran GPUTest ( The stress test with the giant donut looking thing ) at 2560x1080 ( My resolution ), in Fullscreen with AA on MSAA 8x. This ran for close to 5 hours with absolutely 0 problems. I also attempted to use OC Scanner from EVGA, but launching the sress test just made my screen go black with the overlay showing GPU usage and temperature. not sure what's up with that.

Feeling confident clean installing the drivers fixed it, I launched AC Unity, and within 10 minutes, my screen went black, it displayed the "No Signal" message, and something new happened. The LED's on both my KB/M went out. (Both are plugged into my MOBO's (Gigabyte 970A-DS3P) on board USB Ports if that means anything to the situation). This led me to thinking maybe it's the MOBO or PSU (Antec 750M HCG 80+ Bronze) causing the issue. But I realized, the system itself was still running. The lights in the system were still on, the fans were spinning, everything was working just fine. just the monitor was black, and the KB/M LED's were out. After about 20 seconds, it went back to the logging in screen and it was like the computer restarted, just it never lost power.

That got me thinking, the only game I had tested was AC Unity, maybe that was the issue? lol no. AC Syndicate ran for 10 minutes, then I got a BSOD (Error Memory_Management)

I don't know what the issue could be 100% at this point. If I had to guess, I'd say either the GPU, MOBO, or PSU, which limits it a bit, but doesn't make things too easy. Another thing I thought could be maybe the PC isn't getting enough power from the wall outlet, but I'm not an electrician, so I don't know. Is there any way I could test the parts without another system on hand? Any kind of software monitoring or stress tests or whatever would be great. I just don't know at this point. Any game I've tried playing crashes or BSOD within 10 minutes, but a full on stress test had perfect results in the 5 hours it was running. Any help would be appreciated.

List of System Specs:

FX-8350 | No OC
CM Hyper 212 EVO
Gigabyte 970A-DS3P
16GB DDR3
EVGA GTX 980 SC ACX 2.0
3 TB WD Blue + 120 GB Sandisk SSD Plus
Antec 750M HCG 80+ Bronze PSU

Please help!

[KezzaMcFezza]

Try another power supply. I know that those issues can be power supply related. It might be driver related try installing the Nvidia 364.72 is the death of Nvidia cards xD.

I know my answer might not help but hey it might just do.

 

Hope this helps

[daDILLYdog]

I also had problems with 364.72, I would get blank screens needing restarts and also crashes in games constantly, went back to the January release (Can't remember the damn number) and it went fine, I wouldn't know which is most stable past January.

 

I know AC Unity had a lot of flak but my 970 ran it perfectly when I played it in its entirety so it's not that damned games fault at least.

[Nord]

Firestrike is not a good way to check your GPU OC's (at least not until you found a stable one and wanna put it to a somewhat-more realistic testing szenario), the tests it takes are just so short that your GPU will never even remotely reach its max. temperature nor is firestrike a OC "test application", its a benchmark application.

If you know about CPU Overclocking, that would be using Cinebench(=3DMark in this methaphor) instead of Prime95 or Aida Stresstest to check if your OC is working.

If you OC your GPU use something like Unigine Haven or Valley or go oldschool with Furmark or something like that, those also makes it way more easy to OC btw.

 

I'd say it is fairly odd that your problems start exactly after you did an overclock, especially such a minor one.

Assuming you did use MSI Afterburner for your OC'ing and assuming further you do not know the programm alot, I'd suggest you boot it up again and make sure that your card is not overclocked anymore, no "apply OC on system bootup", no Power, Core or Memory increase, no change in the Power/Temp Prev tab (when you click the little button  next to power limit %) and no active profiles... (or just set your overlay to show GPU Core & Mem MHZ, if you know the base clocks on your card - and no, looking them up wont necesseraly help at least for the core because apperantly every card boosts differently high depending on, well luck and what game you currently play.)

Then I would try again, preferable with a game that is not assassins creed or like mentioned above some Unigine bench.

 

You could also try to actually de-clock your card aka go into the - minus values. -100 core & mem should do, just to see if this MIGHT fixes it. - reason behind that, I had an 660 ti factory OC out of the box which i had to declock else some games would crash, benchmakrs would work fine though. I know it doenst make much sense, but if it would actually fix the problem, at least you got something to go on.