My 980TI GPU keeps crashing...but only on my computer? Read on and be mystified...

[gozulin]

***RESOLVED, LOOK AT THE BOTTOM OF THIS POST BEFORE ANSWERING***

So I bought an evga 980TI watercooled on Windows 10 64bit and I would routinely get a black screen with windows event viewer logging the dreaded "Display driver nvlddmkm stopped responding and has successfully recovered.". Windows is still running and I can remote desktop into it just fine. This is definitely limited to the GPU/GPU drivers.

I replaced the original GPU with another and performed all kinds of diagnostics with both evga and nvidia (turning off powersaving modes, DDU, latest drivers, latest "stable" drivers). EVGA sent me a replacement card but I've had the same issue with it...
The card is not overheating (not by a long shot! it's usually around 45 degrees). The crashes don't happen when it's under load. (stress tests of 1h+ would complete just fine). Sometimes the screen goes black while i'm not using the computer for anything. It's just sitting at the desktop and crashes. other times I'm just watching a 720p video, barely using any resources, and it crashes then.

Basically, no correlation between usage and crashes as far as I could tell.
After several months of this, I swapped GPUs with a friend's AMD 280x and I've had 0 crashes in 3 weeks. I was hoping to get crashes so I could blame my mobo or PSU... (I use an asus pro gaming z170l and a silverstone strider gold 650W, which are both supposed to be quality components, recommended by anandtech forum goers).

The incomprehensible thing is that he ALSO didn't have ANY crashes with my GPU. I was fully expecting one of us to get crashes, but nope! since the exchange, nothing. He's running windows 10 64bit too...

So, I don't understand what is going on. I'm a computer engineer and I'm ashamed of being completely stumped! Can any of you think of anything that could explain what I've observed?

If anyone can figure this out, I will vote for them in the King's moot and serve them loyally for as long as my house stands.

***UPDATE***

After I got my GPU back from my friend's computer, I have not had the issue crop up again. I suspect an OS or driver update fixed the symptom. It definitely wasn't my power supply or RAM as some speculated.

 

[gradoman]

Maybe if you have a spare drive, install Win10 on it and try..? If you've got anything overclocked, return it all to stock.

 

[gozulin]

Maybe if you have a spare drive, install Win10 on it and try..? If you've got anything overclocked, return it all to stock.

Already running win10. Nothing is overclocked.

 

[Snarf Snarf]

I think he meant do a clean install on a different hard drive in case there are any bad sectors on the current drive? I had similar issues for a while that I thought were unstable overclock related, but a clean install of Win 10 on a different drive solved the issue.

 

[boozzer]

Already running win10. Nothing is overclocked.

since your friend is not getting any crashes = software problem, I would do as grado stated first.

 

[dark zero]

Mmm... drivers... nVidia ones are crazy buggy last months.... also. Which MoBo are you using?

Haswell to below are giving some problems on Windows 10

 

[Elixer]

Basically, no correlation between usage and crashes as far as I could tell.
After several months of this, I swapped GPUs with a friend's AMD 280x and I've had 0 crashes in 3 weeks. I was hoping to get crashes so I could blame my mobo or PSU... (I use an asus pro gaming z170l and a silverstone strider gold 650W, which are both supposed to be quality components, recommended by anandtech forum goers).

The incomprehensible thing is that he ALSO didn't have ANY crashes with my GPU. I was fully expecting one of us to get crashes, but nope! since the exchange, nothing. He's running windows 10 64bit too...

So, I don't understand what is going on. I'm a computer engineer and I'm ashamed of being completely stumped! Can any of you think of anything that could explain what I've observed?

If a clean install doesn't fix this, then I would fire up OCCT & furmark, and keep an eye on the voltages. Even good PSUs can go bad.

 

[Osiris-TEP]

Are you running any antivirus software -- ESET NOD32, etc?

 

[Despoiler]

New mobo BIOS? One just came out on 5/25 that "Improve system stability".

I would do a clean install of Win10. Could be also be a program that is installed that is causing the instability.

 

[deanx0r]

are you using any kind of monitoring program like MSI afterburner or evga precision? Those two gave display drivers crashes when I was using specific features or with some applications.

 

[Val_]

since your friend is not getting any crashes = software problem, I would do as grado stated first.

Can we be really sure it's a software problem ? Couldn't it be a completely non-obvious hardware incompatibility between the 980 Ti and for example his mainboard, something which does not occur with the 280X and with 980 Ti in the other system ? It's really a question, I have no idea if this kind of thing may happen

 

[plopke]

Maby you can borrow your PSU of your friend? Still can be your card is very very touchy about any power fluctuations.

 

[gozulin]

I think he meant do a clean install on a different hard drive in case there are any bad sectors on the current drive? I had similar issues for a while that I thought were unstable overclock related, but a clean install of Win 10 on a different drive solved the issue.

since your friend is not getting any crashes = software problem, I would do as grado stated first.

I will do that, although it's a huge PITA. I was actually thinking of doing a clean install of win7, but maybe it is too old at this point...

Mmm... drivers... nVidia ones are crazy buggy last months.... also. Which MoBo are you using?

Haswell to below are giving some problems on Windows 10

Mentioned it in the post. Asus pro gaming z170l. Using skylake 6600k.

If a clean install doesn't fix this, then I would fire up OCCT & furmark, and keep an eye on the voltages. Even good PSUs can go bad.

Was not aware of OCCT. Thanks for the suggestion. I will try that.

are you using any kind of monitoring program like MSI afterburner or evga precision? Those two gave display drivers crashes when I was using specific features or with some applications.

Nope. Those were the first things to go.

Maby you can borrow your PSU of your friend? Still can be your card is very very touchy about any power fluctuations.

I can't. I have a microATX tower and the GPU is too long to put in a regular PSU (the silverstone strider PSU I have is shorter and more expensive because of that).

Are you running any antivirus software -- ESET NOD32, etc?

Not in the last 15 years. No. For the exact reason you imply.

New mobo BIOS? One just came out on 5/25 that "Improve system stability".

I have the 2/16 Bios. I'll get the newest one now. Thanks for the tip.

 

[tential]

Have you tried a different cable?

 

[Elixer]

I can't. I have a microATX tower and the GPU is too long to put in a regular PSU (the silverstone strider PSU I have is shorter and more expensive because of that).

You don't have to mount everything in the case, the PSU can be off to the side if you want.
This is just for testing purposes after all.

 

[VulgarDisplay]

Maby you can borrow your PSU of your friend? Still can be your card is very very touchy about any power fluctuations.

Had to get to page two to see this. My guess is your power supply isn't working right.

Sent from my XT1575 using Tapatalk

 

[gozulin]

Have you tried a different cable?

You don't have to mount everything in the case, the PSU can be off to the side if you want.
This is just for testing purposes after all.

Had to get to page two to see this. My guess is your power supply isn't working right.

It could definitely be the PSU. The PSU is obviously on my short list of suspects.

With that said, the PSU is working fine with the 280x and the GPU failures seem completely orthogonal to GPU load (gpu load = increased need for power i'm assuming) so there is no additional evidence that points to it.

I mean, obviously, I could just start replacing the remaining parts component by component (PSU, Ram, Mobo) like I did with the GPU until the issue goes away but I don't have spare parts and the exchange period for my components is over so I'd have to buy replacement parts with no assurance they're at fault. Sounds like a nightmare and exactly what I'm trying to avoid by posting here

That'll be my last resort I guess. If I swap another part, it'll definitely be the PSU.

 

[railven]

Similar story, not the same GPU of course.

GF's rig would always crash running one game, ArcheAge, but every other game she played showed absolutely no issue.
Everything I tried, no solution.
New GPU, new CPU, fresh install of OS, new SSD, fresh drivers, etc, the works.

Finally swapped out her PSU (which showed fine through stress tests, would load CPU 100% with Prime95 and then run a game @ 4K DSR getting 3-10 FPS but GPU load was 100%) with my spare one.

Never had an issue again.

No idea why her other PSU (Corsair GS600) would cause a system crash with only one game, but changing it for my ancient Corsait TX750 fixed it.

 

[Leopardos]

I had the same issue for 1 WHOLE month with new R9 380/GTX 570 with Windows 10...

after 1 month trying this and that ... I fixed it by updating Windows 10, and installing the card on other PCI slot..
i think each PCI slot act differently than the other or the one that the card was installed on was defective...

try installing the card on other pci slot and report back




EDIT: try it before buying/replacing PSU, i was told that it was my PSU, but it wasnt...

 

[Despoiler]

EDIT: try it before buying/replacing PSU, i was told that it was my PSU, but it wasnt...

PSU testers are a cheap and incredibly handy tool. I wouldn't replace or RMA a PSU without knowing it's bad.

 

[plopke]

I had the same issue for 1 WHOLE month with new R9 380/GTX 570 with Windows 10...

after 1 month trying this and that ... I fixed it by updating Windows 10, and installing the card on other PCI slot..
i think each PCI slot act differently than the other or the one that the card was installed on was defective...

try installing the card on other pci slot and report back




EDIT: try it before buying/replacing PSU, i was told that it was my PSU, but it wasnt...

Which motherboard do you have? Almost all latest motherboards these days clearly state a preferred PCIe x16 slot for the primary video card.

 

[Leopardos]

Which motherboard do you have? Almost all latest motherboards these days clearly state a preferred PCIe x16 slot for the primary video card.

ASus P8Z77-v pro

 

[ZGR]

Have you tried the other x16 pcie slot? There is also a x4 you could try; just to see if there is a change in stability.

Are you and your friend using the same cables for video? 980 ti may have a problem with the cable you use OP?

Rather than trying new drivers, what about a driver that is older but still supports w10 and 980 ti?

After installing 980 ti again, reset CMOS.


When computer blackscreens with 980 ti, is there an error code on the motherboard for a clue? Sorry if nothing helps...

Last resort would be to take build out of case and try a different PSU. Or buy a similar sff psu and replace the current one. Good luck!

Other thing is the power cables themselves. Are you sure you are supplying the ti with enough power from the PSU?

 

[Elixer]

PSU testers are a cheap and incredibly handy tool. I wouldn't replace or RMA a PSU without knowing it's bad.

You get what you pay for in PSU testers. Most of them are very basic, and it will not help here, they don't provide enough load.
You can do pretty much the same with a multi-meter.

 

[plopke]

ASus P8Z77-v pro

So from which slot did you go , manual state PCIe16_1 is the preferred single video card setup slot and a 16/x or 8/8 setup, does not mention what you get usingt he PCIe16_2 solo and PCIe16_3 is x4. Interesting case nonetheless and something he can try.