GPU power spikes? Crashing after a minute in any game.

[drive344]

My build is about 4 years old:
Gigabyte GA Z270XP-SLI mobo
i7-7700
Zotac 1080Ti
2x8GB Corsair RAM
Old PSU: evga 750W gold
New PSU: Corsair 750W gold

I've been plagued with random crashes nearly since I built the thing, but I was basically only playing PUBG with it and blamed it on that buggy game. Never had issues with it in any other game, even Half-life Alyx VR, and more technically demanding games. It used to just crash me to desktop, but then it started crashing the whole computer (no BSOD, just completely unresponsive requiring hard reset). In the past few months, however, it now crashes to desktop within a minute of loading up any game.

I've reinstalled the system from scratch several times, rolled back GPU drivers, etc etc. trying to get a stable build but it has just deteriorated. I bit the bullet and bought a new PSU to see if that could be the case, but the recent behavior continues with the new PSU. I've run MEMTEST 86+ and that passed without issue. FurMark3D runs fine and gives no indication of issue. The GPU watts do seem a little odd/noisy when running FurMark, but they seem fine during the minute of game that I can run before it crashes. Screenshots below.

So what do you think? Bad mobo? PCIe slots are shaky? GPU has a terminal disease?

FurMark (old PSU) (2 runs)

FurMark (new PSU)

Game behavior (State of Decay 2)

 

[Selaya]

Check your Event Viewer for logs.

 

[drive344]

Check your Event Viewer for logs.

Good call. This message seems to coincide with the crashes:
"Display driver nvlddmkm stopped responding and has successfully recovered."

 

[Selaya]

Yeah, that's a GPU driver crash. It recovered, but oftentimes games cannot recover w/ it so you will have to shoot 'em down (via the Task Manager, for instance).

As for the root cause, it's a GPU instability issue - it could be either an unstable OC or just a failing GPU. Try underclocking w/ Afterburner, sometimes it works. If not, your GPU's probably dead or at the very least on its way to the graveyard. (Wouldn't surprise me either, w/ a Zotac ...)

 

[drive344]

Yeah, that's a GPU driver crash. It recovered, but oftentimes games cannot recover w/ it so you will have to shoot 'em down (via the Task Manager, for instance).

As for the root cause, it's a GPU instability issue - it could be either an unstable OC or just a failing GPU. Try underclocking w/ Afterburner, sometimes it works. If not, your GPU's probably dead or at the very least on its way to the graveyard. (Wouldn't surprise me either, w/ a Zotac ...)

Hmm, fair enough. I tried my luck on the ZOTAC (good price at the time...), but I suppose I should have stuck to a better maker. Underclocking had seemed to help sometimes in the past, but the clock speed profile would never auto-load whenever I started up the computer, so I just gave up on it and it seemed fine for a while anyway. I always was just blaming it on PUBG. I'll try underclocking again, and in the meantime I'll cross my fingers that I can get my hands on a new GPU.

Thank you for the replies, by the way. Much appreciated

 

[Selaya]

You have to save your profile after you've created it for it to auto-load (I believe? Been a hot moment since I setup mine, and it's autoloading just fine so) by clicking save and then a slot number to save to.

 

[drive344]

You have to save your profile after you've created it for it to auto-load (I believe? Been a hot moment since I setup mine, and it's autoloading just fine so) by clicking save and then a slot number to save to.

Right, which I did (although I was using the ZOTAC Firestorm program instead, probably also a mistake) and it would just fail to load on start-up. I just underclocked with Afterburner just now and I got a good five minutes of gameplay before the crash this time. I'll try a larger underclock, but I think it's just time to build/buy a new rig. 4 years was a good run...

 

[chrcoluk]

First, stop using Furmark.

On the ram, if you want to test, use something decent like karhu ram tester or HCI.

Next how are things configured, any overclocks, voltage changes, custom voltage curves etc. on any of the hardware?

A good way if you suspect the GPU is quite simple, either reduce the power limit to below 100%, or even better underclock it. Safe to do, and if the instability stops, you found your culprit.

 

[Naki]

* Test the GPU in another PC if you can -- it if works fine there, then it could be mobo or PSU failing and your GPU - completely fine.

* Likewise, try another known working GPU on this same PC - if it works, then your Nvidia GPU is likely at fault.
But if other GPU fails too, then again it could be mobo or PSU/etc.