Is my GPU dying?

[KungFuSpaghetti]

I've only installed my 3900x today but have since had the following all evening whilst trying to play pubg...

WHEA_UNCORRECTABLE_ERROR
As posted in another thread, the CPU was at 50 degrees or under showing all cores at 4175 according to hw monitor when it happens. Totally standard setup, no overclocking just xmp on my ram.

I have recently been seeing fairly frequent crash to desktop (on 2700x pre upgrade) with a pubg crash log generated. Usually I attribute this to the game, but it might happen 3 to 4 times in an evening. This above however is happening anything up to 15 minutes into a game...

Temps are low, rarely go above 40 when really pushed, no artefacting etc, just bsod or even black screen followed shortly after by rebooting.

All drivers are up to date, with all mobo and chipset drivers, windows is a fresh install (1903). GPU not overclocked, just running 120% power limit as it boosts to 2126mhz regardless...

Is my GPU dying or could this be ram, what's the best way to test and check before painful RMA process?

 

[KungFuSpaghetti]

Yeah bios on latest version 3.43

I'll try running a memory test and see where I get. Thanks guys.

 

[KungFuSpaghetti]

Thanks @Panos I'll give that a try today. I have 3.40 on a USB already.

 

[KungFuSpaghetti]

Okay I've installed 3.40 bios and still the same issues... just browsing the web and installing aida64 and crashed again, bsod with same exact error. PC was not being taxed in any way at the time.

 

[KungFuSpaghetti]

I didn't hard reset, just flashed back to 3.40, then loaded defaults.

When you say hard reset you mean remove the cmos battery?

I'll try that shortly, I've just put ram back to non xmp and running memtest64 on 20 passes.

 

[KungFuSpaghetti]

Aah yeah okay, can't do that... it will put me back to a non ryzen 3900x compatible bios and my mate put my 2700x in his PC already.

20 loops of memtest passed without errors but running at 2400mhz... going to do some super position runs to stress the GPU and see if it falls over. Then xmp on and same tests again?

 

[KungFuSpaghetti]

Asrock and MSI have said new bios fixing bugs next week.

Next week can't come fast enough! There's no rhyme or reason to it, crashes in games but even doing it in windows when it's doing nothing... then it's ran memtest64 without error. I've gone out and left it on at the desktop doing nothing. If it's okay when I get back then I'll enable xmp and run tests again.

 

[KungFuSpaghetti]

@Panos I was using 3.20 then did the instant flash on boot which put 3.43 on in June or whenever it came out.
Just as a side note, when i put the 3900x in I did a fresh install with 1903 and the first thing I did was put the chipset drivers on followed by nvidia drivers.

I can ddu nvidia drivers, but I'm wondering if it's my ram, as so far since using default none xmp it's been fine. Trying to get it to crash but multiple unigene superposition runs, memtest64 and aida64 extreme stress tests and not a single crash... thinking of trying some gaming later when I'm back home and if still stable, ram issue?

 

[KungFuSpaghetti]

Okay will leave at stock speed, I'm not messing with anything now. Ram at 2400mhz sounds crap but I can live with it for now if it gives some stability.

Looks like this issue is more common than thought and is indeed GPU... but software not hardware related. https://www.pcgamesn.com/nvidia/amd-ryzen-3000-bsod-whea-error

 

[KungFuSpaghetti]

Looks like I'm not the only one with this issue... asrock forum people reporting exactly the same thing. Turning xmp off is making it a lot more stable.