Graphics card crashes

[Gothicwolf]

Hello,

My system specs:

Z790 motherboard (ASRock ASRock Z790 Lightning WiFi)

Intel 12900k

Zotac rtx 3080 trinity oc

CORSAIR DOMINATOR PLATINUM RGB DDR5 RAM 32GB (2x16GB) 5600MHz C36-36-36-76 1.25V

Powerspec 1250w psu

Samsung 970 evo

Windows 11 pro 64bit

 

My problem:

After some time (between 5 and 45min) the screen begins to tear before crashing with a black screen. On reboot the motherboard won't recognize the graphics card until I remove it, boot with the on board graphics and then put it back in. Crashes happen during playing, benchmarks, stress testing and even just browsing the web.

Things I've tried already: swapped motherboard, swapped psu, reinstalled windows, underclocking, lowering power limit for gpu, reapplying thermal paste, different pcie slot, different bios and driver versions, disabling xmp.

Event viewer shows always nvlddmkm errors with various error id's.

Edited July 26, 2024 by Gothicwolf
Card crashed after a while in different pc too

[Mumintroll]

You can try that 3080 in another pc, borrow a friend's pc if you don't have a spare. If it fails there aswell then it's probably bad.

Just to be safe you can test with another gpu in your pc to see what happens.

 

Do you still have warranty on the 3080?

[Gothicwolf]

6 hours ago, Mumintroll said:

You can try that 3080 in another pc, borrow a friend's pc if you don't have a spare. If it fails there aswell then it's probably bad.

Just to be safe you can test with another gpu in your pc to see what happens.

 

Do you still have warranty on the 3080?

Unfortunately 2 months out of warranty

I tried it yesterday in a different pc, it ran for a few hours with heavy load (doom on max graphics in 2k,furmark knot gl etc) but eventually crashed.... At this point I'm certain something is wrong with it. First I thought it is a memory chip but doom pretty much used all 10gig without problems, letting me belive it might be thermals (hot spot temperature gets up in the high 80s,board temp low 80s, power draw 320w in furmark...)

[Mumintroll]

22 minutes ago, Gothicwolf said:

Unfortunately 2 months out of warranty

I tried it yesterday in a different pc, it ran for a few hours with heavy load (doom on max graphics in 2k,furmark knot gl etc) but eventually crashed.... At this point I'm certain something is wrong with it. First I thought it is a memory chip but doom pretty much used all 10gig without problems, letting me belive it might be thermals (hot spot temperature gets up in the high 80s,board temp low 80s, power draw 320w in furmark...)

 

It should start to thermal throttle if the temperature goes too high. I think it's like just above 90 degrees on 3080.

But I don't think the temperature is the cause here, since you said it was crashing during light usage aswell.

 

Sad to hear that it's past warranty....

[Gothicwolf]

I only suspect thermals because a lot of people report thermal crashes and issues with the zotac cards and it takes longer to crash it when it's cold. slapping a waterblock on it would be an easy workaround if it's temp but I don't want to waste $200 if it doesn't fix it.