Question GPU keeps crashing - - I have no idea what's going on ?

[StemelUK]

Hi all,

I'm having a very strange problem with my PC that I think is related to my Asus RTX 4080 GPU, but I'm not sure.

I'll be happily playing a game and all of a sudden my monitors stop receiving input. My PC is still alive, I can hear the game audio and I can still talk on Discord, but I just have no display output. This has been happening intermittently for a few months now and I meant to look into it, but it wasn't happening enough that it was bothering me. And then last night I have 3 crashes within an hour, so either it's getting worse of Helldivers 2 seems to trigger this problem more.

I had a look in Event Viewer based on some Google recommendations and I can see the attached events (these are from the 1st or 2nd occurrence, but the same events pop up in recent crashes).

I've updated firmware, updated drivers, I even under clocked my GPU by a good 10% but still no dice.

I tried to recreate the problem this morning by firing up Furmark @ 2K. It ran solid for 10 minutes, no problem. I then opened Firefox to download Prime 95 to run at the same time and mid way through the PC crashed. After a restart I managed to get both Furmark and Prime 95 going for a solid 30 minutes, but no crash. So I'm left completely scratching my head.

There is one potential clue. During the first Furmark run this morning I could have sworn I saw a red LED on my GPU next to the power socket, and by the time my brain twigged to look properly it was off. I'm now wondering if my problems aren't GPU specifically but maybe PSU related, although I feel like the entire PC would die if it were. Or its cables to my GPU.

Any thoughts or ideas of what I could try to debug this?

Specs:

MBD: ASUS ROG Strix Z690-F
CPU: Intel Core i7-13700K
GPU: ASUS GeForce RTX 4080 TUF 16GB
RAM: G.Skill Trident Z5 32GB (2 x 16GB) 6000MHz DDR5
PSU: 1200W Corsair HX1200

https://imgur.com/a/hR0ATV0

 

[Fix_that_Glitch]

You can try reseating the gpu, using one stick of ram at a time, or do a memtest to rule out your ram as the culprit. And monitor your temps on cpu gpu.

 

[StemelUK]

You can try reseating the gpu, using one stick of ram at a time, or do a memtest to rule out your ram as the culprit. And monitor your temps on cpu gpu.

Reseated the GPU, we'll wait and see on that one I guess. I performed 2 memory tests, the Windows built in one Memtest86, no issues on either. GPU temps never go above 70, CPU does get close to 100 and thermal throttles, but only when it is FULLY loaded.

 

[StemelUK]

You can try reseating the gpu, using one stick of ram at a time, or do a memtest to rule out your ram as the culprit. And monitor your temps on cpu gpu.

Reseat did nothing "Crashed" within 10 minutes of firing up the game.

 

[Unolocogringo]

Any way to try a new power supply?
The HX 1200 is about 7 years old now. Being an older design it may not deal with the power spikes the video card produces.
Also unlimited CPU power is a possibility, your CPU motherboard may be affected.
https://www.tomshardware.com/pc-com...blame-other-high-end-intel-cpus-also-affected

 

[StemelUK]

Any way to try a new power supply?
The HX 1200 is about 7 years old now. Being an older design it may not deal with the power spikes the video card produces.
Also unlimited CPU power is a possibility, your CPU motherboard may be affected.
https://www.tomshardware.com/pc-com...blame-other-high-end-intel-cpus-also-affected

Sadly not, not without buying one.

Interestingly I massively underclocked my GPU tonight, ~80% on power, boost, and memory, and I had no crashes. This could still be the GPU or PSU, but it at least I know it's potentially GPU power causing it. Why Furmark and Prime95 don't trigger it consistently in that case is beyond me.

 

[Phaaze88]

Why Furmark and Prime95 don't trigger it consistently in that case is beyond me.

They're too consistent, making them less stressful than the actual applications you're wanting to run.
Consistent loads are easier on power management systems than sporadic ones.

 

[StemelUK]

They're too consistent, making them less stressful than the actual applications you're wanting to run.
Consistent loads are easier on power management systems than sporadic ones.

That's good to know. It's making me think it's more likely to be the PSU. Putting in the 4070 I bought may "solve" the problem through less power draw, but the PSU may still be having issues. Their website says it's fully compatible with the 4080 so not an issue there. Might see if I can get a different PSU and replace my current one. I just need to ensure the cables are compatible.

 

[Fix_that_Glitch]

That's good to know. It's making me think it's more likely to be the PSU. Putting in the 4070 I bought may "solve" the problem through less power draw, but the PSU may still be having issues. Their website says it's fully compatible with the 4080 so not an issue there. Might see if I can get a different PSU and replace my current one. I just need to ensure the cables are compatible.

Always use the cables supplied with the new psu.

 

[StemelUK]

New PSU orderd. Got one with a native 12+4. I'll try this first and then see if the GPU is still misbehaving. Thanks for all your help, I'll report back.

 

[faalin]

Also download DDU and remove all Nvidia drivers, download and install fresh ones from Nvidia's webpage.

 

[strelok49]

1) Update BIOS
2) Use only default BIOS values until the problem is fixed
3) Check all videocard and PSU Cable again. Remove and install again.
4) Are you using any adapter?

If when you say crashing you're mean that you PC just turn off, this is probably because there's no power enough to keep going or the temps are just too high. Specially CPU temps. Probably not, but keep a eye on it.

If any of these helps you, so you have a faulty adapter (if you're using any), a faulty video card, a faulty PSU, bad power supply from distrution company or some bad contact between PSU, motherboard and videocard.

 

[StemelUK]

Had a crash tonight. I'm 99% certain it hasn't done this before, but the red LED was on my GPU. This smells like PSU and/or cables right?

EDIT: Yes, that's ZIP ties on my CPU cooler. #reasons, and honestly they're not terribly dumb ones

 

[strelok49]

Had a crash tonight. I'm 99% certain it hasn't done this before, but the red LED was on my GPU. This smells like PSU and/or cables right?

It's not clear, but it seems that you're probably using some adapter/extension right?

It's a bad contact, faulty adapter/extension cable (not recommended at all if it's not the original adapter) or faulty PSU.

Also it could be a faulty cable from modular PSU.

This red light, at least from what I know, means that power supply is not being enough.

 

[StemelUK]

I replaced the PSU and all the PSU cables. While I haven't been able to properly test drive the machine yet, I'm hopeful that this has solved the problem.

Interesting, I may know a potential cause. My old PSU has a switch on the back for single rail or multiple and it was set to multiple. My assumption, based on the fact most PSUs I saw when buying the new one was single, is that multiple could have caused problems? Maybe not.

Either way, I'm back in business and saved some $$$ by not wasting it on getting a new GPU. Thanks all!

 

[mazevgg]

Hey, same problem here with same GPU. Nice to read i´m not the only one, but it sucks anyway. I'm really looking forward to your answer buddy.

 

[StemelUK]

Hey, same problem here with same GPU. Nice to read i´m not the only one, but it sucks anyway. I'm really looking forward to your answer buddy.

The new PSU has definately solved this for me. So I recommend looking at your power cables and/or PSU.

 

[mazevgg]

The new PSU has definately solved this for me. So I recommend looking at your power cables and/or PSU.

Same here... new PSU solved this Problem. Seems like Helldivers 2 causes high load peaks.