WHEA_UNCORRECTABLE_ERROR from GPU usage

[Special Agent 星雨]

Hi everyone, first of all, specs:

 

Specs:

R7 3700X STOCK

LianLi Galahad 360mm AIO

Asus X570-E motherboard

KFA2 RTX 2080Ti SG on latest and 2nd/3rd latest drivers

Asus RX 580 8GB Dual on latest Radeon drivers

EVGA 850W P2 PSU

Samsung 970 Pro 512GB NVMe as OS drive.

 

I'd just to start off by saying that I understand the WHEA BSOD is usually because the CPU tells Windows that it has detected a hardware fault, so the issue is usually to do with CPU instability, or temps etc.

 

But hear me out.

 

This ONLY happens when I run a GPU benchmark, or any games that makes GPU hit 100% usage.

 

Things I have tried:

 

CPU:

Reset BIOS to default, even took the CMOS battery out for 5 min.

CPU temps OK (peaks at 40-50C whenever the BSOD occurs)

Aida64 CPU/FPU/Cache OVERNIGHT with NO issues.

Cinebench R15/R20 no issues

P95 small FFT no issues

 

Storage:

Reinstalled Windows using built in reset feature.

Reinstalled using USB stick.

 

GPU:

Happens on both a RTX 2080Ti and RX 580 8GB. Although, if I load up a benchmark, the 2080 Ti will run from anywhere between 1 to 50 seconds before BSOS occurs. RX 580 8GB crashes the PC instantly, as in within 1-3 seconds of loading up Unigine Heaven. I've also tried different PCI-e power cables.

 

Motherboard:

Tried both PCI-e x16 slots, with both GPUs, same result.

 

RAM:

I have tried it at stock and DOCP (3000Mhz C15->C16) settings, ran (at 2133Mhz) Aida64 memory stress test, ran Memtest64, completed OVERNIGHT with no issues at all. Currently running at 2133Mhz and the BSODs still happen.

 

I am confident both my GPUs are fine... because the the RX 580 8GB works perfectly in our home media system (can't fit the 2080Ti in there to test), but why do GPU benchmarks/games cause the BSOD then...?

 

Also, WhoCrashed was unable to find crash logs because the BSOD appears for about 2s before the PC instantly restarts. I was only able to know that the error code was WHEA UNCORRECTABLE ERROR due to using slo mo on my phone to record it.

 

Any insight/input/help will be much appreciated, I am out of ideas...

 

 

Edited September 23, 2020 by Special Agent 星雨

[Special Agent 星雨]

EDIT*

Just to be clear, the PC is perfectly fine when web browsing, doing assignment/coding, even playing light games like Among Us, League of Legends or basically anything that doesn't make the GPU usage hit 100%. But as soon as anything demanding is loaded up, it BSODs instantly.

I've also tried applying a -200Mhz to the core and memory of the 2080Ti and RX 580, still the same thing.


EDIT 2*

Also something else that's weird, each time it BSODs, it seems like... my Windows settings get reset? I turned UAC off because it asks for permission each time I open MSI afterburner, but after each BSOD UAC is back to the default, recommended level.

EDIT 3*

Yup, it literally ONLY crashes if GPU usage hits 100%. If I cap FPS to 30 in PUBG, GPU usage hovers around ~20% and there are no crashes. As soon as I remove FPS limiter, GPU usage hits 100% and the PC freezes within seconds.

 

At least I can still do my assignments, and literally everything else other than play games...

[Hairless Monkey Boy]

Try DDU. Be sure to follow all the recommended procedures.

[Special Agent 星雨]

4 minutes ago, HairlessMonkeyBoy said:

Try DDU. Be sure to follow all the recommended procedures.

I have used DDU in safe mode in between graphics card changes and driver version changes. No effect on my issue. 

 

This is so bizarre...

[Hairless Monkey Boy]

5 minutes ago, Special Agent 星雨 said:

I have used DDU in safe mode in between graphics card changes and driver version changes. No effect on my issue. 

 

This is so bizarre...

I see. At this point, then, I'd start to suspect the motherboard or the PSU, since these are the parts that remain constant.

 

I was going to also suggest a fresh install of Windows, but it looks like you tried that too.

[Special Agent 星雨]

8 hours ago, HairlessMonkeyBoy said:

I see. At this point, then, I'd start to suspect the motherboard or the PSU, since these are the parts that remain constant.

 

I was going to also suggest a fresh install of Windows, but it looks like you tried that too.

I will pull out the motherboard and test it while it is not installed in the case.

 

And yes, I have done a fresh Windows install with a USB as well.

 

I am starting to suspect the motherboard or PSU too, but fingers crossed it is them because I am literally out of ideas...

[Special Agent 星雨]

In an interesting twist, things seems to be working now, after the following changes:

 

- removed Cablemod PSU extensions for CPU, GPU (motherboard still using it)

- removed motherboard from the case, resting on the motherboard packaging box

- disconnected all other drives except the 2 x M.2 on the motherboard

- re-installed W10 via USB

 

Uniginw Heaven benchmark running for 30+ minutes and counting...

[Special Agent 星雨]

Update:

 

CPU power extension cable in use again, no issues.

 

Only GPU is using stock PSU cables now.

[Special Agent 星雨]

Final update:

 

Tested PCI-e cables without the extensions, all OK.

 

Re-assembled motherboard back into PC-O11 XL case, all OK.

 

I'm not sure what went wrong, but the issue seems to have fixed itself.

 

 

[DoctorNick]

6 hours ago, Special Agent 星雨 said:

Final update:

 

Tested PCI-e cables without the extensions, all OK.

 

Re-assembled motherboard back into PC-O11 XL case, all OK.

 

I'm not sure what went wrong, but the issue seems to have fixed itself.

 

 

Could be bad cables or placebo. If you have issues again, try installing windows while being offline and then disable automatic driver installation first thing when booting into windows. Install everything offline and keep automatic driver installation off. Also consider another bios for the motherboard.

[driggsy]

I hate to be that guy and revive this old thread, but I am facing the same exact issues as the OP, down to being unable to take my logs into WhoCrashed because my computer blue screens for 2 seconds before restarting. I also had to record my monitor with my phone and scrub through the video to find the error.

 

I reformatted my C: drive and that seemed to fix the issue for a few days... but then tonight my computer crashed while trying to play a game.

 

@Special Agent 星雨, were you able to find the root cause of this issue?

 

Specs:

 

3900x

3080 FE

64 gb Trident Z Neo 3600

ASUS ROG X570 Crosshair VIII Hero

[DecentSoul]

I am also having the exact same simptoms with my GPU. This is the only place I see other people describing this issue exactly like it happens to me too, which makes me believe there is a common factor or toot cause that needs to be eliminated. 

@driggsy @Special Agent 星雨 any ideas on how to permanently resolve the issue? I'll try to unbuild and rebuild my computer, see if that makes a difference for now. 

[rjabellax5]

I'm also having this issue.

In my case, I tried running unigine valley in linux mint 20.2 mate edition with extreme preset for an entire morning and never crashed. But when I run it on windows version 21H2, I got BSOD after 5mins. In windows, all is working fine when just browsing the web and watching videos youtube and also tried running aida64 stress test with cpu,memory,storage checked for about 12 hours and everything is fine and no BSOD but when I play games like warzone or run unigine valley, it will BSOD after about 5-6 mins. similar issue with the OP

my specs:

motherboard: Gigabyte a320m-h

CPU: Ryzen 3 3100

RAM: Kingston Hyperx 2600

GPU: Zotac OEM RX 580 4GB

Storage: Samsung 970 Evo Plus 250GB(boot drive)

               2x Hitachi 500GB HDD

PSU: Zalman 600W

 

Things I tried:

1. updated the motherboard BIOS to the latest

2. Fresh install of windows and updated to the latest and also the drivers

3. reseating the RAM and GPU

4. ran Memtest86 no errors found and passed

[SoFarKingFast]

I have been having this issue for quite some time. I've found a fix is to usually unplug all the GPU power connectors, or even swap the 8 pins, then put them back. Usually get performance back.

Sounds like it is a power cable issue, after seeing everyone else having the same problems.

[camera]

On 10/25/2021 at 4:25 AM, SoFarKingFast said:

I have been having this issue for quite some time. I've found a fix is to usually unplug all the GPU power connectors, or even swap the 8 pins, then put them back. Usually get performance back.

Sounds like it is a power cable issue, after seeing everyone else having the same problems.

Dude, it seems like you saved my life. I had the same problem and tried EVERYTHING I could software-wise but it kept crashing. After removing GPU, giving it a good ol' canned air wash and putting it back in it works wonders. It's been 2 hours and nothing has crashed, so maybe I'm celebrating too early, but everything points in the right direction. Sorry for the necro but you saved me! Thank you so much!