Question GTX 3080 Random Loss of Display but not full Windows/system crash?

[ralence]

I'm running a PowerSpec pre-built since I couldn't source a GPU back in Feb. As of about 2 months ago the Nvidia FE 3080 has started randomly shutting off. Sometimes after 5 minutes, sometimes after 5 hours. It's seemingly random, almost like it's bad GPU memory, but I'm not sure how to test it?
Other times the GPU will shut down when nothing is running aside from one instance of Chrome. I tried stress testing with Furmark to see if I can create the crash so it's easy to warranty claim, but it ran for 15 minutes just fine without a crash.

It doesn't crash the system so there is nothing in windows event viewer. The LED's on the card shut off, the fan continues running, but there is no video output on either of my monitors. I can still hear the sound output from whatever is running on the computer.

System specs are:
ASUS x570 Plus Wifi (Bios 4408 10/28/22)
AMD 5900x
16g ddr4-3200 ram
FE 3080
2TB SSD
Windows 10

I swapped out the stock 750watt PSU for a new Corsair RM1000e, thinking it may have been a power draw issue, but that hasn't helped.

So far I've tried:
Removed drivers with DDU in safe mode, and re-installed current drivers (as of yesterday) with elevated priveleges.
Updated AMD MB drivers
Swapping two different monitors, tried two different HDMI and DP cables with both.
Disabling Gsync
Plugging in directly to wall outlet to bypass surge protector.
CTRL-SHIFT-WIN-B Beeps but does not reset display.

Monitoring temps the GPU crashes at somehwere around 38C, never gets above 45C. So I don't think it's a cooling issue.

Nearest MicroCenter is a 2 hour drive for me, so I'd rather do as much troubleshooting as I can before I have to make the trip to return under warranty.

Is there anything I haven't tried? I can't replicate the crash which frustrates me. It's seemingly random, including with nothing running except Chrome.

Any help is greatly appreciated! Otherwise I'm driving 4 hours to warranty claim it.

Thanks!

 

[ventsiboi]

I experience the exact same issue.

I purchased 6800xt a month ago - second-hand.

My monitor randomly says No signal and then the motherboard LEDs start going through each component. It usually stays on yellow (RAM) for a while and then turns white and it gets stuck. However, the less on each component remains on.

Sometimes it crashes within 10 min and others take around 24 hours.

I have noticed the crash seems when I have chrome open and where there is little to no activity.

I have not started a game in the background and played YT. It didn't crash so far and my card stays on 20% usage.

Can't wait to find out potential fixes - tried DDU, Chipset updates, disabling freesync, but nothing seems to help

PS
The card doesn't have any weird behaviour in terms of performance, temp or power usage.

Build:
3300x
B450-f strix
2x4GB 3000 DDR4 Teamgroup (also tried 2x8GB 3600 Corsair Vengeance SL)
850 EVO SSD
MSI Gaming X trio 6800xt
Corsair CX750M

 

[JamesSmith79]

this sounds very familiar im currently troubleshooting this on a 6950XT

 

[ventsiboi]

I have found the weirdest fix.

I noticed the issue happens when the gpu has 0 load. I keep an indie game always on windowed mode on the background and this prevents the gpu from disconnecting.

Try this

 

[JamesSmith79]

I have found the weirdest fix.

I noticed the issue happens when the gpu has 0 load. I keep an indie game always on windowed mode on the background and this prevents the gpu from disconnecting.

Try this

Honestly if this fixes it, I'm just gonna RMA it I'm still within the 28 days return window where i should be able to get my money back (EU) because it's honestly unacceptable that it works like this ( at least for me ). I'll just get a 4080 then and call it a day.