Question [GTX 1080 FE] Black screen and 100% GPU fan usage during gaming

[MC23]

Hi all. I bought a used video card (GTX 1080 FE) a few days ago from a friend of mine; after cleaning it from dust and re-applying the thermal paste, I did a couple of benchmarks to test its proper functioning (since it hadn't been used for a couple of months), without any problem.

Today, while I was playing Dying Light 2 for a couple of hours (game which I've been playing for the past few days without any problem), all of a sudden the screen went black, the card's logo (which I set to off on Nvidia LED Visualizer) lit up and the card's fan spiked to 100% (at least I think so, since I only heard the very loud noise of the fans while the screen was just black). When the problem occurred for the first time, the installed driver version (DCH) was 511.79; after this problem, since the new version (512.15 always DCH) was released a few days ago, I proceeded to update it. Despite this, the problem continued to occur, on several games, after a few seconds from startup. I proceeded to uninstall, in safe mode without internet, the video drivers using DDU and then I re-install them (I tried both version 511.79 and 511.65), but not even then I solved the problem.
In the Windows event viewer I only found a critical error Kernel-Power 41, I think due to the fact that I used the reset key (NOTE: sometimes, during the tests I did, I just had to press the power button once to turn off the PC regularly, information perhaps not necessary but that still I prefer to provide).

Following some research on Google, I found a topic on the Microsoft forums and then I discovered a file inside the folder C:\Windows\Minidump (I can't attach to this post, so I uploaded it on wetransfer at https://we.tl/t-vfwdduYX4T). Regarding this file, I noticed that it is generated only in case I don't shut down the PC, if I just wait for it to reboot itself after the problem occurs (this can take some minutes to happen).

Here are my PC specs:
MOBO: MSI PRO Z690-A (BIOS updated to the latest version)
CPU: i5-12600k (no OC done)
Video Card: GTX 1080
PSU: Corsair RM850x
Monitor: Benq GW2765
OS: Windows 11 Pro (updated to the latest version, no Windows Insider)

I have tried, from Windows settings, to set the integrated graphics card to open different games and I have not had any kind of problem with that one.

I ask for your help to know:

1. if I'm struggling with a software or hardware problem (if it's possible to determine this based on the info provided)
2. what other steps I can follow to try to solve the problem

PS: during "normal" use of the PC, the video card works without problems.


Thanks in advance

EDIT: 2022-03-28 I tried running the Render Test from GPU-Z and the problem occured again, this time the fans didn't went to 100% tho
EDIT2: 2022-03-29 the problem happened while using VLC as well

 

[drivinfast247]

What are GPU temps while under load?

 

[Historical Fidelity]

Sounds like a nvidia driver crash, my 1080’s lighting would revert to stock when the drivers crash

 

[MC23]

What are GPU temps while under load?

Between 78° C and 81° C

Sounds like a nvidia driver crash, my 1080’s lighting would revert to stock when the drivers crash

I think I checked this, when I open GeForce Experience the driver is always the same (511.79); unless I need to check the version from Control Panel or another place

 

[Historical Fidelity]

Between 78° C and 81° C



I think I checked this, when I open GeForce Experience the driver is always the same (511.79); unless I need to check the version from Control Panel or another place

Sorry that doesn’t make sense, you can’t tell if the nvidia driver crashes except for black screening out of a game.

 

[MC23]

Sorry that doesn’t make sense, you can’t tell if the nvidia driver crashes except for black screening out of a game.

Hm I don't think I understood your first message then. I thought you meant that, when the crash happens, the driver of the GPU reverts to the stock one: does this mean that the driver goes back to a previous version? Anyway, if it's a driver issue and DDU didn't work, what else can I try?

 

[Historical Fidelity]

Hm I don't think I understood your first message then. I thought you meant that, when the crash happens, the driver of the GPU reverts to the stock one: does this mean that the driver goes back to a previous version? Anyway, if it's a driver issue and DDU didn't work, what else can I try?

Oh okay I see, no the driver will remain the same, when the driver crashes, the driver will close and re-initialize. If it doesn’t re-initialize then the black screen would remain until you restart the computer.

As far as why this is happening, most often it is due to instability. My 1080 causes driver crash with black screen when I push my overclock too high and it becomes unstable. Given that the 1080 is now 6 years old and you do not have a detailed history of it use beforehand, I would say it’s possible that the gpu chip has degraded to the point where stock settings are unstable. Try to underclock the 1080 by like 100-200 MHz and see if it becomes stable while playing dying light 2 like you’ve been playing. Let me know if that works. In the mean time I’ll start thinking of other ways to diagnose the card.

Also, do you have another graphics card to test in your system just to rule out something other than your graphics card like the pci-express slot, cpu pci-express lanes etcetera?

 

[MC23]

Try to underclock the 1080 by like 100-200 MHz and see if it becomes stable while playing dying light 2 like you’ve been playing. Let me know if that works. In the mean time I’ll start thinking of other ways to diagnose the card.

I never played with clock settings, do I just use a software like MSI afterburner? Also, yesterday the video card crashed immediately after I run the "render test" from GPU-Z. I activated the debug option in Nvidia Control Panel, as suggested by an Nvidia assistant who I talked with, but this didn't help.

Also, do you have another graphics card to test in your system just to rule out something other than your graphics card like the pci-express slot, cpu pci-express lanes etcetera?

Unfortunately, I don't. I purchased this card because my 970 died some months ago. I found, however, a friend of mine that will make me test the card on his system. I'll report the results when I do this.

 

[Satan-IR]

As said above the crashes might be due to hardware issues with the card/GPU. Or the driver might be faulty/corrupted. I would use DDU to completely remove the graphics driver and install again.

Also I ran the dump you linked.

The driver that comes up in that dump file is nvlddmkm.sys and the process is RocketLeague.exe. Were you playing when the crash happened and the minidump was dropped? The nvlddmkm.sys is a nvidia driver thing and might be related to issue as the driver being corrupt. Or might be other driver (VPN) or network messing things up and the display diver might be the victim.

The bugcheck is a 133 (DPC_WATCHDOG_VIOLATION). Also SWD\WireGuard\{8D69708D-DDEC-A599-BB02-0475A5D2150E}. Do you have a software or VPN client/app that uses the WireGuard protocol?

 

[MC23]

As said above the crashes might be due to hardware issues with the card/GPU. Or the driver might be faulty/corrupted. I would use DDU to completely remove the graphics driver and install again.

Also I ran the dump you linked.

The driver that comes up in that dump file is nvlddmkm.sys and the process is RocketLeague.exe. Were you playing when the crash happened and the minidump was dropped? The nvlddmkm.sys is a nvidia driver thing and might be related to issue as the driver being corrupt. Or might be other driver (VPN) or network messing things up and the display diver might be the victim.

The bugcheck is a 133 (DPC_WATCHDOG_VIOLATION). Also SWD\WireGuard\{8D69708D-DDEC-A599-BB02-0475A5D2150E}. Do you have a software or VPN client/app that uses the WireGuard protocol?

Hello, first of all, thank you for finding the time to go through the dump file!

As I said in my post, the DDU procedure has been already made, three times now, without success. As for the other questions:

• yes, I was playing when the dump was generated, it was immediately after it happened with DL2, so I ran Rocket League to see if it was a performance/temperature problem caused by running a demanding game
• yes, I do have a VPN, and it does indeed use the WireGuard protocol.

Also, today I had the same problem when I opened VLC to watch a video file.

 

[spentshells]

80c load temps after new thermal compound.

You've put too much paste, that card should never reach 70C at load.
Not ever! You're memory should have new thermal pads and your gpu has far too much paste.

 

[MC23]

80c load temps after new thermal compound.

You've put too much paste, that card should never reach 70C at load.
Not ever! You're memory should have new thermal pads and your gpu has far too much paste.

I put the usual amount I put on CPU as well, but I can do the repasting again even if I'm not sure that's the main problem. I'd expect the card to reach that temperature after at least some minutes of gaming, but the card crashes after like 3/4 seconds while the initial logos are still displaying. Also, it crashes with games like Phoenix Wright or even software like VLC...

 

[MC23]

80c load temps after new thermal compound.

You've put too much paste, that card should never reach 70C at load.
Not ever! You're memory should have new thermal pads and your gpu has far too much paste.

I got curious about this comment and found this article: https://www.anandtech.com/show/10325/the-nvidia-geforce-gtx-1080-and-1070-founders-edition-review/30

It seems like 80/82 °C is not that weird for a single fan founders edition?