Graphics Cards

[Ferroic]

I have been having an issue for a few months where my computer (or possibly just GPU) will crash. The crash consists of a black screen with some audio that will eventually stop. The fans of my GPU will run at maximum speed, and the computer will be completely unresponsive, forcing me to turn it off using the power button.

The crash (so far) only occurs when running certain GPU intensive programs and games, such as Warframe, Vermintide 2, and Furmark. Interestingly, the crash has never occurred when using some other bench-marking software, such as Unigine Heaven, 3Dmark Firestrike, and prime95. To try to fix this issue, I have tried reinstalling drivers, running the pc in safe mode, and rolling back to earlier drivers.

I initially thought that this was a hardware issue with my GPU, however the investigation of Warframe support has lead to the following conclusion based on errors in the even viewer:

'Looks like the same series of errors each time, each leading the AMD Master Driver failing, which because of the black screen and the PC "stopping"'

For reference, I have a MSI RX 580 8GB Armor mk2 and an R5 3600. GPU temperatures are stable.

Application Events file

System Events file

If someone could please try to find a solution that prevents these crashes, I would be very grateful.

[kshutch00]

Are you getting WHEA-logger event 18 errors?

[Ferroic]

No, none of the events with ID 18 are related to the WHEA-logger, and all are only categorized as information, so I think they are unrelated to the issue that I'm having.

Warframe support seemed to think that the relevant error was: 

Event 7000, Service Control Manager

"The AMDRyzenMasterDriverV14 service failed to start due to the following error:
Cannot create a file when that file already exists."

Personally I'm not too sure, since it can be seen that this error was logged after I was forced to restart the PC (which I can see as a critical error "The system has rebooted without cleanly shutting down"), but I'm not 100% sure how the event logger works so I guess it might still be relevant.

[kshutch00]

Do you need RyzenMaster?  If you are using it to overclock or something, why not put the same values in your boards bios?  Maybe try removing RyzenMaster and see what happens.  I do not use it, so I cannot test any of this.

I have never seen an error like this on a service unless I was changing its state or start type and something was wrong with the services permissions or it was corrupted in some way.  That can be changed in the registry by giving the service (HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\[name of the service]\Start) Dword value a 2 for Auto, 3 for Manual, or 4 for Disabled.  Maybe see what happens when you try to change the AMDRyzenMaster service to manual or disabled from your services console (type services.msc into your windows search).  Maybe that could lead a greater understanding of what is going on.

Most reboots I have seen have some Critical or Error related to them in the event viewer.  Did you check to see if you have dump file?