For the past few months I’ve been having problems with the GPU falling off the bus not long after launching 3d games, but only occasionally. When it happens it only happens shortly after a game starts up (Specific games: FF14, Deep Rock Galactic Survivor). If the game launches correctly, I have zero stability issues.
When this happens, the laptop completely fails to see the GPU until I shutdown with the laptop unplugged for a minute or two. The problem persists across reboots, booting into windows, etc. The BIOS won’t show the dedicated gpu, and nothing can see/initialize it.
I’ve attempted to try different driver versions to no avail. I’ve also tried clean driver reinstalls. I can find no specific reason for this. Using nvidia-smi and logging, at no point is the graphics card coming close to its max power usage, nor does it generally even use its max clock.
I would love any information that might help here, including if there’s a way to reinitialize the graphics card from linux without having to power cycle.