You are not logged in.
I've been trying to get my GPU passed to KVM since yesterday, and I just kept running into the problem that the GPU would not use the vfio-pci driver instead of the nouveau driver. I've tried :
- adding vfio-pci as a softdep in modprobe.d (doesn't change anything).
- blacklisting it in modprobe.d (doesn't change anything).
- blacklisting it in the kernelparameters (the GPU has no active drivers loaded, but it still lists nouveau as a kernel module in lspci).
- loading it via initramfs (the system hangs at something like "dev/nvmep0n0: clean, <numbers>/<numbers> files, <numbers>/<numbers> blocks" after starting which I think is from fsck?).
If anyone has ideas about why the vfio drivers weren't being loaded, that could be of great help (... for my next gpu).
More importantly, however, my system shut off out of the blue, and when I powered it on again the GPU sparked. I haven't tried turning it on again since; I assume the GPU is dead - it is clearly burned near the PCI slot (the PCI itself on the GPU and on the motherboard appear undamaged). I have two questions about this:
- Was this likely a faulty GPU, or did messing with the drivers somehow cause this? (I suspect the former)
- Should I get a new PSU? Is there a way to check if *that's* damaged? Would anyone advise against turning my system on again?
The GPU is a GT 1030 since I just needed something small to run music production software. Maybe this GPU isn't worth it, even for a small workload?
Update: the computer no longer turns on ![]()
Update 2: my system seems to work again after reseeding the RAM (what???), but I'm still wondering about the GPU
Last edited by not_woowoo (2024-05-22 00:13:56)
Offline