Hello everyone. I am trying to passthrough 2 x nVIDIA T4s on a Dell 5820 workstation and while I followed all the steps that I think were needed, I get this error when trying to start a vm with one of the gpus attached:
I'm on PVE 9.1.6, Found quite a few posts on the web with the same issue but either they were for AMD hardware or the solutions did not apply in my case. Any help would be greatly appreciated.
Thank you.
UPDATE: Seems that a host reboot allows the vm to start with the gpu attached. But does this mean i would have to reboot the host every time i need to attach the gpu to another vm?
ANOTHER UPDATE: In the vm i set the nvidia-smi to watch every second and after a while (maybe 4-5 minutes) I get this error and that's it, I need to reboot the host if I want the gpu to be detected inside the vm again (or to be able to power the vm (Ubuntu 24.04) on for that matter, if I turn it off and on again, I get the PCI_NUM_PINS error again).
Code:
kvm: ../hw/pci/pci.c:1815: pci_irq_handler: Assertion `0 <= irq_num && irq_num < PCI_NUM_PINS' failed.Thank you.
UPDATE: Seems that a host reboot allows the vm to start with the gpu attached. But does this mean i would have to reboot the host every time i need to attach the gpu to another vm?
ANOTHER UPDATE: In the vm i set the nvidia-smi to watch every second and after a while (maybe 4-5 minutes) I get this error and that's it, I need to reboot the host if I want the gpu to be detected inside the vm again (or to be able to power the vm (Ubuntu 24.04) on for that matter, if I turn it off and on again, I get the PCI_NUM_PINS error again).
Code:
Unable to determine the device handle for GPU0: 0000:01:00.0: Unknown Error
Last edited: