Hi there.
I habe a debian 12 VM on proxmox running - only für ollama usage.
Did the walkthroug for proxmox https://www.youtube.com/watch?v=IE0ew8WwxLM to pass GPU through.
I was able to add the GPU under Proxmox/Datacenter/Resource Mappings.
Adding Hardware und VM/Hardware/PCI Device was also possible
In the VM i was able to list pcie devices properly via 'lspci'.
I Installed driver from Nvidia for Debian 12 and was able to use 'nvidia-smi'.
Constantly monitoring with 'nvidia-smi -lsm' showed, that model was not loaded in ram, when pulling ollama model using "ollama run llama2" or "ollama run llama3.2".
What did I do wrong?
Also tried it with a ubuntu 24.04 .. same result.
Thanks in advance
I habe a debian 12 VM on proxmox running - only für ollama usage.
Did the walkthroug for proxmox https://www.youtube.com/watch?v=IE0ew8WwxLM to pass GPU through.
I was able to add the GPU under Proxmox/Datacenter/Resource Mappings.
Adding Hardware und VM/Hardware/PCI Device was also possible
In the VM i was able to list pcie devices properly via 'lspci'.
I Installed driver from Nvidia for Debian 12 and was able to use 'nvidia-smi'.
Constantly monitoring with 'nvidia-smi -lsm' showed, that model was not loaded in ram, when pulling ollama model using "ollama run llama2" or "ollama run llama3.2".
What did I do wrong?
Also tried it with a ubuntu 24.04 .. same result.
Thanks in advance