ProxMox 7.4.3 Cannot Install Nvida drivers.

[dyl4n20]

Hello

I have tried following these posts to get Nvidia drivers working and unfortunately cannot get passed the following error message.
"ERROR: Unable to load the kernel module 'nvidia.ko'. This happens most frequently when this kernel module was built against the wrong or improperly configured kernel sources, with a version of gcc that differs from the one used to build the target kernel, or if a driver such as rivafb, nvidiafb, or nouveau is present and prevents the NVIDIA kernel module from obtaining ownership of the NVIDIA graphics device(s), or no NVIDIA GPU installed in this system is supported by this NVIDIA Linux graphics driver release."
https://pasteboard.co/wP6w8b0sN0JY.png
https://pasteboard.co/OehiFdmQ2SOk.png

(things I have tried)
https://stackoverflow.com/questions/24734986/unable-to-load-the-kernel-module-nvidia-ko
https://tutorials.technology/tutorials/85-How-to-remove-Nouveau-kernel-driver-Nvidia-install-error.html#:~:text=Introduction 1 Step 1: Cleanup all nvidia package,vim /etc/modprobe.d/blacklist.conf ... 3 Step 3: Update initramfs
https://linuxconfig.org/how-to-disa...idia-driver-on-ubuntu-20-04-focal-fossa-linux

Please let me know if you require any additional information. Thank you for your time and support.

 

[kingwilliam]

which nvidia driver are you install?

 

[dyl4n20]

which nvidia driver are you install?

I tried these three Nvidia drivers. I am using a Geforce RTX 3060 Ti.
NVIDIA-Linux-x86_64-495.44 NVIDIA-Linux-x86_64-525.116.03 NVIDIA-Linux-x86_64-525.116.04

 

[kingwilliam]

what you want to do?
1. direct passthrough 1physical gpu to 1vm?
2. use gpu on pve host?
3. separate gpu to vgpu and then pass to vm?

Q1 if you want direct passthrough gpu to guest, you just add the pci device to vm

 

[emunt6]

hi!

1., Install the "pve-headers" and "pve-headers-`uname -r`" and "build-essential" packages.
2., Run the NVIDIA-Linux-x86_64-x file you downloaded form nvidia website.
3., Create the following file with the content:
/etc/modprobe.d/nvidia.conf blacklist nvidiafb blacklist nouveau
4., Run the following command:
update-initramfs -c -d -u
5., reboot
6., DONE.
7., If you upgrade/update your kernel ( new kernel version), you need to do the process again ( the driver needs to build for the new kernel too ).

 

[dyl4n20]

what you want to do?
1. direct passthrough 1physical gpu to 1vm?
2. use gpu on pve host?
3. separate gpu to vgpu and then pass to vm?

Q1 if you want direct passthrough gpu to guest, you just add the pci device to vm

More along the lines of number 1, but number 3 sounds interesting if it’s possible to use my gpu in multiple VMS that is the route I am looking towards.

 

[dyl4n20]

hi!

1., Install the "pve-headers" and "pve-headers-`uname -r`" and "build-essential" packages.
2., Run the NVIDIA-Linux-x86_64-x file you downloaded form nvidia website.
3., Create the following file with the content:
/etc/modprobe.d/nvidia.conf blacklist nvidiafb blacklist nouveau
4., Run the following command:
update-initramfs -c -d -u
5., reboot
6., DONE.
7., If you upgrade/update your kernel ( new kernel version), you need to do the process again ( the driver needs to build for the new kernel too ).

When I run the Nvidia driver I get the error above, by chance can you share a link to

the "pve-headers" and "pve-headers-`uname -r`" and "build-essential" packages.

Mentioned above.

I am not able to proceed.

Am I supposed to be installing the driver on Proxmox or on the actual VM/ Ubuntu(in my case).

 

[kingwilliam]

Q1. if you just pass single vga card to vm, you can visit below url
A1. Proxmox GPU Passthrough to VM

Q2. if you want seprate single nvidia gpu to multi vgpu
A2. you can visit below url, but you must read supoorted-card section first
Supported cards - "YOUR RTX 30XX or 40XX WILL NOT WORK !!!"
NVIDIA vGPU on Proxmox

 

[emunt6]

When I run the Nvidia driver I get the error above, by chance can you share a link to

Mentioned above.

I am not able to proceed.

Am I supposed to be installing the driver on Proxmox or on the actual VM/ Ubuntu(in my case).

My instruction was for option 2:

1. direct passthrough 1physical gpu to 1vm
2. use gpu on pve host?
3. separate gpu to vgpu and then pass to vm

If you are going to the use option 1, the process is the same:
Guest VM, you need kernel-headers and build-essential (gcc + libs) package

1.,
$> apt install linux-headers-$(uname -r)
$> apt install build-essential

2., ( root user ) run the NVIDIA-Linux-x86_64-x file you downloaded form nvidia website.
3., Create the following file with the content:
/etc/modprobe.d/nvidia.conf
blacklist nvidiafb
blacklist nouveau

4., Run the following command:
update-initramfs -c -d -u
5., reboot VM-machine.
6., DONE.
7., If you upgrade/update your kernel ( new kernel version), you need to do the process again ( the driver needs to build for the new kernel too ).