Sharing nVidia GPU

anto84

New Member
Mar 20, 2025
5
0
1
I have a server with one GPU, and I need to run two vms that both have gpu.

Is it possible to do sharing of one nvidia gpu, between two vms, or is only Passthrough allowed?
 
CPU sharing
You probably mean GPU sharing. To which leesteken responded accordingly.

vGPU is a form of GPU partitioning to enable "sharing" one GPU amongst more than 1 guest.

Read up on the subject to understand it better.

I do believe that you won't be able to use the physical output device/s of the GPU for both VM's at the same time with vGPU technology - since vGPU basically splits up the CPU time slices of the GPU - so it is good for sharing the "processing power" of the GPU - not the physical output device. But vGPU is an evolving technology - so you will need to first check exactly what you want to achieve with these 2 VMs sharing a GPU.
 
Last edited:
  • Like
Reactions: leesteken
You probably mean GPU sharing. To which leesteken responded accordingly.

vGPU is a form of GPU partitioning to enable "sharing" one GPU amongst more than 1 guest.

Read up on the subject to understand it better.

I do believe that you won't be able to use the physical output device/s of the GPU for both VM's at the same time with vGPU technology - since vGPU basically splits up the CPU time slices of the GPU - so it is good for sharing the "processing power" of the GPU - not the physical output device. But vGPU is an evolving technology - so you will need to first check exactly what you want to achieve with these 2 VMs sharing a GPU.
sorry my mistake you also from the title of the post, I meant GPU sharing
 
sorry my mistake you also from the title of the post, I meant GPU sharing
Exactly, like you do for vmware, I was asking if it is possible to share the resources of a Single physical GPU, between two VMs, without having to undergo Passthrough, dedicating the resource only to one VM
 
Exactly, like you do for vmware, I was asking if it is possible to share the resources of a Single physical GPU, between two VMs, without having to undergo Passthrough, dedicating the resource only to one VM
Yes it is possible with NVidia server GPUs: https://docs.nvidia.com/vgpu/latest/index.html . That is why I posted the link to the Proxmox Wikl that explains how to set it up.
If you need only OpenGL support, you could look into VirGL/virtio-gl: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#qm_display
 
Well, first off, you didn't say which NVidia card you had, and I do mean "card" and not GPU and here's why. There are cards that have multiple GPU's on them. Those can be split up to multiple VM's. Let's say its a 4 GPU card ($5k-$10k), then 2 VM's can each get 2 GPU's, or 1 VM gets 3 GPU's the other VM gets only one GPU. You did not mention which card you have. If its a consumer grade "gamer" card it will only have one GPU (albeit really fast still only 1) and can only be assigned to one VM.
 
Well, first off, you didn't say which NVidia card you had, and I do mean "card" and not GPU and here's why. There are cards that have multiple GPU's on them. Those can be split up to multiple VM's. Let's say its a 4 GPU card ($5k-$10k), then 2 VM's can each get 2 GPU's, or 1 VM gets 3 GPU's the other VM gets only one GPU. You did not mention which card you have. If its a consumer grade "gamer" card it will only have one GPU (albeit really fast still only 1) and can only be assigned to one VM.
I have nVidia A30, but i don't find nothig about this card, for compatibility