Support to vGPU

Community Support General
1

Hello everybody,

I have a little clúster with ONE 6.6 and I would like to know if the support that ONE gives to vGPUs is through VM passthrough finally. I have read:
https://docs.opennebula.io/6.8/open_cluster_deployment/kvm_node/vgpu.html
but in the last item “Using vGPU” the documentation jumps to passthrough configuration.

One question is I would like to know if anybody has any experience in using vGPUs in this way as documentation says working with big datasets through VM with vGPU attached.

Other question i how does the performance is?. I know that it depends on very other requirements as a good and quick storage and network but I would like to read any experience about.

Many thanks in advanced and best regards.

2

Hi @jj197005 :wave:

OpenNebula supports vGPU since v6.4. Once your GPU is configured to use vGPU (in case it supports it), you can use the vGPU as any PCI device (as you would do for example with any normal physical GPU). That’s why the documentation redirects you to the PCI devices section.

Best,
Victor.

3

Hello Victor,

Many thanks for your reply. I would like to ask you if you have any experience with GPU nodes and vGPU installed about if the performance is good enough to be applied to datasets used in IA.

Many thanks in advanced and best regards,