Hello,
we want to install a Nvidia GPU in one of our three OpenNebula servers (exactly, server is a Dell PowerEdge R640, but GPU is not still purchased). I would like to know small details about " NVIDIA vGPU support" procedure I have read at https://docs.opennebula.io/6.8/open_cluster_deployment/kvm_node/vgpu.html?highlight=gpu#nvidia-vgpu-support (my servers are running OpenNebula 6.8.0-1 and 6.8.0.1-1).
After reading that, I have understood that:
- SR-IOV is mandatory. My server has this feature, so it seems this won’t be a problem
- IOMMU: after reading some documents in Internet, it seems that IOMMU is VT-d for Intel processors. That server has this feature enabled
- In NVIDIA vGPU support — OpenNebula 6.8.3 documentation, text says " Once everything is set up, you can follow these steps." and link sends you to PCI Passthrough — OpenNebula 6.8.3 documentation… in the middle of the page of https://docs.opennebula.io/6.8/open_cluster_deployment/kvm_node/pci_passthrough.html… so, my question is if I have to do step by step all procedure explained at that page or only from the link sends me, because before that lines, page explains more things about configuration.
Thanks.
link sends you to