NVIDIA GPU Operator with vSphere
- Dominant language
- Go
- Stars
- 2.1k
- Forks
- 328
- Avg merge
- 1d 4h
- Merged PRs (30d)
- 9
Description
Hello!
I saw the document about the [GPU Operator](https://anywhere.eks.amazonaws.com/docs/workloadmgmt/using-gpus/) but it only points to using it with bare metal deployment.
We are experimenting with EKS-A on vSphere. Is it possible to somehow configure the workload VMs from the `VSphereMachineConfig` bind the PCIe device?
I know I can probably shutdown the VM, edit the settings manually and add the PCI device thru passthru, but it would be great if we wouldn't require to do this manual work and could manage it thru the YAML file.
Thanks!
Contributor guide
Research direction
Start with the linked GPU Operator documentation and the VSphereMachineConfig configuration referenced in the issue. Determine whether PCIe passthrough for workload VMs is supported or can be represented there, and identify the relevant configuration or implementation entry points. Done means vSphere GPU setup is either supported through YAML without manual VM edits or the limitation and required documentation are clearly recorded.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- kubernetes
- Domain
- infrastructure
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100