Using OpenStack support for GPU virtualization

[motzzu_2025]

Hi Everyone,

I am planning to purchase an HPE DL380 Gen11 CTO Chassis with dual Intel Xeon Platinum 8462Y+ (2.8GHz, 32-core) processors, ample NVMe storage, and DDR5 memory. Additionally, I will be using the NVIDIA L4 24GB PCIe Accelerator for HPE and would like to leverage GPU virtualization in OpenStack.

I am looking for guidance on setting up OpenStack GPU virtualization with this configuration. Specifically:

• Does the NVIDIA L4 support OpenStack GPU virtualization, including SR-IOV or vGPU configurations?

• What are the best practices for configuring PCI passthrough or vGPU in OpenStack with this setup?

• Are there any known compatibility issues or additional driver/licensing requirements I should be aware of?

I would appreciate any insights or recommendations from those who have deployed a similar setup.

Thanks in advance!

Motzzu

[Rakesh0404]

Hi Motzzu,
Good day! Please find the answeres for the query. 

Does the NVIDIA L4 support OpenStack GPU virtualization, including SR-IOV or vGPU configurations?

The SR-IOV option is available. 

What are the best practices for configuring PCI passthrough or vGPU in OpenStack with this setup?

SR-IOV (PCI Passthrough):

1. Enable SR-IOV in BIOS: Ensure that SR-IOV and VT-d are enabled in your server's BIOS.
2. Install NVIDIA Drivers: Install the correct NVIDIA driver that supports SR-IOV.
3. Involve the correct team for the OPEN Stack, as we are not aware of the whole settings or the configuration. 

Are there any known compatibility issues or additional driver/licensing requirements I should be aware of?
For vGPU configurations, ensure you have the correct NVIDIA GRID vGPU license for your setup. Without this, the vGPU feature won't be fully functional.

SR-IOV doesn’t require special licensing for passthrough, but each VM using GPU-intensive applications may require licenses for those applications (e.g., CUDA).

[support_s]

Hello,

Let us know if you were able to resolve the issue.

If you are satisfied with the answers then kindly click the "Accept As Solution" button for the most helpful response so that it is beneficial to all community members.

Please click on "Thumbs Up/Kudo" icon to give a "Kudo".