Mutiple VGPU types per compute node
For the moment, it's only possible to support only one GPU type per compute service. Operators can support multiple types if they use separate GPU types per compute node and then use aggregates for grouping per type, but they would ideally also like to have a compute service supporting two or more different types if they have more than one physical GPU.
Blueprint information
- Status:
- Complete
- Approver:
- None
- Priority:
- Undefined
- Drafter:
- Sylvain Bauza
- Direction:
- Approved
- Assignee:
- Sylvain Bauza
- Definition:
- Approved
- Series goal:
- Accepted for ussuri
- Implementation:
- Implemented
- Milestone target:
- ussuri-3
- Started by
- Sylvain Bauza
- Completed by
- Sylvain Bauza
Related branches
Related bugs
Sprints
Whiteboard
Gerrit topic: https:/
Addressed by: https:/
Re-proposes multiple vGPU types in libvirt
[efried 20200130] Definition:Approved as spec merged. Target for Ussuri.
[efried 20200220] Agreed in the Nova meeting to Direction:Approve all Definition:Approved blueprints http://
Addressed by: https:/
Provide the parent pGPU when creating a new vGPU
Addressed by: https:/
WIP: Support different vGPU types per pGPU
Addressed by: https:/
DNM: Functional test with pGPUs
Addressed by: https:/
FUP: Support multiple vGPU types