Skip to content

Extend pipeline endpoint with option to change inference device #5042

@leoll2

Description

@leoll2

Endpoint: PATCH /api/projects/{project_id}/pipeline

New body parameter: device

Allowed values: <target>[-<index>] (the index part is optional, default 0)
Examples:

  • cpu -> device from torch.cpu
  • xpu -> default device from torch.xpu
  • cuda -> default device from torch.cuda
  • xpu-2 -> device with index 2 in torch.xpu
  • cuda-1 -> device with index 1 in torch.cuda

The endpoint should return 409 Conflict is the device is not available.

Metadata

Metadata

Assignees

Labels

Geti Tune BackendIssues related to Geti Tune Studio backend

Type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions