As an admin, I would like PBS to autodiscover Xeon Phi and GPU devices on startup and configure the vnode, so that it is all taken care of when the mom report "free"

Description

When a cluster first has PBS installed it can require the admin to configure each node with the proper amount of nmics or ngpus per node. Admins may also want to create a separate numa node per socket to enable the scheduler to determine which socket the job is placed on. This will reduce the burden on the admin when they first configure PBS.

Acceptance Criteria

None

Assignee

Michael Karo

Reporter

Jon Shelley

Severity

None

OS

None

Start Date

None

Pull Request URL

Story Points

1

Epic Link

Fix versions

Priority

Low
Configure