Skip to content

Gpu: Allow/denylist support to ignore devices (e.g. integrated) #2101

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

tkatila
Copy link
Contributor

@tkatila tkatila commented Aug 9, 2025

Fixes #2087

@tkatila
Copy link
Contributor Author

tkatila commented Aug 9, 2025

Depends on the removal PR to be merged first.

@tkatila tkatila changed the title Gpu: Whitelist support to ignore devices (e.g. integrated) Gpu: Allowlist support to ignore devices (e.g. integrated) Aug 11, 2025
@tkatila tkatila marked this pull request as ready for review August 12, 2025 16:55
Copy link
Contributor

@eero-t eero-t left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks OK, but I have few suggestions.

@tkatila tkatila changed the title Gpu: Allowlist support to ignore devices (e.g. integrated) Gpu: Allow/denylist support to ignore devices (e.g. integrated) Aug 14, 2025
Copy link
Contributor

@eero-t eero-t left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved, but I think the added 2 member bools are redundant, unless you intend to use them later also some other function.

Copy link
Contributor

@mythi mythi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, one suggestion

@mythi
Copy link
Contributor

mythi commented Aug 15, 2025

Fixes #1815 too?

@tkatila
Copy link
Contributor Author

tkatila commented Aug 15, 2025

Fixes #1815 too?

Sort of. #1815 is about how to control scheduling of workloads to different GPUs, e.g. lighter workload to integrated GPU, larger workload to a dedicated one. That could be achieved with different resource names per GPU type. This PR can help, if there's no need for the integrated GPU and it's just causing confusion with workloads being scheduled for it.

By defining allowed/denied PCI IDs, it's possible to only
select some GPUs per host. For example, on a desktop with
integrated and discrete graphics, GPU plugin can only
register the discrete one.

Signed-off-by: Tuomas Katila <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Restrict certain devices from being discovered and reported by the plugin
3 participants