[
https://issues.apache.org/jira/browse/YUNIKORN-743?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17379501#comment-17379501
]
Wilfred Spiegelenburg commented on YUNIKORN-743:
------------------------------------------------
YuniKorn supports a generic set of resources already. You can define a resource
like {{nvidia.com/gpu: 1}} on a pod or node. It will be treated like any other
resource (cpu, memory or other custom resource) that is defined.
The other part of the description given in the document uses labels and node
selectors. That is supported already in the scheduler. Node labels and
selectors use the standard predicates from Kubernetes.
All other changes noted in the doc are node install changes of the utilities or
changes to the image that is run. Neither of those changes are linked to the
scheduler.
If you see any other part that is not supported from a scheduler side let me
know but there should be nothing left.
> Support GPU scheduling
> ----------------------
>
> Key: YUNIKORN-743
> URL: https://issues.apache.org/jira/browse/YUNIKORN-743
> Project: Apache YuniKorn
> Issue Type: New Feature
> Components: shim - kubernetes
> Reporter: Chaoran Yu
> Priority: Major
>
> This is to verify the support for GPU scheduling in YuniKorn. Identify and
> bridge gaps with [GPU support offered by the k8s default
> scheduler|https://kubernetes.io/docs/tasks/manage-gpus/scheduling-gpus/].
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
