Skip to content

Conversation

@ldematte
Copy link
Contributor

@ldematte ldematte commented Aug 20, 2025

This PR expands #132670 to account for GPU memory availability: a requesting thread can obtain a resource only if there are enough GPU physical resources available (in this first iteration, memory). Otherwise the requesting thread will be blocked and signalled again to re-check conditions are satisfied when memory is freed (when another thread release a resource).

Depends on rapidsai/cuvs#1267 (which is now merged into branch-25.10)

@ldematte ldematte marked this pull request as ready for review August 27, 2025 07:54
@elasticsearchmachine
Copy link
Collaborator

Pinging @elastic/es-search-relevance (Team:Search Relevance)

@elasticsearchmachine elasticsearchmachine added the Team:Search Relevance Meta label for the Search Relevance team in Elasticsearch label Aug 27, 2025
@ldematte ldematte added the test-gpu Run tests using a GPU label Sep 2, 2025
Copy link
Contributor

@ChrisHegarty ChrisHegarty left a comment

Choose a reason for hiding this comment

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

LGTM

@ldematte ldematte merged commit 83aa729 into elastic:es-gpu Sep 3, 2025
11 of 62 checks passed
@ldematte ldematte deleted the resource-manager-with-memory branch September 3, 2025 09:28
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

>non-issue :Search Relevance/Vectors Vector search Team:Search Relevance Meta label for the Search Relevance team in Elasticsearch test-gpu Run tests using a GPU

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants