[FEA] Support building all-neighbors graphs on disk
Open
@jinsolp is already working on this.
Since Apr 25, 2025.
feature request
- Dominant language
- Cuda
- Stars
- 854
- Forks
- 236
- Avg merge
- 3d 3h
- Merged PRs (30d)
- 62
Description
As the title suggests, the batching algorithm which was recently added to nn-descent enables partitioned, scalable and multi-gpu builds for knn graphs that fit into RAM. We need to enable this to also use the disk. This could be useful for the CAGRA algorithm, in addition to UMAP, TSNE, HDBSCAN, and many others.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.