NVIDIA / NVIDIA/cuvs

[FEA] Pairwise distances to support int64_t internally (and downcast as necessary).

Open
#95 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

feature request
Dominant language
Cuda
Stars
854
Forks
236
Avg merge
3d 3h
Merged PRs (30d)
62

Description

While migrating the pairwise distance API from RAFT, we realized that it's still accepting mdspans w/ indexing type of uint32_t. In order to keep our public facing APIs consistent in the meantime, we're accepting mdspan with int64_t index type on the outer API layer for now and then constructing new mdspans with the uint32_t indexing type for the call to the pairwise distances underneath.

This is a short-term fix and we should revisit this sometime in the near(ish) future to support int64_t efficiently in the pairwise distances layer itself.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start at the pairwise distances API and its underlying pairwise distances layer; inspect how the outer mdspans are converted from int64_t indexing to uint32_t. Done means the API accepts int64_t-indexed mdspans while the underlying call receives the required uint32_t-indexed views, with downcasting handled as necessary.

Written by the indexing model from the issue text.

Assessment

Domain
backend
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.