NVIDIA / NVIDIA/cuvs

[FEA] Improve perf of batched all_neighbors

Open
#1,333 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

For large datasets, batched all_neighbors takes a long time to preprocess the data into clusters.
Currently we are running a brute-force search to assign data points to its nearest cluster centroids.

https://github.com/rapidsai/cuvs/blob/9667a576a3a534a44fec24d7c23d4926eb6a5173/cpp/src/neighbors/all_neighbors/all_neighbors_batched.cuh#L101-L148

This process can be optimized by using a non brute-force approach.

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

Read cpp/src/neighbors/all_neighbors/all_neighbors_batched.cuh around lines 101-148, where batched all_neighbors preprocesses data into clusters and assigns points to their nearest centroids. First understand the current brute-force assignment and measure preprocessing on a large dataset. Done means the assignment uses a non-brute-force approach and batched all_neighbors preprocessing is measurably faster.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
machine-learning, performance, search
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.