NVIDIA / NVIDIA/cuvs

[FEA] Improve support of bitset data type for prefiltering

Open
#539 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

Currently, cuVS is using Raft Bitset to do pre-filtering. Only uint32_t data type is supported for Milvus index, but client might have created their bitset using another data type. For example Milvus create their bitset using uint8_t. This changes the way the bits are ordered.
A solution would be to remove the bitset data type template to always use our preferred data type, and accept any bitset by specifying the data type that was used to create it.

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 by tracing cuVS pre-filtering and its use of the Raft Bitset, then inspect how the current uint32_t template choice reaches the Milvus index interface. Compare the bit ordering produced by uint32_t and uint8_t inputs and define the supported input-type interface before implementing changes; add or update coverage for both representations.

Written by the indexing model from the issue text.

Assessment

Domain
search
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.