NVIDIA / NVIDIA/cuCollections

[FEA] Support concurrent insert/find for static_map when size of K-V pair is greater than 8 bytes

Open
#137 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

helps: rapids type: feature request
Dominant language
Cuda
Stars
667
Forks
120
Avg merge
7d 5h
Merged PRs (30d)
4

Description

An optimized implementation of string renumbering in cuGraph requires building histogram with metadata along with frequency as the payload. The metadata is required for optimal performance of subsequent operations in the renumbering implementation. CUDA kernel concurrently finds/inserts the data in HT.

Since CuCollection currently doesn't support concurrent insert/find for more than 8 byte KV pair size, we currently cuDF concurrent_unordered_map with some manual checking to check update of payload values.

It would be great if CuCo supports concurrent find/insert on complex payloads for static_map.

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 reading the static_map implementation and its current concurrent find/insert behavior, then compare it with the cuDF concurrent_unordered_map usage described in the issue. The work is complete when static_map supports concurrent find/insert operations for key-value pairs larger than 8 bytes, including complex payloads needed by the cuGraph renumbering use case.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
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.