NVIDIA / NVIDIA/raft

[FEA] Move PartDescriptor & Data abstractions from cuML

Open
#10 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement inactive-30d inactive-90d
Dominant language
Cuda
Stars
1k
Forks
251
Avg merge
1d 8h
Merged PRs (30d)
31

Description

These abstractions have been used in the Dask packages of cuML to enable the representation of Dask partitions as input to our multi-node multi-GPU algorithms w/ the communicator. It would be useful for these abstractions to be placed in RAFT in order to create a more unified API for the inputs to primitives.

It would also be nice to include some Cython code for constructing and managing instances of PartDescriptor in order to ease the boilerplate needed and reduce duplication across primitives.

Contributor guide

No contributing guide indexed for this repository

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

No files or tests are named. Start by locating the PartDescriptor and Data abstractions in cuML's Dask packages, then inspect how RAFT primitives represent inputs and use the communicator. Define the scope for moving the abstractions and adding Cython construction and management, then verify affected multi-node, multi-GPU primitives use the unified API.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
distributed-systems, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.