Skip to main content
Ctrl+K
rapidsmpf  documentation - Home rapidsmpf  documentation - Home
  • Quickstart
  • Background
  • API Reference
  • GitHub
  • Twitter
Home
rapidsmpf
cucimcudf-javacudfcugraphcumlcuprojcuspatialcuvscuxfilterdask-cudadask-cudfkvikiolibcudflibcumllibcuprojlibcuspatiallibkvikiolibrmmlibucxxraftrapids-cmakerapidsmpfrmm
nightly (25.10)
nightly (25.10)stable (25.08)
  • Quickstart
  • Background
  • API Reference
  • GitHub
  • Twitter

RapidsMPF documentation#

rapidsmpf provides a collection of multi-GPU, distributed memory algorithms.

Follow the RAPIDS installation guide to install rapidsmpf and the necessary dependencies.

Contents:

  • Quickstart
    • Dask-cuDF Example
  • Background
    • Architecture
      • Shuffle
      • Example: Shuffle a Table on a Column
    • Shuffle Statistics
  • API Reference
    • Integrations
      • Generic
      • Dask
      • Single-process
      • Ray
      • cuDF
    • Shuffler
      • Shuffler
    • Communicator
      • COMMUNICATORS
      • Communicator
      • LOG_LEVEL
      • Logger
      • MPI Communicator
      • UCXX Communicator
    • Buffer
      • MemoryType
      • BufferResource
      • LimitAvailableMemory
      • PackedData
    • Config Options
      • Optional
      • OptionalBytes
      • Options
      • get_environment_variables()
    • Statistics
      • MemoryRecord
      • MemoryRecorder
      • Statistics
    • RMM Resource Adaptor
      • AllocType
      • RmmResourceAdaptor
      • ScopedMemoryRecord

next

Quickstart

This Page

  • Show Source

© Copyright 2025, NVIDIA Corporation.

Created using Sphinx 8.3.0.

Built with the PyData Sphinx Theme 0.16.1.