NVIDIA cuFFTMp documentation¶

Welcome to the cuFFTMp (cuFFT Multi-process) library.

You can find here:

A Quick start guide

A How to use cuFFTMp section, describing the requirements and general usage of cuFFTMp

An API reference section, with a comprehensive description of all of cuFFTMp’s APIs

cuFFTMp is distributed as part of the NVIDIA HPC-SDK.

Highlights¶

2D and 3D distributed-memory FFTs
Slabs (1D) and pencils (2D) data decomposition, with arbitrary block sizes
MPI-compatible interface
Low-latency implementation using NVSHMEM, optimized for single-node and multi-node FFTs
x86_64 and aarch64 support (see Hardware and software requirements)