FFTX documentation

FFTX is the exascale follow-on to the FFTW open-source package for executing the discrete Fast Fourier Transform, as well as higher-level operations composed of FFTs and linear operators. It is based on SPIRAL, a build-time code generator that produces high-performance kernels targeted to specific uses and platform environments.

FFTX is being developed as a cooperative effort between Lawrence Berkeley National Laboratory, Carnegie Mellon University, and SpiralGen, Inc.

Contents:

• Introduction
  • Motivation
  • Concept
  • License
  • Citation
  • Acknowledgments
• Installation with C++ interface
• FFTX Project
  • License
  • Prerequisites
  • Summary
  • 1. Install SPIRAL and associated packages
  • 2. Clone the FFTX repository.
  • 3. Generate library source code.
  • 4. Compile library source code and examples.
  • Building on Windows
  • FFTX Libraries Built
  • Library API
  • Linking Against FFTX Libraries
  • External Application Linking With FFTX
• C++ API
  • FFTX interfaces
  • fftx::point_t
  • fftx::box_t
  • fftx::array_t
  • fftx::global_ptr
  • FFTXProblem
• Python Package for FFTX
• FFTX Distributed 3D FFT Library
  • Scope of FFTX Distributed Library
  • API
  • Running the first example
  • Types of FFTs
  • Directions
  • Batch size
  • Processor Grid
  • Embedded
  • MPI Type
  • Packing
• Demo FFTX External Application
• Collaborations with other projects
  • Spinifel single-particle imaging
  • Plane Wave Distributed FFT
• Contributing to FFTX
  • FFTX Repository
  • Fork the Repository
  • Develop / Make Changes
  • Create the Pull Request (PR)
• FFTX Development Team
  • Lawrence Berkeley National Laboratory
  • Carnegie Mellon University
  • SpiralGen, Inc.
• Further Reading
  • FFTX and SpectralPACK
  • SPIRAL
  • FFTs
  • Distributed and Multicore FFTs
  • Interesting Target Hardware
  • Linear Algebra Libraries
  • Applications
  • 3D FFTs and Plane Wave Codes

Indices and tables

• Index

• Module Index

• Search Page