GitHub

HealpixMPI.jl: an MPI-parallel implementation of the HEALPix tessellation scheme in Julia

Welcome to HealpixMPI.jl, an MPI-parallel implementation of Healpix.jl, an Healpix spherical tessellation scheme written entirely in Julia.

This package constitutes a natural extension of the package Healpix.jl, providing an MPI integration of the main functionalities, allowing for high-performances and better scaling on high resolutions.

More specifically, three main features can be highlighted:

  • High-performance spherical harmonic transforms. This library relies on ducc's state-of-the-art algorithms for performing fast and efficient SHTs.
  • Massively parallelization. The simultaneous usage of modern C++ multithreading (provided by ducc, for shared-memory parallelization) and MPI (for distributed-memory parallelization) allows the code to be run in parallel on a large number of nodes. The code has currently been tested and benchmarked with performance improvements up to 1024 cores. Note that this would be in practice impossible to achieve without the usage of MPI, which allows to distribute maps and harmonic coefficients over different computing nodes, since generally such a high number of computing cores is never available on the same machine.
  • Cross-platform support. This package maintains the same multi-platform compatibility of Healpix.jl, thanks to the cross-platform support of MPI.jl, Ducc0.jl (ducc's wrapper package providing Julia interface) and Julia itself.

Documentation

The documentation was built using Documenter.jl.

Documentation built on 2024-05-20T13:24:38.369 using Julia 1.6.7.

Index