Installer open-mpi avec Homebrew

Liens du projet

• Page d'accueil
• Dépôt

Commandes et alias

• mpiCC
• mpic++
• mpicc
• mpicxx
• mpiexec
• mpif77
• mpif90
• mpifort
• mpirun
• ompi_info
• opal_wrapper
• oshrun

Historique du projet

The project was formed as a consolidation of earlier MPI implementations rather than as a small single-codebase fork. The Open MPI documentation names LAM/MPI, LA-MPI, and FT-MPI as the three prior implementations merged into Open MPI, with the University of Stuttgart team joining through FT-MPI work in late 2004; the project FAQ also describes PACX-MPI contributions as part of the initial technology base.

That origin shaped the project's design goals: a clean new implementation, peer review, vendor and researcher participation, and a component architecture that lets network, scheduler, and platform support evolve without forcing every downstream user to fork the whole MPI stack.

Historique d'adoption

Open MPI grew into one of the standard MPI choices for Linux, macOS, BSD, and HPC software stacks because it provided a permissively licensed implementation with active involvement from academic labs, vendors, and users. The project's own FAQ emphasizes third-party components and commercial or closed-source plugin compatibility as intentional collaboration mechanisms.

Package managers expose Open MPI not only as a library but as a toolchain surface: Homebrew installs compiler wrappers such as mpicc and mpicxx alongside launch commands such as mpirun and mpiexec. That packaging model reflects how developers encounter MPI in practice: by compiling code through wrappers that inject the right headers, libraries, and runtime settings.

Modes d'utilisation

Developers use Open MPI to build MPI programs with wrapper compilers, launch multi-process jobs, inspect runtime configuration with ompi_info, and integrate jobs with cluster networking and process managers. Operators care about the runtime's transport and component selection because MPI performance depends heavily on the fabric, scheduler, host layout, and collective-communication choices.

On a laptop or CI machine Open MPI is often used to test parallel code paths at small process counts. On clusters it becomes part of the site software stack, where application teams expect stable mpirun/mpiexec behavior and admins tune the runtime for the interconnect and batch system.

Pourquoi les passionnés de paquets s'y intéressent

Open MPI is package-nerd interesting because a formula is not just shipping a library: it ships a parallel-programming ABI, compiler wrappers, launchers, runtime daemons, Fortran/C/C++ surfaces, and configuration files. Small packaging changes can affect build reproducibility, linked MPI ABI compatibility, and runtime behavior across many scientific applications.

Its history also explains why it has many moving parts. The project's component model and merger lineage were a deliberate answer to duplicated MPI implementation work, so packagers inherit a system designed to absorb many platforms and hardware transports behind one familiar MPI interface.

Chronologie

• 1994: MPI-1.0 was published by the MPI Forum, establishing the standard API family that Open MPI implements.
• Late 2004: the University of Stuttgart team became effectively part of the Open MPI effort through FT-MPI work, according to Open MPI's history page.
• After version 1.0: the project expanded beyond the founding implementation teams to broader contributor participation, according to the Open MPI FAQ.
• 2026-06-13: the Open MPI FAQ page used for this enrichment was last modified by the project.

Related projects

• LAM/MPI, LA-MPI, FT-MPI, and PACX-MPI are the important historical parents of Open MPI.
• The MPI Forum is the standards body whose MPI specifications define the programming model Open MPI implements.

Sources

• https://docs.open-mpi.org/en/main/history.html
• https://formulae.brew.sh/formula/open-mpi
• https://github.com/open-mpi/ompi
• https://www.mpi-forum.org/docs/
• https://www.open-mpi.org/faq/?category=general