spb/tools-spboucher-ai

1{2  "slug": "open-mpi",3  "name": "Open MPI",4  "aliases": ["OpenMPI", "ompi"],5  "vendor": { "name": "Open MPI consortium", "slug": "open-mpi", "website": "https://www.open-mpi.org/" },6  "homepage_url": "https://www.open-mpi.org/",7  "docs_url": "https://docs.open-mpi.org/en/main/",8  "repo_url": "https://github.com/open-mpi/ompi",9  "license_url": "https://www.open-mpi.org/community/license.php",10  "product_type": "library",11  "product_status": "active",12  "summary": "Open-source implementation of the MPI message-passing standard used to run tightly coupled parallel programs across the nodes of a cluster",13  "description": "Open MPI is an open-source implementation of the Message Passing Interface (MPI) specification, developed and maintained by a consortium of academic, research and industry partners under a 3-clause BSD license variant. MPI is the standard programming model of high-performance computing: processes on different nodes exchange messages and collectives (broadcast, reduce, all-gather) through a library API, and Open MPI provides that library for C, C++ and Fortran programs together with the `mpirun` launcher (built on PRRTE and PMIx) that starts processes across hosts over SSH or under schedulers such as Slurm, PBS and LSF.\n\nIt supports Linux and macOS, many network transports from TCP to InfiniBand, and is packaged by Homebrew and most Linux distributions. On a cluster of Macs it is the usual way to launch MPI-style jobs, and MLX's distributed backend can use MPI as one of its communication options.\n\nIt is a low-level tool: there is no scheduling, fault tolerance is limited, and applications must be written explicitly for message passing, which makes it most relevant to scientific codes rather than general data processing.",14  "capabilities": ["MPI point-to-point and collective communication", "mpirun launcher across hosts via SSH or batch schedulers", "Support for many network transports", "C, C++ and Fortran bindings", "Runtime tuning through MCA parameters"],15  "workflows": ["Run a Fortran or C simulation across several nodes", "Launch MLX distributed jobs with MPI on a Mac cluster", "Integrate MPI jobs into a Slurm allocation"],16  "advantages": ["Standards-based and portable across HPC systems", "Permissive BSD license", "Widely packaged and documented"],17  "limitations": ["Applications must be written for message passing", "No built-in scheduling or elasticity", "Limited fault tolerance for node failures"],18  "prerequisites": ["C/C++/Fortran toolchain", "Passwordless SSH or a scheduler between nodes"],19  "audiences": ["Scientific and HPC developers", "Operators of small compute clusters"],20  "alternatives": ["ray", "dask", "slurm"],21  "categories": [{ "slug": "distributed-computing", "primary": true }],22  "tags": ["hpc", "mpi", "distributed-computing", "open-source", "cli"],23  "use_cases": ["distribute-computation"],24  "operating_systems": ["linux", "macos"],25  "execution_modes": ["local", "self-hosted"],26  "interfaces": ["library", "cli"],27  "deployment_models": ["on-premises"],28  "integrations": ["Slurm", "PBS", "LSF", "PMIx", "Homebrew", "MLX distributed"],29  "data_formats": [],30  "languages": ["C", "C++", "Fortran"],31  "source_model": "open-source",32  "license_spdx": "BSD-3-Clause-Open-MPI",33  "pricing_model": ["open-source", "free"],34  "self_hostable": true,35  "sources": [36    { "url": "https://www.open-mpi.org/", "type": "official_site", "title": "Open MPI: Open Source High Performance Computing", "supports": ["name", "summary"], "checked_at": "2026-11-02" },37    { "url": "https://docs.open-mpi.org/en/main/", "type": "official_docs", "title": "Open MPI documentation", "supports": ["capabilities", "integrations", "vendor"], "checked_at": "2026-11-02" },38    { "url": "https://www.open-mpi.org/community/license.php", "type": "license_page", "title": "Open MPI License", "supports": ["license_spdx", "source_model"], "checked_at": "2026-11-02" }39  ],40  "confidence": "high",41  "discovery_method": "editorial-research",42  "depth": "standard",43  "editorial_status": "reviewed",44  "visibility": "public",45  "featured": false,46  "author": "batch-C"47}48