---
id: mpich
version: "5.0.1.post1"
license: LicenseRef-MPICH
license_treatment: unclear
maintenance: active
---

# mpich — A high performance implementation of MPI

License: unclear · Maintenance: active · Downloads: 91.1K/mo

## What it is and what it does

MPICH is a production-grade implementation of the Message Passing Interface standard, widely deployed on HPC systems and supported by vendors including Intel and HPE. It provides the mpicc compiler wrapper, mpiexec process manager, and core MPI runtime needed to build and run parallel applications across single or multiple nodes. The package comes with the Hydra process manager by default and supports multiple network modules (OFI, UCX) for different interconnect hardware.

Typical use involves configuring MPICH with a chosen device and network module, installing it to a shared filesystem path visible on all compute nodes, then using mpicc to compile parallel C/C++/Fortran code and mpiexec to launch jobs across the cluster. It is not a Python library in the traditional sense—it is a system-level MPI runtime that Python applications can call into via bindings.

Use it for:

- Compile and run parallel C/C++/Fortran applications on HPC clusters using standard MPI calls.
- Set up a multi-node compute environment where applications communicate via message passing.
- Deploy on Top 500 HPC systems or vendor-supported clusters (Intel, HPE, ParTec infrastructure).
- Develop and test parallel algorithms that require inter-process communication across distributed nodes.
- Integrate with existing HPC workflows that expect mpicc, mpiexec, and standard MPI headers.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides a high-performance Message Passing Interface (MPI) implementation for parallel computing applications, with support for multiple network modules and process management.

Yes, if you are building or running parallel HPC applications on a cluster and need a widely-adopted, vendor-supported MPI implementation. No, if you are looking for a Python library for parallel computing—this is a system runtime, not a Python package. Yes-with-conditions if you are unfamiliar with MPI or HPC: expect to configure and install to a shared filesystem, understand process management, and have appropriate compilers available.

## Install

pip install mpich
uv add mpich
poetry add mpich

## Installing mpich

Before you install: Medium install friction due to platform-specific wheels (macOS arm64/x86_64, Linux aarch64/x86_64) and no runtime Python dependencies, but requires system-level C compiler and optional Fortran/C++ compilers depending on use case.

License in practice: License treatment is unclear—the package uses a custom LicenseRef-MPICH identifier with no SPDX mapping. Review the project's actual license terms before adopting in proprietary or restricted-distribution contexts.

Quickstart:

# After installation, add to PATH:
export PATH=/path/to/mpich-install/bin:$PATH

# Compile an MPI program:
mpicc -o myapp myapp.c

# Run on local machine with n processes:
mpiexec -n 4 ./myapp

Requires a C99-compliant C compiler; optional Fortran and C++ compilers if building Fortran or C++ applications. Installation directory must be at the same path on all nodes in a cluster (typically on shared NFS).

Verify before relying:

- Whether the PyPI wheel includes pre-built MPICH binaries or requires local compilation.
- Python version requirements or compatibility constraints not specified in the fact sheet.
- Whether the package works with Windows or only Unix-like systems.

## Package facts

- License: LicenseRef-MPICH (unclear)
- Python support: unspecified
- Install friction: medium
- Maintenance: active
- Downloads: 91.1K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

MPI message passing interface, parallel computing framework, distributed computing library, high-performance computing MPI, process communication parallel, HPC cluster computing, multi-node application launcher, hpc, parallel-computing, message-passing

[View on SkillFed](https://skillfed.io/packages/mpich) · [View on PyPI](https://pypi.org/project/mpich/)