High Performance Computing for Mac OS

Purpose

A practical Mac scientific-computing shelf

This site is intentionally small. It keeps the resources that still matter for scientific macOS users and removes old project listings that no longer help someone get work done.

Last content review: July 2026.

Mac HPC stack

Compiler, MPI, math libraries, solvers, and research code

A useful scientific Mac setup usually starts with a consistent compiler and MPI stack, then adds numerical libraries, solver frameworks, interactive tools, and larger research codes. The notes below prefer standalone Mac binaries where credible ones exist and fall back to package managers or source builds when they do not.

Toolchain: GCC, GFortran, SDK paths, and runtime libraries
Parallel jobs: Open MPI wrappers for local testing and Cactus builds
Fourier transforms: FFTW libraries for serial, threaded, OpenMP, and MPI codes
Solvers: PETSc pointers for scalable linear, nonlinear, and time-dependent problems
Analysis: Octave for interactive numerical work and plotting
Research codes: Cactus Code, Einstein Toolkit, and GRAVSIM pointers

Resource map

Start with the tool you need

Compilers

GCC and GFortran

Prebuilt GNU compiler tarballs for scientific users who want a simple toolchain under /usr/local.

View compiler builds

MPI

Open MPI

Prebuilt Open MPI for the HPC GCC/GFortran stack, plus package-manager options for newer downstream builds.

View MPI builds

Fourier transforms

FFTW

Standalone Apple Silicon FFTW package with serial, threaded, OpenMP, and MPI libraries for C and Fortran codes.

View FFTW build

Solver toolkit

PETSc

Mac-specific install pointers for PETSc, including real and complex-scalar package-manager builds.

View PETSc notes

Numerical analysis

Octave

Standalone app-bundle option where it helps, plus current macOS package-manager guidance.

View Octave notes

Relativity

Cactus Code

General-purpose Cactus computational toolkit links, with Einstein Toolkit workstation pointers.

View Cactus links

N-body code

GRAVSIM

A preserved gravitational N-body package for studying algorithmic, vector, and multiprocessor optimization.

View GRAVSIM

Core resource

GCC and GFortran for macOS

All compiler files on SourceForge

Latest listed build

GCC 16.1 and GFortran 16.1

Current Apple Silicon build from the HPC project files. These tarballs install under /usr/local and are intended for users who want a straightforward GNU C, C++, and Fortran toolchain outside a package manager.

GCC 16.1 tarball GFortran 16.1 tarball

Basic install

Install Apple's command-line tools first, then extract the selected tarball from your downloads folder.

xcode-select --install
cd ~/Downloads
gunzip gcc-16.1-mx-bin.tar.gz
sudo tar -xvf gcc-16.1-mx-bin.tar -C /
gfortran --version

If your browser already expanded the archive, start with the .tar file and adjust the file name.

SDK paths on newer macOS releases

Some builds on Catalina and newer macOS releases may need explicit SDK include and library paths.

-I/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include \
-L/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/lib

Use these flags only when the compiler cannot find system headers or libraries.

Legacy compiler builds for older macOS systems

Build	macOS target noted on old site	Updated	Downloads
GCC/GFortran 15.2	Tahoe - Apple Silicon	Jan 2026	GCC GFortran
GCC/GFortran 14.1	Sonoma, Sequoia - Apple Silicon	June 2024	GCC GFortran
GCC/GFortran 13.2	Sonoma - M1/M2 Apple Silicon	Jan 2024	GCC GFortran
GCC/GFortran 13 M2	Ventura - M2 Apple Silicon	May 2023	GCC GFortran
GCC/GFortran 13 M1	Ventura - M1 Apple Silicon, experimental	Jan 2023	GCC GFortran
GCC/GFortran 12.1	Monterey - M1 Apple Silicon	June 2022	GCC GFortran
GCC/GFortran 11.2	Monterey	Nov 2021	GCC GFortran
GCC/GFortran M1	Big Sur - M1 Apple Silicon	Dec 2020	GCC GFortran
GCC/GFortran 10.2	Big Sur	Nov 2020	GCC GFortran
GCC/GFortran 9.2	Catalina	Oct 2019	GCC GFortran
GCC/GFortran 8.3	Mojave, Catalina	April 2019	GCC GFortran
GCC/GFortran 8.1	High Sierra, Mojave	June 2018	GCC GFortran
GCC/GFortran 7.3	High Sierra	June 2018	GCC GFortran
GCC/GFortran 7.1	El Capitan, Sierra	June 2017	GCC GFortran
GCC/GFortran 5.1	Yosemite, El Capitan	June 2015	GCC GFortran
GCC/GFortran 4.9	Mavericks, Yosemite	Nov 2014	GCC GFortran
GCC/GFortran 4.8	Mountain Lion, Mavericks	Oct 2013	GCC GFortran
GCC/GFortran 4.7	Lion, Mountain Lion	July 2012	GCC GFortran

Upstream documentation: GCC project and GCC online documentation.

Parallel computing

Open MPI on macOS

Open MPI provides the mpicc, mpicxx, mpifort, and mpirun tools used by many C, C++, and Fortran HPC codes. The HPC project now provides a standalone Apple Silicon Open MPI build for users who want the same /usr/local layout as the GCC and GFortran tarballs.

Latest listed build

Open MPI 5.0.10

Built with /usr/local/bin/gcc, g++, and gfortran from the HPC GCC/GFortran 16.1 stack. Install those compiler tarballs first if this is a fresh Mac setup.

Open MPI 5.0.10 tarball

Basic install and quick checks

xcode-select --install
cd ~/Downloads
gunzip openmpi-5.0.10-mx-bin.tar.gz
sudo tar -xvf openmpi-5.0.10-mx-bin.tar -C /
mpicc --version
mpifort --version
mpirun -np 4 hostname

Keep the MPI wrappers, C/C++ compiler, and Fortran runtime from the same /usr/local toolchain family when building Cactus or other mixed-language codes.

All MPI files on SourceForge Official source and binary-package note Open MPI source downloads Homebrew Open MPI MacPorts Open MPI

Fourier transforms

FFTW on macOS

FFTW is the standard C and Fortran FFT library behind many spectral methods, signal processing workflows, and Fourier-space solvers. The HPC project provides an Apple Silicon standalone package for users who want an install layout that matches the /usr/local GCC, GFortran, and Open MPI tarballs.

Latest listed build

FFTW 3.3.11

Built on the Mac Studio with the HPC GCC/GFortran 16.1 stack and Open MPI 5.0.10. Includes double, single, and long-double libraries with pthreads, OpenMP, MPI, CMake, pkg-config, man pages, info docs, and upstream license files.

FFTW 3.3.11 tarball

Basic install and quick checks

xcode-select --install
cd ~/Downloads
gunzip fftw-3.3.11-mx-bin.tar.gz
sudo tar -xvf fftw-3.3.11-mx-bin.tar -C /
gcc test.c -lfftw3 -lm
mpicc mpi-test.c -lfftw3_mpi -lfftw3 -lm

Serial FFTW can be used on its own. Install the matching Open MPI package first when building against the FFTW MPI interfaces.

All FFTW files on SourceForge FFTW project FFTW source downloads FFTW documentation Homebrew FFTW MacPorts FFTW

Solver toolkit

PETSc on macOS

PETSc provides scalable data structures and solvers for linear, nonlinear, and time-dependent scientific applications. A single frozen Mac binary is less useful here because PETSc builds are often tied to choices such as real versus complex scalars, MPI, BLAS/LAPACK, HDF5, Hypre, MUMPS, SuiteSparse, and petsc4py.

Mac install paths

Real scalars: Homebrew's petsc formula is the simplest current Apple Silicon route for many C, C++, and Fortran users.
Complex scalars: use Homebrew's petsc-complex formula when the application expects complex-valued PETSc.
Variant control: use PETSc's official source install when you need a particular MPI, external package, precision, or Python stack.

brew install petsc
brew install petsc-complex

Keep PETSc, MPI, compiler, and dependent solver libraries from one coherent stack whenever possible.

PETSc documentation Official PETSc install guide Homebrew PETSc Homebrew PETSc complex MacPorts PETSc petsc4py documentation

Numerical analysis

GNU Octave on macOS

Octave is the open-source, Matlab-like environment many researchers use for numerical prototyping, plotting, and exploratory analysis. For users who want a standalone Mac application, Octave.app is the best current DMG route: its 9.2 release is much newer than the old MacOSX binary archives. No maintained Octave 10 or 11 standalone Mac DMG was found in the July 2026 review, so use Homebrew or MacPorts when you need the newest upstream release.

Mac install paths

Standalone app: try the Octave.app 9.2 DMGs first if that release is current enough for your work.
Recent binaries: Octave.app provides newer Apple Silicon and Intel app bundles than the old legacy archives.
Newest Octave: use Homebrew or MacPorts when you need the current upstream release instead of a standalone app.
Apple Silicon: Homebrew installs command-line tools under /opt/homebrew by default.

Octave.app DMGs currently require the release-page xattr -c quarantine step before opening the installer on recent macOS releases.

Octave.app 9.2 standalone DMGs Octave.app release archive Octave downloads Official macOS install notes Homebrew Octave MacPorts Octave Octave documentation

Research code framework

Cactus Code on macOS

Cactus Code is a general-purpose, component-based framework for high-performance scientific computing. It is used by projects such as the Einstein Toolkit and expects a Unix-like HPC software stack: C, C++, Fortran, MPI, Python, Perl, git, and subversion. There is no maintained standalone Mac binary distribution; for macOS workstations, use current Cactus and project-specific requirements, then choose the matching Homebrew or MacPorts option list after checkout.

Mac workstation notes

Use the current Cactus and project release pages for download commands; avoid old fixed-release tutorial commands.
For Einstein Toolkit work, inspect simfactory/mdb/optionlists/osx-homebrew.cfg or osx-macports.cfg after checkout.
Use laptop builds for learning, testing, and small examples; move production runs to real HPC systems.

Cactus Code Cactus download and requirements Einstein Toolkit download and requirements SimFactory workstation setup notes Jupyter/ET container materials

Scientific code

GRAVSIM

GRAVSIM files on SourceForge

GRAVSIM is a gravitational N-body package originally written by Mark Bellon and distributed through this project. It contains solvers for tracking bodies moving under mutual gravity and was designed as a readable study of mathematical, vector, and multiprocessor optimization techniques.

The package includes Newtonian and Barnes-Hut methods, X11-based viewing support, and a data-access library for analysis tools. It is preserved here as scientific code and as a useful example of performance-oriented Mac and Unix-era numerical programming.