OpenBLAS

OpenBLAS is an open-source high-performance implementation of the Basic Linear Algebra Subprograms (BLAS) API. It provides optimized kernels for a wide range of CPUs and is designed to be a drop-in replacement for other BLAS libraries.

OpenBLAS implements BLAS routines in C and Fortran and exposes a stable BLAS interface for use by applications. It includes architecture-specific kernels and can automatically detect the host CPU to select appropriate optimizations at run time. It supports multi-threading to utilize multi-core processors, with thread count controlled via environment variables such as OPENBLAS_NUM_THREADS.

The project originated as a fork of the GotoBLAS2 project and is developed by a community of contributors. It is distributed under a permissive BSD-style license.

OpenBLAS is widely used in scientific computing on Linux, macOS, and Windows. It is available in many operating system package repositories and can be built from source. It is commonly used as the default BLAS in various numerical ecosystems, including Python data science stacks that rely on compiled numerical libraries.