Provide high quality, high performance numerical methods for matrix arithmetic and for solving linear algebra problems.
This project is the SimTK Core implementation of the extremely reliable, high speed linear algebra package LAPACK and the underlying BLAS library on which LAPACK is built. We make use of ATLAS to generate "hand tuned" BLAS kernels for a variety of hardware platforms, including multiprocessors, using a variety of operating systems including Windows, Mac, and Red Hat Linux. We pre-build on all these platforms and make the binaries available as a single shared library which can be conveniently used by any program. This means that users who are not experts in high performance scientific computation can nonetheless use the fastest linear algebra methods available for their machines.
Our download page provides libraries that are specifically tuned for your hardware platform -- come and get 'em. The project also provides example programs, documentation and regression tests.
Oct 23, 2009