PSBLAS (Parallel Sparse Basic Linear Algebra Subprograms)

PSBLAS provides a framework to enable easy, efficient and portable implementations of iterative solvers for linear systems on parallel computers. It implements parallel versions of most of the Sparse BLAS computational kernels (sparse matrix by dense matrix multiplication, sparse block diagonal system solution, sparse matrix norms, vector norms, etc.), sparse matrix management functionalities and parallel implementations of preconditioned Krylov solvers.

The PSBLAS library is built with a user interface making full use of the modern features of Fortran 95, including facilities for data encapsulation, functional overloading and dynamic memory management; the parallel implementation is based on the SPMD paradigm, using MPI and BLACS for inter-process data exchange.