The Toeplitz algebra library routines allow the user to take advantage of both multithreaded and memory-distributed programming paradigms and are therefore adapted to run efficiently on heteregeneous computer architectures. The multithreading is implemented using openMP directives, while distributed programming uses MPI. Both shared and/or distributed parallelism can be switched of, at the compilation time, if so desired. Moreover, the user has always access to two versions of each of the routines: openMP/MPI and openMP-only.

We note that the MPI version of the routines are essentially just wrappers on openMP/sequential versions of the corresponding routines, which facilitate necessary data exchanges between distributed MPI processes.