Parallelisation on HP Systems

To program parallel computers different concepts can be used. Nowadays programming for computers with distributed memory and the explicitly exchange of messages between separated tasks of a parallel application are widely spread. MPI is the standard for this purpose. All manufacturers of parallel computers nowadays offer efficient implementations of MPI.

An older communication library often used on workstation clusters is PVM. The most significant advantage of MPI is its fast communication within a parallel computer and the advantage of PVM is the management of separate tasks within a workstation cluster. 

OpenMP is an extension of Fortran and C/C++ for the programming of parallel shared memory systems. OpenMP is a de facto standard for the parallelization of programs on these  systems.

• Message Passing Interface (MPI)
• OpenMP (only in German)