This book discusses complex performance evaluation of various typical parallel algorithms (shared memory, distributed memory) and their practical implementations. As real application examples it shows the various influences during the process of modelling and performance evaluation and the consequences of their distributed parallel implementations.

The current trends in High Performance Computing (HPC) are to use networks of workstations (NOW, SMP) or a network of NOW networks (Grid) as a cheaper alternative to the traditionally-used, massive parallel multiprocessors or supercomputers. Individual workstations could be single PCs (personal computers) used as parallel computers based on modern symmetric multicore or multiprocessor systems (SMPs) implemented inside the workstation.

With the availability of powerful personal computers, workstations and networking devices, the latest trend in parallel computing is to connect a number of individual workstations (PCs, PC SMPs) to solve computation-intensive tasks in a parallel way to typical clusters such as NOW, SMP and Grid. In this sense it is not yet correct to consider traditionally evolved parallel computing and distributed computing as two separate research disciplines.

To exploit the parallel processing capability of this kind of cluster, the application program must be made parallel. An effective way of doing this for (parallelisation strategy) belongs to the most important step in developing an effective parallel algorithm (optimisation). For behaviour analysis we have to take into account all the overheads that have an influence on the performance of parallel algorithms (architecture, computation, communication etc.).