Numerical Libraries and Tools for Scalable Parallel Cluster Computing
Jack J. Dongarra, Shirley Moore, Anne Trefethen · The International Journal of High Performance Computing Applications · 2001
For cluster computing to be effective within the scientific community it is essential that there are numerical libraries and programming tools available to application developers. Cluster computing may mean a cluster of heterogeneous components or hybrid architecture with some SMP nodes. This is clear at the high end, for example the latest IBM SP architecture, as well as in clusters of PC-based workstations. However, these systems may present very different software environments on which to build libraries and applications, and indeed require a new level of flexibility in the algorithms if they are to achieve an adequate level of performance. We will consider here the libraries and software tools that are already available and offer directions that might be taken in the light of cluster computing.