It is now widely realized that large capital investments in acquiring a single HPC facility for shared memory multiprocessor parallel computing is not an economically viable route for modest budget research areas such as astrophysics. An attractive route to retaining competitive edge in high performance computing is based on clusters of modest computing units linked over a fast network. We plan to adopt the cluster concept where very high computational performance is achieved using a number of modest computing units clustered over a very fast network. Distributed memory parallel computing is cost effective and readily scalable. Both cost and performance scale linearly with number of units. This allows for sustained enhancement of computing power with modest periodic capital investments that allow addition and/or upgrade of units.
High-Performance Computing
Research in a number of areas of Astronomy and Astrophysics (A&A) now rely on high performance computing (HPC) and capability to manage and process large volume of data. This trend is reflected in the research of a number of researchers in IUCAA. We have set up a high performance computational facility for astrophysical sciences in IUCAA.