International Journal of Advanced Research in Computer Science and Software Engineering (IJARCSSE)
Presently distributed cluster computations are increasingly used in resolution of various research and practical problems. However, clusters are quite often developed for resolution of some set of problems of a single or of as little as two or three considerably narrow subject areas. In such cases the attention is paid to the issue of development of clusters which would feature the best capacity for the particular set of problems under cost and input power constraints. A way to such an optimization is to install one or more hardware accelerators in parallel with the main processor. Such accelerators could be custom microprocessors, programmable logic microcell arrays that perform a particular set of functions.