ABSTRACT
This chapter discusses directions for development that many computational scientists and researchers feel are most promising for yielding long-term, feasible platforms on which to build a wide range of supercomputing applications: parallel computing in distributed computing environments. It presents some observations on trends for delivering supercomputing services to scientists and engineers. In general, the development of hardware and software, and the structure of computing service organizations in the 1990s should reflect the needs of the computing applications for research endeavors that will provide breakthroughs for science and eventual economic gains for US industries. For the long-term future, one of many developments to watch is supercomputing that integrates a large number of microprocessors. A distributed rather than single-machine solution of networked data and computing resources is one of the most practical and financially viable solutions for future computing configurations. Solutions for the future will be refined as new technical alternatives come forth.
