Behind the Scenes of HPCC

In Suggesting Computer Science Agendas for High Performance Computing, U. Viskin, ed., ACM, 1994, pp. 156-158.

Behind the Scenes of HPCC

Quentin F. Stout
EECS Department, University of Michigan

Abstract: This volume is a collection of suggestions, evaluations, and position papers on the role of computer science in high performance computing. In this paper, the emphasis is on the lack of awareness of and appreciation for the computer science work that has already produced large parallel machines, the Web, stunning graphics, etc. In particular, by not understanding the large body of work that built up to the current situation, many planners make unreasonably optimistic forecasts for the rate at which new breakthroughs can be achieved. Many of the decisions have been made with little computer science input, and inadequate computer science support (for example, many of the large systems are poorly instrumented, making it impossible to collect data which would be valuable in designing future systems). Rather than flooding money on projects in expectation of quick successes, a steadier stream of long-term funding, coupled with better interaction and planning with computer scientists, is more likely to produce the desired results.

Computer scientists, for their part, must push to increase their role in planning for high performance computing, need to be more concerned about creating artifacts that others can utilize, and need to improve their MA/MS programs so that can help develop users who can better exploit these machines.

Keywords: HPCC, funding, planning, instrumentation, parallel computing, supercomputing, graphics, professional training and degrees