Do parallel languages respond to the needs of scientific programmers?

摘要

Parallel programming is considered from the viewpoint of scientific researchers, with particular reference to their requirements for language support. The questions of how scientists go about developing parallel applications, what role language plays in determining the success of their programming efforts, how much scientific programmers should be expected to know about parallel languages and machines and what computer scientists can do to facilitate parallel scientific programming are addressed. The discussion centers on the 'mainline' supercomputers for scientific and engineering applications: vector and scalar MIMD (multiple instruction, multiple data) multiprocessors. New research efforts that are tackling the problems facing scientific programmers are discussed.