Research directions in compiling for massive parallelism

Abstract

in this paper we summarize and amplify the discussion of research directions in compiling for massive parallelism that took place at New Frontiers: A Workshop on Future Directions in Massively Parallel Processing which preceded the Frontier's '92 Conference. In addition to those problems specific to compilation for massively parallel machines, problems discussed included issues in language design^ the success of current vectorizing compilers and the implications for compiling for massively parallel machinesj and the necessity of optimizing for uniprocessor performance. Center for Supercomputing Research and Development.