Parallel Processing Implications for Management Scientists

As parallel processing systems become more accessible, management scientists need to be aware of their implications for algorithm design. Parallel processing systems can simultaneously execute several programs and can be categorized by the number of processors available or the number of data streams that can be used. Efficient interprocessor communication is essential. Fine-grained parallel programs explicitly make use of these communication functions to improve speed, while coarse-grained programs interact less frequently and distribute similar subprograms among processors. In either case, procedures can be used to enable serial programs to execute on parallel processors. Management scientists have used these systems to investigate linear programming problems, simulation models, and network analysis. They offer users opportunities to address problems whose solutions were previously limited by serial-processing architectures.