Guest Editors' Introduction Parallel Processing

AS we approach the limits of how much a single processor's speed can be increased through advances in technology, there is no question that parallelism will be the key to building tomorrow's fastest, mostpowerful computers. Current parallel processing studies demonstrate that parallelism has the potential to provide solutions to a wide range of computationally intensive problems. Nonetheless, at a time when "supercomputer" is almost a household word, fundamental issues in parallel processing continue to demand attention. This Special Issue is evidence of the scope of the field of parallel processing. Topics addressed include synchronous and asynchronous architectures; communications via buses, trees, pipelines, crossbars, and multistage inter-connection networks; algorithms for numerical matrix operations, image generation (with Star Trek II cited as an example), dictionary searches, and database queries; efficient scheduling of hundreds of tasks; automatic vectorization of constructs that are not directly vectorizable; and performance evaluations of all of the above. A brief introduction to the papers follows.