High-performance computing: Successes, failures, and future directions

The field of high-performance computing has seen many turns. The 1980s and the 1990s were golden years for academic research in parallel processing. With the promise of massively parallel systems, the early 1990s saw an array of aggressive vendors producing a range of cutting edge parallel computers. Soon after, many of those vendors closed their doors and the research community diverted most of its attention to other areas. With the introduction of clusters and their march into the top 500 supercomputers list, the field started to re-emerge. With the realization that there is a tremendous need for more high-end computing systems and the birth of the Earth simulator, the world's fastest computer in Japan, the competition heated up and many research and development projects are now underway to produce not only new large installations but also engender new architectures of parallel computers that can be in production before the end of the decade. Given this roller coaster of events in this domain we assemble a panel of leading experts from academia, research labs, and industry to debate what the failures were successes were and predict the future of this domain from the perspectives of both basic research and developments at industry.