Interconnection Networks for Large-scale Parallel Processing: Theory and Case StudiesParallel computer systems are being used to forecast the weather, make maps, simulate chemical reactions, control air traffic, guide missiles, provide robots with vision, and manage ballistic missile defense. A major problem in designing large-scale parallel systems is the construction of an interconnection network to provide interprocessor communications. This book presents the theoretical basis and a number of case studies that demonstrate how this work is done. This revision includes the most recent research in the field. |
Contents
Machine Models | 5 |
Model of SIMD Machine Architectures | 13 |
The Interconnection Networks | 20 |
Copyright | |
18 other sections not shown
Other editions - View all
Common terms and phrases
ADM and IADM ADM network algorithm assumed box in stages broadcast-routing tag chapter Consider crossbar switches cycle dashed lines data manipulator data transfers delta network disabled DTR contents ESC network example exclusive-or executed extra-stage fault model fault-tolerance criterion generalized-cube network high-order bit IEEE Transactions implementation independent subnetworks indirect binary n-cube input/output input/output ports interchange box interconnection function interconnection network lower bound mapping memory MIMD modulo multiple-SIMD multistage cube network negative dominant network output network switches network topology node omega network one-to-one connections output link output port packet Parallel Processing partition partitionable permutation PEs execute PM2+ PM2I positive dominant primary path processors R₁ reliability reroute method routing tag secondary path set to straight shown in figure shuffle shuffle-exchange Siegel SIMD machine simulate solid lines stage 0 permutations STARAN straight link swap switch tag scheme tion TNPUT Transactions on Computers two's complement