On the tradeoff between node degree and communication channel width in shuffle-exchange networks

The architecture of direct connected multiprocessor using the K-ary shuffle-exchange topology is investigated in the presence of node pin-out constraints. These constraints translate into a tradeoff between node degree and width of the communication channels between nodes. A higher order shuffle uses a higher node degree to reduce inter-node distances, but it also results in a lower bandwidth per channel by decreasing its width. A generalization of the K-ary shuffle exchange architecture to non-power-of-K system sizes permits the study of this tradeoff at a fine level, for values of K=2, 3, 4, . . . This study points out that the tradeoff is more complex here than in the family of multidimensional meshes. Very low order shuffles (K<0.5 log/sub 2/N) do not perform well in any of the situations investigated. Among higher order shuffles, the best structure depends upon both the system size and the pin-out constraint, and monotonic reduction in message delays is not guaranteed as the order of the shuffle is increased.<<ETX>>

[1]  Krishnan Padmanabhan,et al.  Design and Analysis of Even-Sized Binary Shuffle-Exchange Networks for Multiprocessors , 1991, IEEE Trans. Parallel Distributed Syst..

[2]  Duncan H. Lawrie,et al.  A Class of Redundant Path Multistage Interconnection Networks , 1983, IEEE Transactions on Computers.

[3]  Kenneth C. Sevcik,et al.  Reduced distance routing in single-state shuffle-exchange interconnection networks , 1987, SIGMETRICS '87.

[4]  Tomás Lang,et al.  Interconnections Between Processors and Memory Modules Using the Shuffle-Exchange Network , 1976, IEEE Transactions on Computers.

[5]  Leonard Kleinrock,et al.  Virtual Cut-Through: A New Computer Communication Switching Technique , 1979, Comput. Networks.

[6]  Marc Snir,et al.  The Performance of Multistage Interconnection Networks for Multiprocessors , 1983, IEEE Transactions on Computers.

[7]  Krishnan Padmanabhan,et al.  Performance of Multicomputer Networks under Pin-out Constraints , 1991, J. Parallel Distributed Comput..

[8]  Duncan H. Lawrie,et al.  Access and Alignment of Data in an Array Processor , 1975, IEEE Transactions on Computers.

[9]  C. Thomborson,et al.  Area-time complexity for VLSI , 1979, STOC.

[10]  S. Chittor,et al.  Link switching: a communication architecture for configurable parallel systems , 1989, Eighth Annual International Phoenix Conference on Computers and Communications. 1989 Conference Proceedings.

[11]  Pen-Chung Yew,et al.  An Enhancement Scheme for Hypercube Interconnection Networks , 1987, ICPP.

[12]  Harold S. Stone,et al.  Parallel Processing with the Perfect Shuffle , 1971, IEEE Transactions on Computers.