Towards a more realistic comparative analysis of multicomputer networks

Several studies have examined the relative performance merits of the torus and hypercube taking into account the channel bandwidth constraints imposed by implementation technology. While the torus has been shown to outperform the hypercube under the constant wiring density constraint, the opposite conclusion has been reached when the constant pin‐out constraint is considered. However, these studies have assumed a pure uniform traffic pattern and deterministic routing. The ‘uniform traffic’ assumption is not always justifiable in practice as there are many real‐world parallel applications that exhibit non‐uniform traffic patterns, which can create unbalanced traffic such as hotspots in the network. This paper re‐examines the performance merits of the torus and hypercube in the presence of hotspot traffic. The comparative analysis is based on fully adaptive routing as this has been gaining popularity in recent practical multicomputers. Moreover, it uses a new cost model that takes into account the implementation cost of the network and its routers. The results reveal that for moderate and large system sizes, lower dimensional k‐ary n‐cubes (e.g. 2D torus) always outperform their higher dimensional counterparts even under the pin‐out constraint. Copyright © 2004 John Wiley & Sons, Ltd.

[1]  Xiaola Lin,et al.  The Message Flow Model for Routing in Wormhole-Routed Networks , 1995, IEEE Trans. Parallel Distributed Syst..

[2]  Debashis Basak,et al.  Designing Clustered Multiprocessor Systems under Packaging and Technological Advancements , 1996, IEEE Trans. Parallel Distributed Syst..

[3]  Hamid Sarbazi-Azad,et al.  An Analytical Model of Adaptive Wormhole Routing in Hypercubes in the Presence of Hot Spot Traffic , 2001, IEEE Trans. Parallel Distributed Syst..

[4]  Bruno Ciciani,et al.  Performance evaluation of deterministic wormhole routing in k-ary n-cubes , 1998, Parallel Comput..

[5]  William J. Dally Virtual-Channel Flow Control , 1992, IEEE Trans. Parallel Distributed Syst..

[6]  Pen-Chung Yew,et al.  Performance Evaluation of Wire-Limited Hierarchical Networks , 1997, J. Parallel Distributed Comput..

[7]  Hamid Sarbazi-Azad,et al.  Analytical Modeling of Wormhole-Routed k-Ary n-Cubes in the Presence of Hot-Spot Traffic , 2001, IEEE Trans. Computers.

[8]  Lionel M. Ni,et al.  The turn model for adaptive routing , 1998, ISCA '98.

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

[10]  Daniel H. Linder,et al.  An Adaptive and Fault Tolerant Wormhole Routing Strategy for k-Ary n-Cubes , 1994, IEEE Trans. Computers.

[11]  Joydeep Ghosh,et al.  A Comprehensive Analytical Model for Wormhole Routng in Multicomputer Systems , 1994, J. Parallel Distributed Comput..

[12]  Hamid Sarbazi-Azad,et al.  An accurate analytical model of adaptive wormhole routing in k-ary n-cubes interconnection networks , 2001, Perform. Evaluation.

[13]  Anant Agarwal,et al.  Limits on Interconnection Network Performance , 1991, IEEE Trans. Parallel Distributed Syst..

[14]  José Duato,et al.  A New Theory of Deadlock-Free Adaptive Routing in Wormhole Networks , 1993, IEEE Trans. Parallel Distributed Syst..

[15]  Michael D. Noakes,et al.  The J-machine multicomputer: an architectural evaluation , 1993, ISCA '93.

[16]  Ronald I. Greenberg,et al.  Modeling and Comparison of Wormhole Routed Mesh and Torus Networks , 1997 .

[17]  Sudhakar Yalamanchili,et al.  Interconnection Networks: An Engineering Approach , 2002 .

[18]  A. A. Chein,et al.  A cost and speed model for k-ary n-cube wormhole routers , 1998 .

[19]  William J. Dally,et al.  The M-machine multicomputer , 1997, Proceedings of the 28th Annual International Symposium on Microarchitecture.

[20]  Manuel P. Malumbres,et al.  Optimal Topology for Distributed Shared-Memory Multiprocessors: Hypercubes Again? , 1996, Euro-Par, Vol. I.

[21]  Chita R. Das,et al.  Hypercube Communication Delay with Wormhole Routing , 1994, IEEE Trans. Computers.

[22]  William J. Dally,et al.  The Reliable Router: A Reliable and High-Performance Communication Substrate for Parallel Computers , 1994, PCRCW.

[23]  William J. Dally,et al.  Deadlock-Free Message Routing in Multiprocessor Interconnection Networks , 1987, IEEE Transactions on Computers.

[24]  William J. Dally,et al.  Performance Analysis of k-Ary n-Cube Interconnection Networks , 1987, IEEE Trans. Computers.

[25]  Mohamed Ould-Khaoua,et al.  A Performance Model for Duato's Fully Adaptive Routing Algorithm in k-Ary n-Cubes , 1999, IEEE Trans. Computers.

[26]  R. E. Kessler,et al.  Cray T3D: a new dimension for Cray Research , 1993, Digest of Papers. Compcon Spring.

[27]  Donald Yeung,et al.  The MIT Alewife machine: architecture and performance , 1995, ISCA '98.

[28]  Seth Abraham,et al.  Performance-Based Constraints for Multidimensional Networks , 2000, IEEE Trans. Parallel Distributed Syst..

[29]  E. Anderson,et al.  Performance of the CRAY T3E Multiprocessor , 1997, ACM/IEEE SC 1997 Conference (SC'97).

[30]  Gregory F. Pfister,et al.  “Hot spot” contention and combining in multistage interconnection networks , 1985, IEEE Transactions on Computers.

[31]  James R. Goodman,et al.  The Impact of Pipelined Channels on k-ary n-Cube Networks , 1994, IEEE Trans. Parallel Distributed Syst..