This Article 
 Bibliographic References 
 Add to: 
The Impact of Pipelined Channels on k-ary n-Cube Networks
January 1994 (vol. 5 no. 1)
pp. 2-16

In a pipelined-channel interconnection network, multiple bits may be simultaneously inflight on a single wire. This allows the cycle time of the network to be independent of thewire lengths, significantly affecting the network design trade-offs. This paper investigatesthe design and performance of pipelined channel k-ary n-cube networks, with particularemphasis on the choice of dimensionality and radix. Networks are investigated under theconstant link width, constant node size and constant bisection constraints. We find thatthe optimal dimensionality of pipelined-channel networks is higher than that ofnonpipelined-channel networks, with the difference being greater under looser wiringconstraints. Their radix should remain roughly constant as network size is grown,decreasing slightly for some unidirectional tori and increasing slightly for some bi-directional meshes. Pipelined-channel networks are shown to provide lower latency and higher bandwidth than their nonpipelined-channel counterparts, especially for high-dimensional networks. The paper also investigates the effects of switching overhead and message lengths, indicating where results agree with and differ from previous results obtained for nonpipelined-channel networks.

[1] S. Abraham and K. Padmanabhan, "Performance of the direct binaryn- cube network for multiprocessors,"IEEE Trans. Comput., vol. 38, no. 7, pp. 1000-1011, July 1989.
[2] S. Abraham and K. Padmanabhan, "Performance of multicomputer networks under pin-out constraints,"J. Parallel Distributed Computing, vol. 12, pp. 237-248, July 1991.
[3] V. S. Adve and M. K. Vemon, "Performance analysis of multiprocessor mesh interconnection networks with wormhole routing," Comput. Sci. Tech. Rep. #1001, Univ. of Wisconsin-Madison, Madison, WI, Feb. 1991.
[4] A. Agrawal et al., "APRIL: A Processor Architecture for Multiprocessing,"17th Int'l Symp. Computer Architectures, 1990, IEEE Computer Soc. Press, Los Alamitos, Calif., Order No. 2047, pp. 104-114.
[5] A. Agarwal, "Limits on interconnection network performance,"IEEE Trans. Parallel Distributed Syst., pp. 398-412, Oct. 1991.
[6] R. Alversonet al., "The Tera computer system," inProc. Int. Conf. on Supercomput. ACM Press, 1990, pp. 1-6.
[7] S. Borkar et al., "Supporting Systolic and Memory Communication in iWarp,"Proc. 17th Ann. Int'l Symp. Computer Architecture, IEEE Computer Society Press, Los Alamitos, 1990, pp. 70-81.
[8] W.J. Dally and C.L. Seitz, "Deadlock-Free Message Routing in Multiprocessor Interconnection Networks,"IEEE Trans. Computers, Vol. C-36, No. 5, May 1987, pp. 547-553.
[9] W. J. Dally, "Performance analysis ofk-aryn-cube interconnection networks,"IEEE Trans. Comput.vol. 39, no. 6, pp. 775-785, June 1990.
[10] W. J. Dally, "Virtual-channel flow control,"IEEE Trans. Parallel Distributed Syst.vol. 3, no. 2, pp. 194-205, Mar. 1992.
[11] D. M. Dias and J. R. Jump, "Packet switching interconnection networks for modular systems,"IEEE Comput.vol. 14, pp. 43-53, Dec. 1981.
[12] T. Feng, "A survey of interconnection networks,"Comput., pp. 12-27, Dec. 1981.
[13] D. Hillis,The Connection Machine. Cambridge, MA: M.I.T. Press, 1985.
[14] IEEE Scalable Coherent Interface, IEEE Standard 1596-1992, 1992.
[15] P. Kermani and L. Kleinrock, "Virtual cut-through: A new computer communication switching technique,"Comput. Net., vol. 3, pp. 267-286, 1979.
[16] C. P. Kruskal and M. Snir, "The performance of multistage interconnection networks for multiprocessors,"IEEE Trans. Comput., vol. C-32, no. 12, pp. 1091-1098, Dec. 1983.
[17] D. H. Lawrie, "Access and alignment of data in an array processor,"IEEE Trans. Comput., vol. C-24, no. 12, pp. 1145-1155, Dec. 1975.
[18] M. Lee and C.-L. Wu, "Performance analysis of circuit switching baseline interconnection network," inProc. 11th Annu. Symp. Comput. Architecture, 1984, pp. 82-90.
[19] D. Lenoski, J. Laudon, K. Gharachorloo, A. Gupta, J. Hennessy, M. Horowitz, and M. Lam, "Design of the Stanford DASH multiprocessor," CSL Tech. Rep. 89-403, Stanford Univ., Dec. 1989.
[20] J. H. Patel, "Performance of processor-memory interconnections for multiprocessors,"IEEE Trans. Comput., vol. C-30, no. 10, pp. 771-780, Oct. 1981.
[21] S. L. Scott, "A cache coherence mechanism for scalable, shared memory multiprocessors," inProc. 1991 Int. Symp. Shared Memory Multiprocessing, Apr. 1991, pp. 49-59.
[22] S. L. Scott, J. R. Goodman, and M. K. Vernon, "Performance of the SCI ring," inProc. 18th Annu. Int. Symp. Comput. Architecture, Gold Coast, Australia, May 1992, pp. 403-414.
[23] C. L. Seitz, "Concurrent VLSI architectures,"IEEE Trans. Comput., vol. C-33, no. 12, pp. 775-785, Dec. 1984.
[24] C. L. Seitz, "The Cosmic Cube,"Commun. ACM, pp. 22-33, Jan. 1985.
[25] C. L. Seitz, Personal correspondence, Apr. 1992.
[26] H. J. Siegal, Interconnection networks for SIMD machines,Comput., vol. 12, no. 6, pp. 57-65, June 1979.
[27] J. E. Smith, Personal correspondence, Apr. 1992.
[28] H. Sulivan and T. R. Bashkov, "A large scale homogeneous, fully distributed parallel machine, I," inProc. 4th Symp. Comput. Arch., March 1977, pp. 105-117.
[29] A. Tanenbaum,Computer Networks. Englewood Cliffs, NJ: Prentice-Hall, 1988.
[30] TMC,CM5 Reference Manual. Cambridge, MA: Thinking Machines, Inc. 1991.
[31] H. Yoon, K. Y. Lee, and M. T. Liu, "Performance analysis and comparison of packet switching interconnection networks," inProc. 1987 Int. Conf. Parallel Processing, Aug. 1987, pp. 542-545.

Index Terms:
Index Termsmultiprocessor interconnection networks; performance evaluation; pipeline processing;pipelined channels; k-ary n-cube networks; cycle time; bisection constraints; optimaldimensionality; pipelined-channel networks; switching overhead; message lengths
S.L. Scott, J.R. Goodman, "The Impact of Pipelined Channels on k-ary n-Cube Networks," IEEE Transactions on Parallel and Distributed Systems, vol. 5, no. 1, pp. 2-16, Jan. 1994, doi:10.1109/71.262584
Usage of this product signifies your acceptance of the Terms of Use.