Recursive Dual-Net: A New Universal Network for Supercomputers of the Next Generation

  • Yamin Li
  • Shietung Peng
  • Wanming Chu
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5574)

Abstract

In this paper, we propose a new universal network, called recursive dual-net (RDN), as a potential candidate for the interconnection network of supercomputers with very large scale. The recursive dual-net is based on a recursive dual-construction of a base network. A k-level dual-construction for k > 0 creates a network containing \((2m)^{2^k}/2\) nodes with node-degree d + k, where m and d are the number of nodes and the node-degree, respectively, of the base network. The recursive dual-net is node and edge symmetric and can contain huge number of nodes with small node-degree and short diameter. For example, we can construct a symmetric RDN connecting more than 3-million nodes with only 6 links per node and a diameter of 22. We investigate the topological properties of the recursive dual-net and compare it to that of other networks including 3D torus, WK-recursive network, hypercube, cube-connected-cycle, and dual-cube. We also establish the basic routing and broadcasting algorithms for the proposed recursive dual-net.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Adiga, N.R., Blumrich, M.A., Chen, D., Coteus, P., Gara, A., Giampapa, M.E., Heidelberger, P., Singh, S., Steinmacher-Burow, B.D., Takken, T., Tsao, M., Vranas, P.: Blue gene/l torus interconnection network. IBM Journal of Research and Development 49(2/3), 265–276 (2005), http://www.research.ibm.com/journal/rd/492/tocpdf.html CrossRefGoogle Scholar
  2. 2.
    Aki, S.G.: Parallel Computation: Models and Methods. Prentice-Hall, Englewood Cliffs (1997)Google Scholar
  3. 3.
    Beckman, P.: Looking toward exascale computing, keynote speaker. In: International Conference on Parallel and Distributed Computing, Applications and Technologies (PDCAT 2008), University of Otago, Dunedin, New Zealand, December 2 (2008)Google Scholar
  4. 4.
    Chen, G.H., Duh, D.R.: Topological properties, communication, and computation on wk-recursive networks. Networks 24(6), 303–317 (1994)MathSciNetCrossRefGoogle Scholar
  5. 5.
    Ghose, K., Desai, K.R.: Hierarchical cubic networks. IEEE Transactions on Parallel and Distributed Systems 6(4), 427–435 (1995)CrossRefGoogle Scholar
  6. 6.
    Leighton, F.T.: Introduction to Parallel Algorithms and Architectures: Arrays, Trees, Hypercubes. Morgan Kaufmann, San Francisco (1992)MATHGoogle Scholar
  7. 7.
    Li, Y., Peng, S.: Dual-cubes: a new interconnection network for high-performance computer clusters. In: Proceedings of the 2000 International Computer Symposium, Workshop on Computer Architecture, ChiaYi, Taiwan, December 2000, pp. 51–57 (2000)Google Scholar
  8. 8.
    Li, Y., Peng, S., Chu, W.: Efficient collective communications in dual-cube. The Journal of Supercomputing 28(1), 71–90 (2004)CrossRefMATHGoogle Scholar
  9. 9.
    Preparata, F.P., Vuillemin, J.: The cube-connected cycles: a versatile network for parallel computation. Commun. ACM 24, 300–309 (1981)MathSciNetCrossRefGoogle Scholar
  10. 10.
    Saad, Y., Schultz, M.H.: Topological properties of hypercubes. IEEE Transactions on Computers 37(7), 867–872 (1988)CrossRefGoogle Scholar
  11. 11.
    TOP500. Supercomputer Sites (June 2008), http://top500.org/
  12. 12.
    Varma, A., Raghavendra, C.S.: Interconnection Networks for Multiprocessors and Multicomputers: Theory and Practice. IEEE Computer Society Press, Los Alamitos (1994)MATHGoogle Scholar
  13. 13.
    Vicchia, G., Sanges, C.: A recursively scalable network vlsi implementation. Future Generation Computer Systems 4(3), 235–243 (1988)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Yamin Li
    • 1
  • Shietung Peng
    • 1
  • Wanming Chu
    • 2
  1. 1.Department of Computer ScienceHosei UniversityTokyoJapan
  2. 2.Department of Computer HardwareUniversity of AizuAizu-WakamatsuJapan

Personalised recommendations