Advertisement

A Modest Proposal for Communication Costs in Multicomputers

  • Paul M. B. Vitányi

Abstract

Getting rid of the ‘von Neumann’ bottleneck in the shift from sequential to non-sequential computation, a new communication bottleneck arises because of the interplay between locality of computation, communication, and the number of dimensions of physical space. As a consequence, realistic models for non-sequential computation should charge extra for communication, in terms of time and space. We give a proposal for this that is more subtle, but mutatis mutandis similar to the transition from unit cost to logarithmic cost in the sequential Random Access Machine model. The space cost of communication is related with the topology of the communication graph. We use a new lower bound on the average interconnect (edge) length in terms of diameter and symmetry of the topology. This lower bound is technology independent, and shows that many interconnection topologies of today’s multicomputers do not scale well in the physical world with 3 dimensions.

Keywords

Communication Cost Space Cost Defense Advance Research Project Agency Complete Binary Tree Random Access Machine 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    Harary, F., Graph Theory, Addison-Wesley, Reading, Mass., 1969.Google Scholar
  2. [2]
    Leighton, F.T. and A. Rosenberg, “Three-dimensional circuit layouts,” Tech. Rept. TM-262, MIT, Laboratory for Computer Science, Cambridge, Mass., June, 1984.Google Scholar
  3. [3]
    Mead, C. and L. Conway, Introduction to VLSI Systems, Addison-Wesley, Reading, Mass., 1980.Google Scholar
  4. [4]
    Mead, C. and M. Rem, “Minimum propagation delays in VLSI,” IEEE J. on SolidState Circuits, vol. SC-17, 1982, 773–775; Correction: ibid, vol. SC-19, 1984, 162.Google Scholar
  5. [5]
    Ullman, J.D., Computational Aspects of VLSI, Computer Science Press, Rockville, Maryland, 1984.Google Scholar
  6. [6]
    Vitányi, P.M.B., “Non-sequential computation and Laws of Nature,” in VLSI Algorithms and Architectures, Lecture Notes in Computer Science, vol. 227, Springer Verlag, Berlin, 1986, 108–120.CrossRefGoogle Scholar
  7. [7]
    Vitányi, P.M.B., “Locality, Communication, and Interconnect Length in Multicomputers,” SIAM J. on Computing, to appear.Google Scholar
  8. [8]
    Vitányi, P.M.B., “Area penalty for sublinear signal propagation delay on chip,” Proceedings 26th Annual IEEE Symposium on Foundations of Computer Science, 1985, 197–207.Google Scholar

Copyright information

© Plenum Press, New York 1988

Authors and Affiliations

  • Paul M. B. Vitányi
    • 1
  1. 1.Centrum voor Wiskunde en InformaticaAmsterdamThe Netherlands

Personalised recommendations