Advertisement

On Validating Parallel Architectures via Graph Embeddings

  • Arnold L. Rosenberg

Abstract

A viable multi-purpose parallel architecture must exhibit many qualities, including: technological feasibility, low communication overhead, robustness in the face of faults, and programmability. The maturity and reliability of techniques for detecting and studying these desiderata decrease rapidly in the order of their listing. The issues of technological feasibility and communication overhead are relatively well understood; progress is being made on the problem of detecting and enhancing the robustness of proposed architectures; but the issue of programmability has not yet even been formulated in a generally accepted way. We extrapolate here from joint work with S. N. Bhatt, F. R. K. Chung, L. S. Heath, and F. T. Leighton, to propose an avenue for assessing the programmability of a proposed parallel architecture, using the formal tool of graph embeddings. As is common, we represent the communication structure of a parallel architecture as an undirected graph, and we seek efficient embeddings of a variety of specific graph families in the architecture graph. We choose the graph families to abstract the intertask dependency structures of popular algorithmic strategies, such as divide-and-conquer and convolution. We illustrate the approach by describing recent studies of optimal embeddings in the Hypercube architecture.

Keywords

Binary Tree Rectangular Grid Parallel Architecture Graph Embedding Technological Feasibility 
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]
    A.V. Aho, J.E. Hoperoft, J.D. Ullman (1974): The Design and Analysis of Computer Algorithms. Addison-Wesley, Reading, MA.Google Scholar
  2. [2]
    F. Berman and L. Snyder (1984): On mapping parallel algorithms into parallel architectures. Intl. Conf. on Parallel Processing.Google Scholar
  3. [3]
    S.N. Bhatt, F.R.K. Chung, F.T. Leighton, A.L. Rosenberg (1986): Optimal simulations of tree machines. 27th IEEE Symp. on Foundations of Computer Science, 274–282.Google Scholar
  4. [4]
    S.N. Bhatt and F.T. Leighton (1984): A framework for solving VLSI graph layout problems. J. Comp. Syst. Sci. 28, 300–343.CrossRefGoogle Scholar
  5. [5]
    N. Blum (1983): An area-maximum edge length tradeoff for VLSI layout. 16th ACM Symp. on Theory of Computing, 92–97.Google Scholar
  6. [6]
    S.H. Bokhari (1981): On the mapping problem. IEEE Trans. Comp., C-30, 207–214.CrossRefGoogle Scholar
  7. [7]
    F.R.K. Chung, F.T. Leighton, A.L. Rosenberg (1983): DIOGENES — A methodology for designing fault-tolerant processor arrays. 13th Intl. Conf. on Fault-Tolerant Computing, 26–32.Google Scholar
  8. [8]
    J.W. Greene and A. El Gamal (1984): Configuration of VLSI arrays in the presence of defects. J. ACM 31, 694–717.CrossRefGoogle Scholar
  9. [9]
    J. Hastad, F.T. Leighton, M. Newman (1986): Reconfiguring a hypercube in the presence of faults. Typescript, MIT.Google Scholar
  10. [10]
    L.S. Heath and A.L. Rosenberg (1987): An optimal mapping of the FFT algorithm onto the Hypercube architecture. Tech. Rpt., Univ. of Massachusetts; submitted for publication.Google Scholar
  11. [11]
    J.-W. Hong, K. Mehlhorn, A.L. Rosenberg (1983): Cost tradeoffs in graph embeddings. J. ACM 30, 709–728.CrossRefGoogle Scholar
  12. [12]
    L. Johnsson (1985): Basic linear algebra computations on hypercube architectures. Tech. Rpt., Yale Univ.Google Scholar
  13. [13]
    F.T. Leighton (1984): Parallel computation using meshes of trees. 1983 Workshop on Graph-Theoretic Concepts in Computer Science, Trauner Verlag, Linz, pp. 200–218.Google Scholar
  14. [14]
    F.T. Leighton and C.E. Leiserson (1985): Wafer-scale integration of systolic arrays. IEEE Trans. Comp., C-34, 448–461.CrossRefGoogle Scholar
  15. [15]
    R.J. Lipton, A.L. Rosenberg, A.C. Yao (1980): External hashing schemes for collections of data structures. J. ACM 27, 81–95.CrossRefGoogle Scholar
  16. [16]
    F.P. Preparata and J.E. Vuillemin (1981): The cube-connected cycles: a versatile network for parallel computation. C ACM 24, 300–309.Google Scholar
  17. [17]
    A.L. Rosenberg (1981): Issues in the study of graph embeddings. In Graph-Theoretic Concepts in Computer Science: Proceedings of the International Workshop WG80, Bad Honnef, Germany (H. Noltemeier, ed.) Lecture Notes in Computer Science 100, Springer-Verlag, New York 150–176.Google Scholar
  18. [18]
    A.L. Rosenberg (1983): The Diogenes approach to testable fault-tolerant arrays of processors. IEEE Trans. Comp., C-32, 902–910.Google Scholar
  19. [19]
    L. Snyder (1981): Overview of the CHiP computer. In VLSI 81: Very Large Scale Integration (J.P. Gray, ed.) Academic Press, London, pp. 237–246.Google Scholar

Copyright information

© Plenum Press, New York 1988

Authors and Affiliations

  • Arnold L. Rosenberg
    • 1
  1. 1.University of MassachusettsAmherstUSA

Personalised recommendations