Advertisement

SLiD — A cost-effective and Scalable Limited-Directory scheme for cache coherence

  • Guoying Chen
Paper Sessions Architectures: Caches
Part of the Lecture Notes in Computer Science book series (LNCS, volume 694)

Abstract

In this paper, we propose a hybrid scheme — dubbed SLiD (Scalable Limited Directory) that combines the good features of the chained directory and limited directory schemes and minimizes the effect of the weak points of each individual scheme, namely the performance penalty in the limited directory when pointer overflow occurs, and the linear latency in the chained directory scheme when performing invalidations due to a write. Moreover, the hybrid scheme provides designers with three structural options for different requirements of performance and implementation cost.

Simulation studies demonstrate that the SLiD scheme is promising. Compared with other related schemes, the SLiD scheme is scalable and cost-effective.

Keywords

Limited Directory Cache Line Memory Block Storage Overhead Multiple Link 
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.
    Anant Agarwal, David Chaiken, Guoying Chen, and Allan Gottlieb. Private Communications, November 1992.Google Scholar
  2. 2.
    Anant Agarwal, Richard Simoni, John Hennessy, and Mark Horowitz. An evaluation of directory schemes for cache coherence. In Proceedings of the 15th Annual International Symposium of Computer Architecture, pages 280–289, May 1988.Google Scholar
  3. 3.
    L. M. Censier and P. Feautrier. A new solution to coherence problems in multicache systems. IEEE Trans. on Computers, C-27(12): 1112–1118, December 1978.Google Scholar
  4. 4.
    David Chaiken, Craig Fields, Kiyoshi Kurihara, and Anant Agarwal. Directory-based cache coherence in large-scale multiprocessors. Computer, 23(6):49–58, June 1990.Google Scholar
  5. 5.
    David Chaiken, John Kubiatowics, and Anant Agarwal. LimitLESS directories: A scalable cache coherence scheme. In Proceedings of the 4th International Conference on Architectural Support for Programming Languages and Systems, pages 224–234, April 1991.Google Scholar
  6. 6.
    Guoying Chen. SLiD — A cost-effective and scalable limited-directory scheme for cache coherence. Ultracomputer Note 186, Courant Institute, NYU, November 1992.Google Scholar
  7. 7.
    Guoying Chen. Design and Simulation Analysis of Directory-Based Cache Coherence Schemes for Large-Scale Multiprocessors. PhD thesis, Courant Institute, NYU, New York, 1993. (In preparation).Google Scholar
  8. 8.
    Guoying Chen. An evaluation of the SLiD and its related cache coherence schemes. Ultracomputer note, Courant Institute, NYU, March 1993. (submitted for publication).Google Scholar
  9. 9.
    Anoop Gupta, Wolf-Dietrich Weber, and Todd Mowry. Reducing memory and traffic requirements for scalable directory-based cache coherence schemes. In Proceedings of the 1990 International Conference on Parallel Processing, pages 1312–1321, Auguest 1990.Google Scholar
  10. 10.
    David V. James, Anthony T. Laundrie, Stein Gjessing, and Gurindar S. Sohi. Distributeddirectory scheme: Scalable coherent interface. Computer, 23(6):74–77, June 1990.Google Scholar
  11. 11.
    Håkan Nilsson and Per Stenström. The scalable tree protocol — a cache coherence approach for large-scale multiprocessors. In Proceedings of the Fourth IEEE Symposium on Parallel and Distributed Processing, November 1992.Google Scholar
  12. 12.
    Richard Simoni and Mark Horowitz. Dynamic pointer allocation for scalable cache directories. In Proceedings of 1st International Symposium on Shared Memory Multiprocessing, pages 309–318, April 1991.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1993

Authors and Affiliations

  • Guoying Chen
    • 1
  1. 1.Ultracomputer Research LaboratoryNew York UniversityNew York

Personalised recommendations