Abstract
This paper presents an adaptive limited pointers directory scheme which works better than the broadcast and the non-broadcast limited pointers schemes. The former is good at workloads comprising mostly-read data and the latter is good at workloads comprising frequently read / write data. The proposed adaptive scheme dynamically switches between the two according to the application sharing patterns. It works well for both types of workloads, and outperforms the previous schemes for the workloads having both the mostly-read and the frequently read / write data at the same time. The adaptive limited pointers scheme reduces the traffic volume of real workloads by up to 33%. Further, it can be implemented with minor addition to the directory memory overhead.
Chapter PDF
References
L.M. Censier and P. Feautrier, “A new solution to coherence problems in multicache systems,” IEEE Trans. Computers, vol. C-27, pp. 1112–1118, Dec. 1978.
W.-D. Weber and A. Gupta, “Analysis of cache invalidation patterns in shared-memory multiprocessors,” in 3rd International Conference on Architectural Support for Programming Languages and Operating Systems, pp. 243–256, 1989.
A. Agarwal, R. Simoni, J. Hennessy, and M. Horowitz, “An evaluation of directory schemes for cache coherence,” in 15th Annual International Symposium on Computer Architecture, pp. 280–289, 1988.
S.C. Woo, M. Ohara, E. Torrie, J.P. Singh, and A. Gupta, “The SPLASH-2_programs: Characterization and methodological considerations,” in 22nd Annual International Symposium on Computer Architecture, pp. 24–36, 1995.
S. Goldschmidt, Simulation of Multiprocessors: Accuracy and Performance. PhD thesis, Stanford University, 1993.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
HoPark, C., HyukChoi, J., Park, K., Park, D. (1999). An Adaptive Limited Pointers Directory Scheme for Cache Coherence of Scalable Multiprocessors. In: Amestoy, P., et al. Euro-Par’99 Parallel Processing. Euro-Par 1999. Lecture Notes in Computer Science, vol 1685. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-48311-X_105
Download citation
DOI: https://doi.org/10.1007/3-540-48311-X_105
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-66443-7
Online ISBN: 978-3-540-48311-3
eBook Packages: Springer Book Archive