Abstract
The Data Diffusion Machine (DDM) is a scalable shared virtual memory multiprocessor in which the location of a datum in the machine is completely decoupled from its virtual address. In particular, there is no distinguished home location where a datum must normally reside. Instead data migrates automatically to where it is needed, reducing access times and traffic.
The hardware organisation consists of a hierarchy of buses and data controllers linking an arbitrary number of processors each having a large set-associative memory. Each data controller has a set-associative directory containing status bits for data under its control. The controller supports remote data access by “snooping” on the buses above it and below it. The data access protocol it uses provides for the automatic migration, duplication and replacement of data while maintaining data coherency.
The machine is scalable in that there may be any number of levels in the hierarchy. Only a few levels are necessary in practice for a very large number of processors. Most memory requests are satisfied locally. Requests requiring remote access generally cause only a limited amount of traffic over a limited part of the machine, and are satisfied within a small time that is logarithmic to the number of processors. Although designed particularly to provide good support for the parallel execution of logic programs, the architecture is very general in that it does not assume any particular processor, language or class of application.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Masatoshi Sato Akira Matsumoto, Takayuki Nakagawa. Locally Parallel Cache Design. TR 327, ICOT, 1988.
Robert A Iannucci Arvind. Two Fundamental Issues in Multiprocessing: the Dataflow Solution. MIT/LCS/TM 241, MIT, 1983.
J.R. Goodman and P.J. Woest. The Wisconsin Multicube: a new large-scale cache-coherent multiprocessor. In Proceedings of the 15th Annual International Symposium on Computer Architecture, Honolulu, Hawaii, pages 442–431, IEEE, 1988.
Seif Haridi and Per Brand. Andorra Prolog-an integration of Prolog and committed choice languages. In International Conference on Fifth Generation Computer Systems 1988, ICOT, 1988.
M. Hermenegildo and P. McGehearty. Address Escaping and Reference Classification in the Design of a Cached, Multiple Cluster, Shared-Memory Architecture. PP-SRS-Technical Memo 12, MCC, 1987.
Ewing Lusk, David H. D. Warren, Seif Haridi, et al. The Aurora or-parallel Prolog system. In International Conference on Fifth Generation Computer Systems 1988, ICOT, 1988.
Gurindar S Sohi Mary K Venon, Rajeev Jog. Performance analysis of hierarchical cacheconsistent multiprocessors. In Conference Proceedings of International Seminar on Performance of Distributed and Parallel Systems, pages 111–126, 1988.
G.F. Pfister et al. The IBM Research Parallel Processor Prototype (RP3). In Proceedings of the 1985 International Conference on Parallel Processing, Chicago, IEEE, 1985.
David H. D. Warren and Seif Haridi. Data Diffusion Machine—a scalable shared virtual memory multiprocessor. In International Conference on Fifth Generation Computer Systems 1988, ICOT, 1988.
A. Wilson. Hierarchical cache/bus architecture for shared memory multiprocessor. Technical report ETR 86-006, Encore Computer Corporation, 1986.
Andrew W Wilson. Organization and Statistical Simulation of Hierarchical Multiprocessors. PhD thesis, CMU, 1985.
Rong Yang. Programming in Andorra-I. August 1988. Internal Report, Gigalips Project.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1989 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Haridi, S., Hagersten, E. (1989). The cache coherence protocol of the Data Diffusion Machine. In: Odijk, E., Rem, M., Syre, JC. (eds) PARLE '89 Parallel Architectures and Languages Europe. PARLE 1989. Lecture Notes in Computer Science, vol 365. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3540512845_29
Download citation
DOI: https://doi.org/10.1007/3540512845_29
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-51284-4
Online ISBN: 978-3-540-46183-8
eBook Packages: Springer Book Archive