Advertisement

Performance Modeling of a Coarse Grain Dataflow Machine

  • Cornelia Riefers
  • Rainer Feix
Conference paper
Part of the Informatik-Fachberichte book series (INFORMATIK, volume 286)

Abstract

This paper deals with the formulation of an (extended) queueing network model for the performance evaluation of a coarse grain dataflow machine. The model comprises dynamic load balancing features which require a simulative approach for the model evaluation. The average parallelism of the dataflow graph is related to the population of the proposed model which is validated by comparing the simulation results to the measurements obtained from a prototype implementation. An exhaustive experimental study identifies the potential bottlenecks and the performance limits of the investigated dataflow machine.

Keywords

data flow machine coarse grain parallelism dynamic load balancing performance evaluation (extended) queueing network models simulation 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    Arvind, D.E. Culler: Dataflow Architectures. In: Annual Reviews in Computer Science (1988) 1, 225–253CrossRefGoogle Scholar
  2. [2]
    Arvind, R. Nikisch: Executing a program on the MIT tagged-token dataflow architecture. In: Parallel Architectures and Languages Europe, Vol. 2 (Lecture Notes in Computer Science 259), J.W. de Bakker, A.J. Nijman, P.C. Trelcavcn (eds.), Springer-Verlag (1987), 1–29Google Scholar
  3. [3]
    P. Barahona, J.R. Gurd: Simulated Performance of the Manchester Multi-Ring Dataflow Machine. In: Parallel Computing (1985), Elsevier Science Publisher North-Holland (1986), 419–424Google Scholar
  4. [4]
    J.B. Dennis, C.K. Leung, D.P. Misunas: A highly parallel processor using a dataflow machine language. Technical Report CSG Memo 134–1, Laboratory for Computer Science, Massachusetts Institute of Technology (Cambridge, 1979)Google Scholar
  5. [5]
    G. Färber (ed.): Bussysteme — Parallele und serielle Bussysteme in Theorie und Praxis. Oldenbourg (München, Wien, 1984)Google Scholar
  6. [6]
    R. Feix, M. Jobmann: MAOS — Model Analysis and Optimization System. Ultrix 2.2 User Guide Version 1.0. Technical Report 137/89. Department of Computer Science, University of Hamburg (1989)Google Scholar
  7. [7]
    P. Friedrich, E. Glück-Hiltrop, M. Ramlow, et al.: High Level Language. ESP-415E-STO-056, Deliverable 8 of Stollmann GmbH in ESPRIT Project 415 (Hamburg, 1989)Google Scholar
  8. [8]
    J.R. Gurd, C.C. Kirkham, I. Watson: The Manchester Prototype Data Flow Machine. In: Communications of the ACM 28 (1985)1,34–52Google Scholar
  9. [9]
    E. Glück-Hiltrop, M. Ramlow, U. Schürfeld: The Stollmann Data Flow Machine. In: Parallel Architectures and Languages Europe 1 (Lecture Notes in Computer Science 365), E. Odijk, M. Rem, J.-C. Syre (eds.), Springer-Verlag (1989), 433–457Google Scholar
  10. [10]
    D. Ghosal, L.N. Bhuyan: Performance Evaluation of a Dataflow Architecture. In: IEEE Transactions on Computers 39 (1990)5,615–627CrossRefGoogle Scholar
  11. [11]
    D. Ghosal, S.K. Tripathi, L.N. Bhuyan: Analysis of Computation-Communication Issues In Dynamic Dataflow Architectures. In: ACM SIGARCH 17 (1989) 3,325–333CrossRefGoogle Scholar
  12. [12]
    J.R. Gurd: Dataflow Computers and Languages. ESPRIT Summer School on Future Parallel Computers, Università di Pisa (1986)Google Scholar
  13. [13]
    A.R. Hurson, B. Lee, B. Shirazi: Hybrid Structure: A Scheme for Handling Data Structures in a Data Flow Environment. In: Parallel Architectures and Languages Europe 1 (Lecture Notes in Computer Science 365), E. Odijk, M. Rem, J.-C. Syre (eds.), Springer-Verlag (1989), 433–457Google Scholar
  14. [14]
    F. Hutner, R. Holzner: Architektur,Programmierung und Leistungsbewertung des MIT-Datenflußrechners. In: Informatik-Spektrum 12 (1989) 3,147–157Google Scholar
  15. [15]
    T. Jipp, P. Friedrich, H. Oldach, et al.: Definition of Principles and Basic Language. ESP-415E-STO-022, Deliverable 4 of Stollmann GmbH in ESPRIT Project 415 (Hamburg, 1987)Google Scholar
  16. [16]
    M. Jobmann: ILM AOS — Eine Sprache zur Formulierung von Rechensystemmodellen. Technical Report 91/82, Department of Computer Science, University of Hamburg (1982)Google Scholar
  17. [17]
    M. Jobmann: Modellbildung und -analyse von Rechensystemen mit Hilfe des Programmsystems MAOS. In: Messung, Modellierung, Bewertung von Rechensystemen (Informatik-Fachberichte 110), H. Beilner (ed.), Springer-Verlag (1985), 51–64Google Scholar
  18. [18]
    M. Jöhnk, U. Schürfeld: The Stollmann Data Flow Machine. In: Vernetzte und komplexe Informatik-Systeme (Informatik-Fachberichte 189), B. Wolfinger (ed.), Springer-Verlag (1989), 6–19Google Scholar
  19. [19]
    S. Lavenberg (ed.): Computer Performance Modeling Handbook. Academic Press (New York, 1983)zbMATHGoogle Scholar
  20. [20]
    E.D. Lazowska, J. Zahorjan, K.C. Sevcik: Computer System Performance Evaluation Using Queueing Network Models. In: Annual Reviews in Computer Science (1986), 107–137Google Scholar
  21. [21]
    A. Pias: LAU System Architecture: A Parallel Data Driven Processor Based On Single Assignment. In: Proceedings of the 1976 International Conference on Parallel Processing (1976), 293–302Google Scholar
  22. [22]
    C.H. Sauer, K.M. Chandy: Computer Systems Performance Modeling. Prentice-Hall (Englewood Cliffs, 1981)Google Scholar
  23. [23]
    G. Schäffler: A Survey of Dataflow Languages. ESP-415E-STO-010, Deliverable 1 of Slollmann GmbH in ESPRIT Project 415 (Hamburg, 1986)Google Scholar
  24. [24]
    Stollmann GmbH: SRTX Documentation and Reference Manual. Stollmann GmbH (Hamburg, 1987)Google Scholar
  25. [25]
    P.C. Treleaven, D.R. Brownbridge, R.P. Hopkins: Data Driven and Demand Driven Computer Architectures. In: ACM Computing Surveys 14 (1982) 1,93–139CrossRefGoogle Scholar
  26. [26]
    A.H. Veen: Dataflow Machine Architectures. In: ACM Computing Surveys 18 (1986) 4,365–393CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1991

Authors and Affiliations

  • Cornelia Riefers
    • 1
  • Rainer Feix
    • 1
  1. 1.Fachbereich InformatikUniversität HamburgHamburg 50Deutschland

Personalised recommendations