Performance evaluation of fork and join synchronization primitives
- 109 Downloads
The paper presents a performance model of fork and join synchronization primitives. The primitives are used in parallel programs executed on distributed systems. Three variants of the execution of parallel programs with fork and join primitives are considered and queueing models are proposed to evaluate their performance on a finite number of processors. Synchronization delays incurred by the programs are represented by a state-dependent server with service rate depending on a particular synchronization scheme. Closed form results are presented for the two processor case and a numerical method is proposed for many processors. Fork-join queueing networks having more complex structure i.e., processors arranged in series and in parallel, are also analyzed in the same manner. The networks can model the execution of jobs with a general task precedence graph corresponding to a nested structure of the fork-join primitives. Some performance indices of the parallel execution of programs are studied. The results show that the speedup which can be obtained theoretically in a parallel system may be decreased significantly by synchronization constraints.
KeywordsComputational Mathematic Performance Model Closed Form Performance Index Service Rate
Unable to display preview. Download preview PDF.
- 1.Baccelli, F., Makowski, A.M.: Simple Computable Bounds for the Fork-Join Queue. Proc. John Hopkins Conf. Information Sciences and Systems, John Hopkins University, 1985Google Scholar
- 2.Baccelli, F., Massey, W.A.: Series-Parallel, Fork-Join Queueing Networks and their Stochastic Ordering. Bell Labs. Technical Memorandum 11211-851101-36, November 1985 Murray HillGoogle Scholar
- 6.Conway, M.: A Multiprocessor System Design. In: Proc. AFIPS 1963 Fall Joint Computer Conference, 1963 Vol. 24, pp. 139–146. Baltimore, Maryland: Spartan BooksGoogle Scholar
- 10.Dijkstra, E.W.: Co-operating Sequential Processes. In: Programing Languages, F. Genuys, (ed.). pp. 43–112. New York: Academic Press 1968Google Scholar
- 14.Kanakia, H., Tobagi, F.A.: Theoretical Results on Distributing Processing with Limited Computing Resources. Stanford University, SEL Technical Report No. 85-273, April 1985Google Scholar
- 15.Kleinrock, L.: Distributed Systems. Comm. Assoc. Comput. Mach. 28, 1200–1213 (1985)Google Scholar
- 16.Miller, B.P.: Performance Characterization of Distributed Programs. Ph.D. Dissertation, Report No. UCB/CSD84/197. University of California, Berkeley, August 1984Google Scholar
- 17.Nelson, R., Tantawi, A.N.: Approximate Analysis of Fork/Join Synchronization in Parallel Queues, IBM Research Report RC 11481, to be published in IEEE Trans. on Comput. October 1985Google Scholar
- 19.Potier, D., Veran, M.: The Markovian Solver of QNAP2 and Examples. In: Proc. Int. Seminar Computer Networking and Performance Evaluation, T. Hasegawa (ed.) 1985Google Scholar
- 20.Sevcik, K.C., Levy, A.I., Tripathi, S.K., Zahorjan, J.L.: Improving Approximations of Aggregated Queuing Network Subsystems. In: Computer Performance. K.M. Chandy, M. Reiser (eds.) pp. 1–22, Amsterdam: North-Holland, 1977Google Scholar
- 22.Veran, M., Potier, D.: QNAP2: a Portable Environment for Queueing Systems Modelling. In: Modelling Techniques and Tools for Performance Analysis, D. Potier (ed.) Amsterdam: North-Holland 1985Google Scholar