Abstract
Coherent-cache shared-memory architectures often give disappointing performance which can be alleviated by manual tuning. We describe a new trace analysis tool, Clarissa, which helps diagnose problems and pinpoint their causes. Unusually, Clarissa works by analysing potential contention, instead of measuring predicted contention by simulating a specific memory system design. This is important because, after tuning, the software will be executed on different inputs and different configurations. The goal is to produce a program with robustly good performance. This paper explains the principle behind cautious trace analysis, describes our implementation, and presents our experience of using the tool.
Chapter PDF
Similar content being viewed by others
Keywords
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.
References
Mats Brorsson. SM-prof: A tool to visualise and find cache coherence performance bottlenecks in multiprocessor programs. In Proceedings of the ACM SIGMETRICS and Performance '95, pages 178–187, May 1995.
Susan J. Eggers and Randy H. Katz. A characterisation of sharing in parallel programs and its application to coherency protocol evaluation. 15th Annual International Symposium on Computer Architecture, Honolulu, May, in Computer Architecture News, 16(2):373–382, May 1988.
M. Martonosi, A. Gupta, and T. Anderson. Tuning memory performance of sequential and parallel programs. IEEE Computer, 28(4):32–40, April 1995.
Jaswinder Pal Singh, Wolf-Dietrich Weber, and Anoop Gupta. SPLASH: Stanford parallel applications for shared-memory. Computer Architecture News, 20(1):5–44 March 1992.
Sarah A. M. Talbot. Performance tuning of programs for shared-memory multi-processors. Master's thesis, Department of Computing, Imperial College, London U.K., 1995.
B. A. Tanyi. Iterative Solution of the Incompressible Navier-Stokes Equations on a Distributed Memory Parallel Computer. PhD thesis, UMIST, 1993.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1996 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Talbot, S.A.M., Bennett, A.J., Kelly, P.H.J. (1996). Cautious, machine-independent performance tuning for shared-memory multiprocessors. In: Bougé, L., Fraigniaud, P., Mignotte, A., Robert, Y. (eds) Euro-Par'96 Parallel Processing. Euro-Par 1996. Lecture Notes in Computer Science, vol 1123. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-61626-8_13
Download citation
DOI: https://doi.org/10.1007/3-540-61626-8_13
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-61626-9
Online ISBN: 978-3-540-70633-5
eBook Packages: Springer Book Archive