OVERHEAD COMPENSATION IN PERFORMANCE PROFILING
Abstract
Measurement-based profiling introduces intrusion in program execution. Intrusion effects can be mitigated by compensating for measurement overhead. Techniques for compensation analysis in performance profiling are presented and their implementation in the TAU performance system described. Experimental results on the NAS parallel benchmarks demonstrate that overhead compensation can be effective in improving the accuracy of performance profiling. However, parallel time profiling requires the execution delay introduced by measurement on individual processes to be communicated between processes when they interact. Parallel execution scenarios are given to model the effects and to determine the analysis procedures to be applied online.
References
- , International Conference on Computational Science,
LNCS 2074, eds.V. Alexandrov (Springer, 2001) pp. 751–760. Google Scholar - F. Wolf and B. Mohr, "Automatic Performance Analysis of SMP Cluster Applications," Technical Report IB 2001-05, Research Centre Juelich, 2001 . Google Scholar
- Software Practice and Experience 1, 105 (1971). Crossref, Google Scholar
- L. De Rose, "The Hardware Performance Monitor Toolkit," Euro-Par Conference, 2001 . Google Scholar
- T. Fahringer and C. Seragiotto, "Experience with Aksum: A Semi-Automatic Multi-Experiment Performance Analysis Tool for Parallel and Distributed Applications," Workshop on Performance Analysis and Distributed Computing, 2002 . Google Scholar
- , Distributed and Parallel Systems, From Instruction Parallelism to Cluster Computing, Third Workshop on Distributed and Parallel Systems (DAPSYS 2000), eds.
G. Kotsis and P. Kacsuk (Kluwer, 2000) pp. 37–46. Google Scholar - P. Mucci, "Dynaprof," http://www.cs.utk.edu/~mucci/dynaprof . Google Scholar
- D. Reed, L. DeRose, and Y. Zhang, "SvPablo: A Multi-Language Performance Analysis System," International Conference on Performance Tools, pp. 352–355, September 1998 . Google Scholar
- S. Graham, P. Kessler, and M. McKusick, "gprof: A Call Graph Execution Profiler," SIGPLAN Symposium on Compiler Construction, pp. 120–126, June 1982 . Google Scholar
- IBM, "Profiling Parallel Programs with Xprofiler," IBM Parallel Environment for AIX: Operation and Use, Volume 2 . Google Scholar
- C. Janssen, "The Visual Profiler," http://aros.ca.sandia.gov/~cljanss/perf/vprof/ . Google Scholar
- Journal of Supercomputing 23, 81 (2002). Crossref, ISI, Google Scholar
- Unix Programmer's Manual, "prof command," Section 1, Bell Laboratories, Murray Hill, NJ, January 1979 . Google Scholar
- A. Malony, "Performance Observability," Ph.D. thesis, University of Illinois, Urbana-Champaign, 1991 . Google Scholar
- D. Bailey, T. Harris, W. Saphir, R. van der Wijngaart, A. Woo, M. Yarrow, "The NAS Parallel Benchmarks 2.0," Technical Report NAS-95-020, NASA Ames Research Center, 1995 . Google Scholar
- International Journal of High Performance Computing Applications 14(3), 189 (2000). Crossref, ISI, Google Scholar
- M. Zagha, B. Larson, S. Turner, and M. Itzkowitz, "Performance Analysis Using the MIPS R10000 Performance Counters," Supercomputing Conference, November 1996 . Google Scholar
- R. Hall, "Call Path Profiling," International Conference on Software Engineering, pp. 296–306, 1992 . Google Scholar
L. De Rose and F. Wolf , Euro-Par Conference,LNCS 2400 (Springer, 2002) pp. 167–176. Google Scholar- D. Kranzlmüller, R. Reussner, and C. Schaubschläger, "Monitor Overhead Measurement with SKaMPI," EuroPVM/MPI Conference, LNCS 1697, pp. 43–50, 1999 . Google Scholar
- Alain Fagot and Jacques Chassin de Kergommeaux, "Systems Assessment of the Overhead of Tracing Parallel Programs," Euromicro Workshop on Parallel and Distributed Processing, pp. 179–186, 1996 . Google Scholar
- J. Hollingsworth and B. Miller, "An Adaptive Cost System for Parallel Program Instrumentation," Euro-Par Conference, Volume I, pp. 88–97, August 1996 . Google Scholar
- , Performance Analysis and Grid Computing, eds.
V. Getov (Kluwer, Norwell, MA, 2003) pp. 129–144. Google Scholar - IEEE Transactions on Parallel and Distributed Systems 3(4), 433 (1992). Crossref, ISI, Google Scholar
- A. Malony and D. Reed, "Models for Performance Perturbation Analysis," ACM/ONR Workshop on Parallel and Distributed Debugging, pp. 1–12, May 1991 . Google Scholar
- A. Malony, "Event Based Performance Perturbation: A Case Study," Principles and Practices of Parallel Programming (PPoPP), pp. 201–212, April 1991 . Google Scholar
-
W. Williams , T. Hoel and D. Pase , Programming Environments for Massively Parallel Distributed Systems ( North-Holland , 1994 ) . Google Scholar - S. Sarukkai and A. Malony, "Perturbation Analysis of High-Level Instrumentation for SPMD Programs," Principles and Practices of Parallel Programming (PPoPP), pp. 44–53, May 1993 . Google Scholar
-
J. Vetter , ACM SIGMETRICS Joint International Conference on Measurement and Modeling of Computer Systems ( ACM , 2002 ) . Google Scholar - G. Bronevetsky, D. Marques, K. Pingali, and P. Stodghill, "Collective Operations in an Application-level Fault Tolerant MPI System," International Conference on Supercomputing (ICS), 2003 . Google Scholar
- G. Bronevetsky, D. Marques, K. Pingali, and P. Stodghill, "Automated Application-level Checkpointing of MPI Programs," Principles and Practice of Parallel Programming (PPoPP), 2003 . Google Scholar


