Evaluating similarity-based trace reduction techniques for scalable performance analysis

Event traces are required to correctly diagnose a number of performance problems that arise on today's highly parallel systems. Unfortunately, the collection of event traces can produce a large volume of data that is difficult, or even impossible, to store and analyze. One approach for compressing a trace is to identify repeating trace patterns and retain only one representative of each pattern. However, determining the similarity of sections of traces, i.e., identifying patterns, is not straightforward. In this paper, we investigate pattern-based methods for reducing traces that will be used for performance analysis. We evaluate the different methods against several criteria, including size reduction, introduced error, and retention of performance trends, using both benchmarks with carefully chosen performance behaviors, and a real application.

[1]  Barton P. Miller,et al.  Dynamic program instrumentation for scalable performance tools , 1994, Proceedings of IEEE Scalable High Performance Computing Conference.

[2]  James S. Walker,et al.  A Primer on Wavelets and Their Scientific Applications , 1999 .

[3]  Bernd Mohr,et al.  Automatic analysis of inefficiency patterns in parallel applications , 2007, Concurr. Comput. Pract. Exp..

[4]  Laxmikant V. Kalé,et al.  Scaling Molecular Dynamics to 3000 Processors with Projections: A Performance Analysis Case Study , 2003, International Conference on Computational Science.

[5]  B. G. Martínez,et al.  Historia general de México , 1978 .

[6]  Matthias Hauswirth,et al.  Automating vertical profiling , 2005, OOPSLA '05.

[7]  Ada Wai-Chee Fu,et al.  Efficient time series matching by wavelets , 1999, Proceedings 15th International Conference on Data Engineering (Cat. No.99CB36337).

[8]  F. Petrini,et al.  The Case of the Missing Supercomputer Performance: Achieving Optimal Performance on the 8,192 Processors of ASCI Q , 2003, ACM/IEEE SC 2003 Conference (SC'03).

[9]  Martin Schulz,et al.  Preserving time in large-scale communication traces , 2008, ICS '08.

[10]  W. Call.  Plan Puebla-Panama , 2002 .

[11]  Petra Perner,et al.  Data Mining - Concepts and Techniques , 2002, Künstliche Intell..

[12]  Dieter Kranzlmüller,et al.  Pattern Matching of Collective MPI Operations , 2004, PDPTA.

[13]  Jerry C. Yan,et al.  Performance Data Gathering and Representation from Fixed-Size Statistical Data , 1997 .

[14]  Patricia J. Teller,et al.  A systematic multi-step methodology for performance analysis of communication traces of distributed applications based on hierarchical clustering , 2006, Proceedings 20th IEEE International Parallel & Distributed Processing Symposium.

[15]  Jeffrey S. Vetter,et al.  Dynamic statistical profiling of communication activity in distributed applications , 2002, SIGMETRICS '02.

[16]  A. Jensen,et al.  Ripples in Mathematics - The Discrete Wavelet Transform , 2001 .

[17]  J PritchardD.  Concurrency: Practice and Experience , 1991 .

[18]  Daniel P. Spooner,et al.  Performance feature identification by comparative trace analysis , 2006, Future Gener. Comput. Syst..

[19]  Dieter Kranzlmüller,et al.  Event graph visualization for debugging large applications , 1996, SPDT '96.

[20]  Jesús Labarta,et al.  A dynamic periodicity detector: application to speedup computation , 2001, Proceedings 15th International Parallel and Distributed Processing Symposium. IPDPS 2001.

[21]  Danny B. Lange,et al.  Object-Oriented Program Tracing and Visualization , 1997, Computer.

[22]  Laura Carrington,et al.  A performance prediction framework for scientific applications , 2003, Future Gener. Comput. Syst..

[23]  J. C. Yan,et al.  Constructing Space-Time Views from Fixed Size Trace Files - Getting the Best of Both Worlds , 1997, PARCO.

[24]  F. Mueller,et al.  Scalable Compression and Replay of Communication Traces in Massively P arallel E nvironments , 2007, 2007 IEEE International Parallel and Distributed Processing Symposium.

[25]  Karsten Schwan,et al.  Falcon: on-line monitoring and steering of large-scale parallel programs , 1995, Proceedings Frontiers '95. The Fifth Symposium on the Frontiers of Massively Parallel Computation.

[26]  Karen L. Karavanic,et al.  Towards Scalable Event Tracing for High End Systems , 2007, HPCC.

[27]  Philip C. Roth,et al.  Real-Time Statistical Clustering for Event Trace Reduction , 1997, Int. J. High Perform. Comput. Appl..

[28]  Jack J. Dongarra,et al.  An algebra for cross-experiment performance analysis , 2004, International Conference on Parallel Processing, 2004. ICPP 2004..

[29]  Jesús Labarta,et al.  Automatic Phase Detection of MPI Applications , 2007, PARCO.

[30]  Timothy Sherwood,et al.  Wavelet-based phase classification , 2006, 2006 International Conference on Parallel Architectures and Compilation Techniques (PACT).

[31]  Bernd Mohr,et al.  A test suite for parallel performance analysis tools , 2007, Concurr. Comput. Pract. Exp..

[32]  Martin Schulz,et al.  Scalable compression and replay of communication traces in massively parallel environments , 2006, SC.

[33]  Martin Schulz,et al.  Scalable load-balance measurement for SPMD codes , 2008, 2008 SC - International Conference for High Performance Computing, Networking, Storage and Analysis.

[34]  Andreas Knüpfer,et al.  A New Data Compression Technique for Event Based Program Traces , 2003, International Conference on Computational Science.

[35]  Julio César López-Hernández,et al.  Stardust: tracking activity in a distributed storage system , 2006, SIGMETRICS '06/Performance '06.

[36]  Angewandte Mathematik,et al.  Knowledge Specification for Automatic Performance Analysis APART Technical Report Revised Version , 2001 .

[37]  Laxmikant V. Kalé,et al.  Towards scalable performance analysis and visualization through data reduction , 2008, 2008 IEEE International Symposium on Parallel and Distributed Processing.

[38]  Robert J. Fowler,et al.  Scalable methods for monitoring and detecting behavioral equivalence classes in scientific codes , 2008, 2008 IEEE International Symposium on Parallel and Distributed Processing.