ОБЗОР МЕТОДОВ АВТОМАТИЗИРОВАННОЙ ОЦЕНКИ ЭФФЕКТИВНОСТИ ВЫПОЛНЕНИЯ ПАРАЛЛЕЛЬНЫХ ПРОГРАММ
Аннотация
According to the report of the USA High-end computing revitalization task force (HECRTF) sustained performance of contemporary parallel machines is less than 10 %. To gain acceptable parallel performance programmers often use performance analysis instruments. Most of them entirely rely on developer in finding parallel performance problems but some automate this process more or less. This paper is dedicated to such kind of instruments.
Ключевые слова
004.051
Список литературы
1. Federal Plan for High-End Computing: Report of the High-End Computing Revitalization Task Force (HECRTF) [Электронный ресурс], 2004. URL: http://www.nitrd.gov/pubs/2004_hecrtf/20040702_hecrtf.pdf (дата обращения: 2.03.09).
2. Leko, A. Performance Analysis Strategies [Электронный ресурс]. URL: <http://www.hcs.ufl.edu/prj/ upcgroup/upcperf/documents/20050302AnalysisDraft.pdf> (дата обращения: 2.03.09).
3. Miller, B. PIPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Transactions on Parallel and Distributed Systems, IEEE Computer Society / В. Р. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. S. Lim, T. Torzewski. - 1990. - С. 206 - 217.
4. Vetter, J. Performance Analysis of Distributed Applications Using Automatic Classification of Communication Inefficiencies / J. Vetter // Труды XIV International Conference on Supercomputing, ACM Press. - 2000. - С. 245 - 254.
5. Truong, H.-L.. On using SCALEA for Performance Analysis of Distributed and Parallel Programs / H.-L. Truong, T. Fahringer, G. Madsen, A. D. Malony, H. Moritsch, S. Shende // Труды конференции SC2001, ACM Press, 2001. - С. 37.
6. Truong, H.-L. Scalea - a Performance Analysis System for Distributed and Parallel Programs / H.-L. Truong, T. Fahringer // Труды VIII International Euro-Par Conference, Springer, 2002. - С. 75 - 85.
7. Amdahl, D. Validity of the Single-processor Approach to Achieving Large-scale Computer Capabilities / D. Amdahl // Труды конференции AFIPS, AFIPS Press, 1967. - С. 483 - 485.
8. Wolf, F. Automatic Search for Patterns of Inefficient Behavior in Parallel Applications / F. Wolf, B. Mohr, J. Dongarra, S. Moore. [Электронный ресурс]. URL: <http://www.netlib.org/netlib/utk/people/ JackDongarra/PAPERS/auto-apart-2005.pdf> (дата обращения: 2.03.09).
9. Wolf, F. Automatic Performance Analysis of MPI Applications Based on Event Traces / F. Wolf, B. Mohr. // Труды VI International Euro-Par Conference, Springer. - 2000. - C. 123. - 132.
10. Shende, S. S.. The Role of Instrumentation and Mapping in Performance Measurement: докторская диссертация, University of Oregon, 2001.
11. DeRose, L. The Dynamic Probe Class Library - An Infrastructure for Developing Instrumentation for Performance Tools. / L. DeRose, Т. Hoover, J. Hollingsworth // Труды XV International Parallel and Distributed Processing Symposium, IEEE Computer Society, 2001.
12. Mohr, B. Design and Prototype of a Performance Tool Interface for OpenMP / B. Mohr, A. Malony, S. Shende, F. Wolf // Journal of Supercomputing, Kluwer Academic Publishers, 2002. - С. 105 - 128.
13. F. Wolf. EARL - API Documentation. [Электронный ресурс]. URL: http://www.fzjuelich.de/jsc/ datapool/Kojak/earl-2.0.pdf (дата обращения: 2.03.09).
14. F. Wolf. Automatic Performance Analysis on Parallel Computers with SMP Nodes: докторская диссертация, RWTH Aachen University, 2003.
Для цитирования:
Андреев Н.Е. ОБЗОР МЕТОДОВ АВТОМАТИЗИРОВАННОЙ ОЦЕНКИ ЭФФЕКТИВНОСТИ ВЫПОЛНЕНИЯ ПАРАЛЛЕЛЬНЫХ ПРОГРАММ. Вестник Кемеровского государственного университета. 2009;(2):5-13.