Electrotechnical and Computer Engineering
Vol. 35 No. 11 (2020): Proceedings of the Faculty of Technical Sciences
A TOOL FOR MEASURING THE PERFORMANCE OF OPENMP PROGRAMS
Abstract
The hpcrun has been extended with OMPT interface support in order to facilitate performance analysis of OpenMP programs’ source code. The usage of OMPT call back functions may lead to drastic increase in overhead while measuring the performance of OpenMP programs with fine-grained parallelism.
References
[1] https://www.openmp.org/wp-content/uploads/OpenMP-API-Specification-5.0.pdf (pristupljeno u novembru 2018.)
[2] N. Froyd, J. Mellor-Crummey, R. Fowler. Low overhead call path profiling of unmodified, optimized code. In Proceedings of the 19th annual international conference on Supercomputing, pages 81–90, 2005.
[3] https://www.openmp.org/wp-content/uploads/ompt-tr2.pdf (pristupljeno u julu 2017.)
[4] L. Adhianto, et all. Hpctoolkit: Tools for performance analysis of optimized parallel programs. Concurrency and Computation: Practice and Experience, 22(6):685–701, 2010.
[5] X. Liu, J. Mellor-Crummey, and M. Fagan. A new approach for performance analysis of openmp programs. In Proceedings of the 27th international ACM conference on International conference on supercomputing, pages 69–80, 2013.
[6] https://www.openmp.org/wp-content/uploads/openmp-TR9.pdf (pristupljeno u martu 2020.)
[7] https://www.spec.org/omp2012/ (pristupljeno u junu 2020.)
[8] https://www.epcc.ed.ac.uk/research/computing/-performance-characterisation-and-benchmarking/epcc-openmp-micro-benchmark-suite (pristupljeno u junu 2018.)
[9] A. E. Eichenberger, K. O’Brien. Experimenting with low-overhead openmp runtime on ibm blue gene/q. IBM Journal of Research and Development, 57(1/2):8–1, 2013.