Holistic Hardware Counter Performance Analysis of Parallel Programs

@inproceedings{wylie_et_al:DSP:2006:507, abstract = {The KOJAK toolkit has been augmented with refined hardware performance counter support, including more convenient measurement specification, additional metric derivations and hierarchical structuring, and an extended algebra for integrating multiple experiments. Comprehensive automated analysis of a hybrid OpenMP/MPI parallel program, the ASC Purple sPPM benchmark, is demonstrated with performance experiments on equisized POWER4-II-based IBM Regatta p690+ cluster, Opteron-based Cray XD1 cluster and UltraSPARC-IV-based Sun Fire E25000 systems. Automatically assessed communication and synchronisation performance properties, combined with a rich set of measured and derived counter metrics, provide a holistic analysis context and facilitate multi-platform comparison.}, author = {Wylie, Brian J. N. and Mohr, Bernd and Wolf, Felix }, booktitle = {Automatic Performance Analysis}, citeulike-article-id = {2435309}, editor = {Gerndt, Hans M. and Malony, Allen and Miller, Barton P. and Nagel, Wolfgang }, keywords = {mpi, mpi\_tools, performance\_analysis}, number = {05501}, posted-at = {2008-02-27 10:59:06}, priority = {0}, publisher = {Internationales Begegnungs- und Forschungszentrum fuer Informatik (IBFI), Schloss Dagstuhl, Germany}, series = {Dagstuhl Seminar Proceedings}, title = {Holistic Hardware Counter Performance Analysis of Parallel Programs}, url = {http://drops.dagstuhl.de/opus/volltexte/2006/507/pdf/05501.WylieBrian.Paper.507.pdf}, year = {2006} }

TY - CONF ID - wylie_et_al:DSP:2006:507 L3 - citeulike-article-id:2435309 TI - Holistic Hardware Counter Performance Analysis of Parallel Programs T2 - Automatic Performance Analysis SE - Dagstuhl Seminar Proceedings IS - 05501 PB - Internationales Begegnungs- und Forschungszentrum fuer Informatik (IBFI), Schloss Dagstuhl, Germany N2 - The KOJAK toolkit has been augmented with refined hardware performance counter support, including more convenient measurement specification, additional metric derivations and hierarchical structuring, and an extended algebra for integrating multiple experiments. Comprehensive automated analysis of a hybrid OpenMP/MPI parallel program, the ASC Purple sPPM benchmark, is demonstrated with performance experiments on equisized POWER4-II-based IBM Regatta p690+ cluster, Opteron-based Cray XD1 cluster and UltraSPARC-IV-based Sun Fire E25000 systems. Automatically assessed communication and synchronisation performance properties, combined with a rich set of measured and derived counter metrics, provide a holistic analysis context and facilitate multi-platform comparison. KW - mpi KW - mpi_tools KW - performance_analysis AU - Wylie, Brian AU - Mohr, Bernd AU - Wolf, Felix A2 - Gerndt, Hans A2 - Malony, Allen A2 - Miller, Barton A2 - Nagel, Wolfgang PY - 2006/// UR - http://drops.dagstuhl.de/opus/volltexte/2006/507/pdf/05501.WylieBrian.Paper.507.pdf ER -