View in EDS

Holistic Hardware Counter Performance Analysis of Parallel Programs

Saved in:
Bibliographic Details
Title: Holistic Hardware Counter Performance Analysis of Parallel Programs
Authors: Wylie, Brian J. N., Mohr, Bernd, Wolf, Felix
Contributors: Brian J. N. Wylie and Bernd Mohr and Felix Wolf
Publisher Information: Schloss Dagstuhl – Leibniz-Zentrum für Informatik
Publication Year: 2006
Collection: DROPS - Dagstuhl Research Online Publication Server (Schloss Dagstuhl - Leibniz Center for Informatics )
Subject Terms: Parallel program execution tracing, hardware performance counters, automated structured analysis, performance comparison
Description: 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.
Document Type: article in journal/newspaper
conference object
File Description: application/pdf
Language: English
Relation: Is Part Of Dagstuhl Seminar Proceedings, Volume 5501, Automatic Performance Analysis (2006); https://drops.dagstuhl.de/entities/document/10.4230/DagSemProc.05501.4
DOI: 10.4230/DagSemProc.05501.4
Availability: https://doi.org/10.4230/DagSemProc.05501.4
https://nbn-resolving.org/urn:nbn:de:0030-drops-5071
https://drops.dagstuhl.de/entities/document/10.4230/DagSemProc.05501.4
Rights: https://creativecommons.org/licenses/by/4.0/legalcode
Accession Number: edsbas.1F21D80E
Database: BASE
Description
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.
DOI:10.4230/DagSemProc.05501.4