Stefano Markidis Erwin Laure (Eds.) 9 5 7 Solving Software Challenges 8 S C for Exascale N L International Conference on Exascale Applications and Software, EASC 2014 Stockholm, Sweden, April 2–3, 2014 Revised Selected Papers 123 Lecture Notes in Computer Science 8759 Commenced Publication in 1973 Founding and Former Series Editors: Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen Editorial Board David Hutchison Lancaster University, Lancaster, UK Takeo Kanade Carnegie Mellon University, Pittsburgh, PA, USA Josef Kittler University of Surrey, Guildford, UK Jon M. Kleinberg Cornell University, Ithaca, NY, USA Friedemann Mattern ETH Zurich, Zürich, Switzerland John C. Mitchell Stanford University, Stanford, CA, USA Moni Naor Weizmann Institute of Science, Rehovot, Israel C. Pandu Rangan Indian Institute of Technology, Madras, India Bernhard Steffen TU Dortmund University, Dortmund, Germany Demetri Terzopoulos University of California, Los Angeles, CA, USA Doug Tygar University of California, Berkeley, CA, USA Gerhard Weikum Max Planck Institute for Informatics, Saarbrücken, Germany More information about this series at http://www.springer.com/series/7407 Stefano Markidis Erwin Laure (Eds.) (cid:129) Solving Software Challenges for Exascale International Conference on Exascale Applications and Software, EASC 2014 – Stockholm, Sweden, April 2 3, 2014 Revised Selected Papers 123 Editors Stefano Markidis Erwin Laure KTHRoyal InstituteofTechnology KTHRoyal InstituteofTechnology Stockholm Stockholm Sweden Sweden ISSN 0302-9743 ISSN 1611-3349 (electronic) Lecture Notesin ComputerScience ISBN 978-3-319-15975-1 ISBN 978-3-319-15976-8 (eBook) DOI 10.1007/978-3-319-15976-8 LibraryofCongressControlNumber:2015932683 LNCSSublibrary:SL1–TheoreticalComputerScienceandGeneralIssues SpringerChamHeidelbergNewYorkDordrechtLondon ©SpringerInternationalPublishingSwitzerland2015 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartofthe material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodologynow knownorhereafterdeveloped. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbookare believedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsortheeditors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissionsthatmayhavebeenmade. Printedonacid-freepaper SpringerInternationalPublishingAGSwitzerlandispartofSpringerScience+BusinessMedia (www.springer.com) Preface Exascale supercomputers will deliver an unprecedented computing power of 1018 floating point operations per second through extreme parallelism likely achieved from hybridcomputerarchitectures.Softwareandscientificapplicationsrunningonexascale supercomputer face the challenge of effectively exploiting this computing power. To address this challenge, many potentially disruptive changes are needed in software and applications. The Exascale Applications and Software Conference (EASC) brings together all developers and researchers involved in solving the software challenges oftheexascaleera.ThisvolumecollectedselectedcontributionsfromthesecondEASC inStockholmduringApril2–3,2014.Thevolumeisintendedforusebyresearchersand students of computer science and computational physics. In particular, the volume is verywellsuitedforusebydevelopersofparallelcodes,newprogrammingmodels,run- time systems, andtoolsfor petascale and exascalesupercomputers. Thisvolumeisorganizedintotwoparts.Thefirstseriesofarticlespresentsthenew developments and algorithms in large scientific applications from different scientific domains, such as biochemistry, computational fluid dynamics, and neutronics. In particular, these articles show how to exploit different levels of parallelism (vector instructions, intra-node and inter-node levels) on hybrid supercomputers in the molecular dynamics GROMACS and computational fluid dynamics Nek5000 codes. Innovative algorithms for reducing memory storage in Monte Carlo methods for neutronics and for improving the quality of sparse domain decomposition in lattice- Boltzmann methods are presented. The second part of the volume presents advance- mentsinsoftwaredevelopmentenvironmentsforexascale.Theperformancemodeling oftheHPX-5run-timesystemfortheLULESHproxyapplicationispresentedandthe co-design work is explained. One article analyzes the effect of system noise on dis- tributed applications at large scale. The new developments in the MUST MPI cor- rectnesscheckerandtheVAMPIRperformancemonitoringtoolarepresented.Finally, an article on the Vistle visualization tool for distributed environments concludes the volume. January 2015 Stefano Markidis Organization EASC 2014 was organized by the European Commission funded projects CRESTA (Grant Agreement No. 287703, cresta-project.eu), EPiGRAM (Grant Agreement No. 610598, epigram-project.eu), and by the Swedish e-Science Research Center SeRC (e-science.se) Steering Group Erwin Laure KTH Royal Institute of Technology, Sweden Stefano Markidis KTH Royal Institute of Technology, Sweden William D. Gropp University of Illinois at Urbana–Champaign, USA Mark Parsons EPCC, University of Edinburgh, Scotland, UK Lorna Smith EPCC, University of Edinburgh, Scotland, UK Bastian Koller HLRS, Germany Program Committee Erwin Laure KTH Royal Institute of Technology, Sweden Stefano Markidis KTH Royal Institute of Technology, Sweden William D. Gropp University of Illinois, USA Satoshi Matsuoka Tokyo Institute of Technology, Japan Mark Parsons EPCC, University of Edinburgh, Scotland, UK Lorna Smith EPCC, University of Edinburgh, Scotland, UK Daniel Holmes EPCC, University of Edinburgh, Scotland, UK Bastian Koller HLRS, Germany Pavan Balaji Argonne National Laboratory, USA Jed Brown Argonne National Laboratory, USA Robert Clay Sandia National Laboratories, USA Roberto Gioiosa Pacific Northwest National Laboratory, USA Katie Antypas NERSC, USA Leroy Drummond-Lewis Lawrence Berkeley National Laboratory, USA Alec Johnson Katholieke Universiteit Leuven, Belgium Sponsoring Institutions Cray Inc., Seattle, WA, USA Mellanox Technologies, Sunnyvale, CA, USA Contents Towards Exascale Scientific Applications Tackling Exascale Software Challenges in Molecular Dynamics Simulations with GROMACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Szilárd Páll, Mark James Abraham, Carsten Kutzner, Berk Hess, and Erik Lindahl Weighted Decomposition in High-Performance Lattice-Boltzmann Simulations: Are Some Lattice Sites More Equal than Others? . . . . . . . . . . . 28 Derek Groen, David Abou Chacra, Rupert W. Nash, Jiri Jaros, Miguel O. Bernabeu, and Peter V. Coveney Performance Analysis of a Reduced Data Movement Algorithm for Neutron Cross Section Data in Monte Carlo Simulations . . . . . . . . . . . . . . . . . . . . . 39 John R. Tramm, Andrew R. Siegel, Benoit Forget, and Colin Josey Nek5000 with OpenACC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Jing Gong, Stefano Markidis, Michael Schliephake, Erwin Laure, Dan Henningson, Philipp Schlatter, Adam Peplinski, Alistair Hart, Jens Doleschal, David Henty, and Paul Fischer Auto-tuning an OpenACC Accelerated Version of Nek5000. . . . . . . . . . . . . 69 Luis Cebamanos, David Henty, Harvey Richardson, and Alistair Hart Development Environment for Exascale Applications Towards Exascale Co-Design in a Runtime System. . . . . . . . . . . . . . . . . . . 85 Thomas Sterling, Matthew Anderson, P. Kevin Bohan, Maciej Brodowicz, Abhishek Kulkarni, and Bo Zhang Overcoming Asynchrony: An Analysis of the Effects of Asynchronous Noise on Nearest Neighbor Synchronizations. . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Adam Hammouda, Andrew Siegel, and Stephen Siegel Memory Usage Optimizations for Online Event Analysis. . . . . . . . . . . . . . . 110 Tobias Hilbrich, Joachim Protze, Michael Wagner, Matthias S. Müller, Martin Schulz, Bronis R. de Supinski, and Wolfgang E. Nagel Towards Detailed Exascale Application Analysis — Selective Monitoring and Visualisation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Jens Doleschal, Thomas William, Bert Wesarg, Johannes Ziegenbalg, Holger Brunst, Andreas Knüpfer, and Wolfgang E. Nagel VIII Contents Performance Analysis of Irregular Collective Communication with the Crystal Router Algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 Michael Schliephake and Erwin Laure The Architecture of Vistle, a Scalable Distributed Visualization System. . . . . 141 Martin Aumüller Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Towards Exascale Scientific Applications