High Performance Computing in Science and Engineering: 4th International Conference, HPCSE 2019, Karolinka, Czech Republic, May 20–23, 2019, Revised ... (Lecture Notes in Computer Science, 12456) PDF
Preview High Performance Computing in Science and Engineering: 4th International Conference, HPCSE 2019, Karolinka, Czech Republic, May 20–23, 2019, Revised ... (Lecture Notes in Computer Science, 12456)
Tomáš Kozubek · Peter Arbenz · ˇ Jiˇrí Jaroš · Lubomír Ríha · Jakub Šístek · Petr Tichý (Eds.) 6 5 4 2 1 High Performance Computing S C N in Science and Engineering L 4th International Conference, HPCSE 2019 Karolinka, Czech Republic, May 20–23, 2019 Revised Selected Papers Lecture Notes in Computer Science 12456 Founding Editors Gerhard Goos Karlsruhe Institute of Technology, Karlsruhe, Germany Juris Hartmanis Cornell University, Ithaca, NY, USA Editorial Board Members Elisa Bertino Purdue University, West Lafayette, IN, USA Wen Gao Peking University, Beijing, China Bernhard Steffen TU Dortmund University, Dortmund, Germany Gerhard Woeginger RWTH Aachen, Aachen, Germany Moti Yung Columbia University, New York, NY, USA More information about this subseries at http://www.springer.com/series/7407 áš Tom Kozubek Peter Arbenz (cid:129) (cid:129) ří š í Ří Ji Jaro Lubom r ha (cid:129) (cid:129) Ší ý Jakub stek Petr Tich (Eds.) (cid:129) High Performance Computing in Science and Engineering 4th International Conference, HPCSE 2019 – Karolinka, Czech Republic, May 20 23, 2019 Revised Selected Papers 123 Editors TomášKozubek PeterArbenz VSB- Technical University of Ostrava ETHZurich Ostrava-Poruba, Czech Republic Zurich, Switzerland JiříJaroš Lubomír Říha BrnoUniversity ofTechnology VSB- Technical University of Ostrava Brno, Czech Republic Ostrava-Poruba, Czech Republic JakubŠístek Petr Tichý Institute of Mathematics of the CAS CharlesUniversity Prague,Czech Republic Prague,Czech Republic ISSN 0302-9743 ISSN 1611-3349 (electronic) Lecture Notesin Computer Science ISBN 978-3-030-67076-4 ISBN978-3-030-67077-1 (eBook) https://doi.org/10.1007/978-3-030-67077-1 LNCSSublibrary:SL1–TheoreticalComputerScienceandGeneralIssues ©SpringerNatureSwitzerlandAG2021 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, expressed or implied, with respect to the material contained herein or for any errors or omissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictionalclaimsin publishedmapsandinstitutionalaffiliations. ThisSpringerimprintispublishedbytheregisteredcompanySpringerNatureSwitzerlandAG Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland Preface This volume comprises the proceedings of the 4th International Conference on High Performance Computing in Science and Engineering – HPCSE 2019, which was held intheHotelSoláňintheheartoftheBeskydyMountains,CzechRepublic,onMay20– 23, 2019. The biennial conference was organized by the IT4Innovations National Supercomputing Center at VSB - Technical University of Ostrava with the support of partner institutions: Brno University of Technology, Charles University, Czech TechnicalUniversityinPrague,InstituteofGeonicsandInstituteofMathematicsofthe CAS,andCzechNetworkforMathematicsinIndustryeu-maths-in.cz.Theaimwasto bring together specialists in high performance computing from fields such as applied mathematics, numerical methods, and parallel computing, toexchange experience and toinitiatenewresearchcollaborations.Wearegladthatourinvitationwasacceptedby distinguished experts from world-leading research institutions. This conference has become an international forum for exchanging ideas among researchers involved in scientific and parallel computing, including theory and appli- cations,aswellasappliedandcomputationalmathematics.ThefocusofHPCSE2019 was on models, algorithms, and software tools that facilitate efficient and convenient utilization of modern parallel and distributed computing architectures, as well as on large-scale applications. The members of the Scientific Committee of HPCSE 2019 were Radim Blaheta, Martin Čermák, Zdeněk Dostál, Jaroslav Hron, Jiří Jaroš, Tomáš Kozubek, Jaroslav Kruis, Tomáš Oberhuber, Ivan Šimeček, Jakub Šístek, Petr Tichý, and Vít Vondrák. The invited plenary talks were presented by: (cid:129) Andrea Bartolini (University of Bologna), (cid:129) Steffen Börm (Christian-Albrechts-Universität zu Kiel), (cid:129) Ben T. Cox (University College London), (cid:129) Dominik Göddeke (University of Stuttgart), (cid:129) Frédéric Hecht (Université Pierre et Marie Curie), (cid:129) Jakub Kurzak (University of Tennessee), (cid:129) Günther Of (Technische Universität Graz), (cid:129) Daniela di Serafino (University of Campania Luigi Vanvitelli), (cid:129) Garth Wells (University of Cambridge), (cid:129) Stefano Zampini (KAUST). We gratefully acknowledge the support of the Ministry of Education, Youth and Sports from the National Programme of Sustainability (NPU II) through project “IT4Innovations excellence in science – LQ1602”. The HPCSE 2019 conference was a fruitful event, providing interesting lectures, showcasing new ideas, demonstrating the beauty of applied mathematics, presenting numericallinearalgebra,optimizationmethods,andhighperformancecomputing,and starting or strengthening collaborations and friendships. vi Preface This meeting attracted about 100 participants from 10 countries. All participants were invited to submit an original paper to this book of proceedings. We give thanks forallcontributionsaswellasfortheworkofthereviewers,andhopethatthisvolume will be useful for readers. The proceedings were edited by Tomáš Kozubek, Peter Arbenz, Jiří Jaroš, Lubomír Říha, Jakub Šístek, and Petr Tichý. Finally, we would like to cordially invite readers to participate in the next HPCSE conference, which is planned to be held at the same place on May 17–20, 2021. On behalf of the organizers, Tomáš Kozubek. Tomáš Kozubek Peter Arbenz Jiří Jaroš Lubomír Říha Jakub Šístek Petr Tichý Organization Conference Chair Tomáš Kozubek VSB - Technical University of Ostrava, Czech Republic Scientific Committee Radim Blaheta Institute of Geonics of the CAS, Czech Republic Martin Čermák VSB - Technical University of Ostrava, Czech Republic Zdeněk Dostál VSB - Technical University of Ostrava, Czech Republic Jaroslav Hron Charles University, Czech Republic Jiří Jaroš Brno University of Technology, Czech Republic Tomáš Kozubek VSB - Technical University of Ostrava, Czech Republic Jaroslav Kruis CzechTechnicalUniversityinPrague,CzechRepublic Tomáš Oberhuber CzechTechnicalUniversityinPrague,CzechRepublic Ivan Šimeček CzechTechnicalUniversityinPrague,CzechRepublic Jakub Šístek Institute of Mathematics of the CAS, Czech Republic Petr Tichý Charles University, Czech Republic Vít Vondrák VSB - Technical University of Ostrava, Czech Republic Contents Thermal Characterization of a Tier0 Datacenter Room in Normal and Thermal Emergency Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Mohsen Seyedkazemi Ardebili, Carlo Cavazzoni, Luca Benini, and Andrea Bartolini Towards Local-Failure Local-Recovery in PDE Frameworks: The Case of Linear Solvers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Mirco Altenbernd, Nils-Arne Dreier, Christian Engwer, and Dominik Göddeke Complexity Analysis of a Fast Directional Matrix-Vector Multiplication . . . . 39 Günther Of and Raphael Watschinger Fast Large-Scale Boundary Element Algorithms . . . . . . . . . . . . . . . . . . . . . 60 Steffen Börm Solving Large-Scale Interior Eigenvalue Problems to Investigate the Vibrational Properties of the Boson Peak Regime in Amorphous Materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Giuseppe Accaputo, Peter M. Derlet, and Peter Arbenz Performance Evaluation of Pseudospectral Ultrasound Simulations on a Cluster of Xeon Phi Accelerators. . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Filip Vaverka, Bradley E. Treeby, and Jiri Jaros Estimation of Execution Parameters for k-Wave Simulations . . . . . . . . . . . . 116 Marta Jaros, Tomas Sasak, Bradley E. Treeby, and Jiri Jaros Analysis and Visualization of the Dynamic Behavior of HPC Applications. . . 135 Ondrej Vysocky, Ivo Peterek, Martin Beseda, Matej Spetko, David Ulcak, and Lubomir Riha A Convenient Graph Connectedness for Digital Imagery . . . . . . . . . . . . . . . 150 Josef Šlapal Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 Thermal Characterization of a Tier0 Datacenter Room in Normal and Thermal Emergency Conditions B Mohsen Seyedkazemi Ardebili1( ) , Carlo Cavazzoni2 , Luca Benini1,3 , and Andrea Bartolini1 1 Universit´a degli Studi di Bologna, Viale Risorgimento, 2, 40136 Bologna, Italy {mohsen.seyedkazemi,luca.benini,a.bartolini}@unibo.it 2 CINECA, Via Magnanelli 6/3, Casalecchio di Reno, 40033 Bologna, Italy [email protected] 3 Eidgen¨ossische Technische Hochschule Zu¨rich, Gloriastrasse 35, 8092 Zu¨rich, Switzerland [email protected] https://ee.ethz.ch/ https://www.dei.unibo.it/ https://www.cineca.it/ Abstract. Datacenters are at the heart of the AI, Industry 4.0 and cloud revolution. A datacenter contains a large number of computing nodeshostedinalargetemperature-controlledroom.Duetotheincreas- ing total power and power density of computing nodes, the overall dat- acenter compute capacity is often capped by peak power consumption andtemperaturebottlenecks.Topreservethehomogeneousperformance assumptionbetweenallthenodes,complexcoolingsolutionarerequired, but they might not be sufficient. In this work, we analysed and charac- terisedthethermalpropertiesofaTier0datacenterdeployingadvanced hybridcoolingtechnologies:specifically,westudiedthespatialandtem- poral heterogeneity during production and cooling emergency hazards. This paper gives first quantitative evidence of thermal bottlenecks in real-life production workload, showing the presence of significant spa- tial thermal heterogeneity which could be exploited by thermal-aware job scheduling and datacenter-room run-time workload adaptation and distribution. · · Keywords: HPC Thermal characterization Power consumption 1 Introduction and Related Works Withthegrowthofthecomputingdemandfromabroadsetofsocietal,industrial and science application, datacenter have become a key component of the whole This work has been partially supported by the EU H2020 ICT/2018 project IoTwins (g.a. 857191). (cid:2)c SpringerNatureSwitzerlandAG2021 T.Kozubeketal.(Eds.):HPCSE2019,LNCS12456,pp.1–16,2021. https://doi.org/10.1007/978-3-030-67077-1_1
The list of books you might like
