Automatic Performance Modeling of HPC Applications
Many existing applications suffer from inherent scalability limitations that will prevent them from running at exascale. Current tuning practices, which rely on diagnostic experiments, have drawbacks because (i) they detect scalability problems relatively
- PDF / 15,377,641 Bytes
- 557 Pages / 439.42 x 683.15 pts Page_size
- 88 Downloads / 217 Views
Hans-Joachim Bungartz Philipp Neumann Wolfgang E. Nagel Editors
Software for Exascale Computing – SPPEXA 2013-2015 Editorial Board T. J.Barth M.Griebel D.E.Keyes R.M.Nieminen D.Roose T.Schlick
Lecture Notes in Computational Science and Engineering Editors: Timothy J. Barth Michael Griebel David E. Keyes Risto M. Nieminen Dirk Roose Tamar Schlick
113
More information about this series at http://www.springer.com/series/3527
Hans-Joachim Bungartz • Philipp Neumann • Wolfgang E. Nagel Editors
Software for Exascale Computing – SPPEXA 2013-2015
123
Editors Hans-Joachim Bungartz Philipp Neumann Institut fRur Informatik Technische UniversitRat MRunchen Garching Germany
Wolfgang E. Nagel Technische UniversitRat Dresden Dresden Germany
ISSN 1439-7358 ISSN 2197-7100 (electronic) Lecture Notes in Computational Science and Engineering ISBN 978-3-319-40526-1 ISBN 978-3-319-40528-5 (eBook) DOI 10.1007/978-3-319-40528-5 Library of Congress Control Number: 2016951949 Mathematics Subject Classification (2010): 65-XX, 68-XX, 70-XX, 76-XX, 85-XX, 86-XX, 92-XX © Springer International Publishing Switzerland 2016 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the 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 storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. Cover illustration: Cover figure by courtesy of iStock.com/nadla Printed on acid-free paper This Springer imprint is published by Springer Nature The registered company is Springer International Publishing AG Switzerland
Preface
One of the grand challenges with respect to current developments in highperformance computing (HPC) lies in exploiting the upcoming exascale systems, i.e. systems with 1018 floating point operations per second and beyond. Moore’s law has proven to be astonishingly robust so far. However, these days, the respective progress can no longer be obtained via faster chips or higher clock rates, but only via a massive use of parallelism and by increasingly complex ways how this parallelism is arranged in the large systems. Extreme-scale supercomputers will be made of heterogeneous hardware and will comprise
Data Loading...
