Algorithmic Differentiation of Pragma-Defined Parallel Regions Diffe
Numerical programs often use parallel programming techniques such as OpenMP to compute the program's output values as efficient as possible. In addition, derivative values of these output values with respect to certain input values play a crucial role. To
- PDF / 4,633,715 Bytes
- 411 Pages / 419.52 x 595.2 pts Page_size
- 40 Downloads / 244 Views
Michael Förster
Algorithmic Differentiation of Pragma-Defined Parallel Regions Differentiating Computer Programs Containing OpenMP
Michael Förster RWTH Aachen University Aachen, Germany
D 82, Dissertation RWTH Aachen University, Aachen, Germany, 2014
ISBN 978-3-658-07596-5 DOI 10.1007/978-3-658-07597-2
ISBN 978-3-658-07597-2 (eBook)
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. Library of Congress Control Number: 2014951338 Springer Vieweg © Springer Fachmedien Wiesbaden 2014 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. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law. 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. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Printed on acid-free paper Springer Vieweg is a brand of Springer DE. Springer DE is part of Springer Science+Business Media. www.springer-vieweg.de
Abstract The goal of this dissertation is to develop a source code transformation that exploits the knowledge that a given input code is parallelizable in a way that it generates derivative code efficiently executable on a supercomputer environment. There is barely a domain where optimization does not play a role. Not only in science and engineering, also in economics and industry it is important to find optimal solutions for a given problem. The size of these optimization problems often requires large-scale numerical techniques that are capable of running on a
Data Loading...
