Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
:52 - : - S:: =: - , - - .~ OF ~ _ WE HA F Committee on the Future of Supercomputing . . Computer Science and Telecommunications Board Division on Engineering and Physical Sciences NATIONAL RESEARCH COUNCIL OF ME INTO - L ARIES THE NATIONAL ACADEMIES PRESS Washington, D.C. www.nap.edu
THE NATIONAL ACADEMIES PRESS 500 Fifth Street, N.W. Washington, DC 20001 NOTICE: The project that is the subject of this report was approved by the Governing Board of the National Research Council, whose members are drawn from the councils of the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine. The members of the committee responsible for the report were chosen for their special competences and with regard for appropriate balance. Support for this project was provided by the DepaWnent of Energy. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the sponsor. International Standard Book Number 0-309-08995-6 (Book) International Standard Book Number 0-309-52675-2 (PDF) Cover designed by Jennifer M. Bishop. Copies of this report are available from the National Academies Press, 500 Fifth Street, N.W., Lockbox 285, Washington, DC 20055, (800) 624-6242 or (202) 334-3313 in the Washington metropolitan area. Internet, http://www.nap.edu Copyright 2003 by the National Academy of Sciences. All rights reserved. Printed in the United States of America
Admers to the Nation on 5oenre, t~gineer/ng, and Medirine The National Academy of Sciences is a private, nonprofit, self-perpetuating society of distinguished scholars engaged in scientific and engineering research, dedicated to the furtherance of science and technology and to their use for the general welfare. Upon the authority of the charter granted to it by the Congress in 1863, the Academy has a mandate that requires it to advise the federal government on scientific and technical matters. Dr. Bruce M. Alberts is president of the National Academy of Sciences. The National Academy of Engineering was established in 1964, under the charter ofthe National Academy of Sciences, as a parallel organization of outstanding engineers. It is autonomous in its administration and in the selection of its members, sharing with the National Academy of Sciences the responsibility for advising the federal government. The National Academy of Engineering also sponsors engineering programs aimed at meeting national needs, encourages education and research, and recognizes the superior achievements of engineers. Dr. Wm. A. Wulf is president of the National Academy of Engineering. The Institute of Medicine was established in 1970 by the National Academy of Sciences to secure the services of eminent members of appropriate professions in the examination of policy matters pertaining to the health of the public. The Institute acts under the responsibility given to the National Academy of Sciences by its congressional charter to be an adviser to the federal government and, upon its own initiative, to identify issues of medical care, research, and education. Dr. Harvey V. Fineberg is president of the Institute of Medicine. The National Research Council was organized by the National Academy of Sciences in 1916 to associate the broad community of science and technology with the Academy's purposes of furthering knowledge and advising the federal government. Functioning in accordance with general policies determined by the Academy, the Council has become the principal operating agency of both the National Academy of Sciences and the National Academy of Engineering in providing services to the government, the public, and the scientific and engineering communities. The Council is administered jointly by both Academies and the Institute of Medicine. Dr. Bruce M. Alberts and Dr. Wm. A. Wulf are chair and vice chair, respectively, of the National Research Council. www. national-academies . org
COMMITTEE ON THE FUTURE OF SUPERCOMPUTING SUSAN L. GRAHAM, University of California, Berkeley, Co-chair MARC SNIR, University of Illinois at Urbana-Champaign, Co-chair WILLIAM J. DALLY, Stanford University JAMES DEMMEL, University of California, Berkeley JACK J. DONGARRA, University of Tennessee, Knoxville KENNETH S. FLAMM, University of Texas at Austin MARY JANE IRWIN, Pennsylvania State University CHARLES KOELBEL, Rice University BUTLER W. LAMPSON, Microsoft Corporation ROBERT LUCAS, University of Southern California PAUL C. MESSINA, Argonne National Laboratory (part-time) JEFFREY PERLOFF, University of California, Berkeley WILLIAM H. PRESS, Los Alamos National Laboratory ALBERT J. SEMTNER, Naval Postgraduate School SCOTT STERN, Kellogg School of Management, Northwestern University SHANKAR SUBRAMANIAM, University of California, San Diego LAWRENCE C. TARBELL, JR., Eagle Alliance STEVEN J. WALLACH, Chiaro Networks Staff CYNTHIA A. PATTERSON, Study Director and Program Officer PHIL HILLIARD, Research Associate MARGARET MARSH HUYNH, Senior Project Assistant IV
COMPUTER SCIENCE AND TELECOMMUNICATIONS BOARD DAVID D. CLARK, Massachusetts Institute of Technology, Chair ERIC BENHAMOU, 3Com Corporation ELAINE COHEN, University of Utah THOMAS E. DARCIE, University of Victoria MARK E. DEAN, IBM JOSEPH FARRELL, University of California, Berkeley JOAN FEIGENBAUM, Yale University HECTOR GARCIA-MOLINA, Stanford University RANDY H. KATZ, University of California, Berkeley WENDY A. KELLOGG, IBM T.J. Watson Research Center SARA KIESLER, Carnegie Mellon University BUTLER W. LAMPSON, Microsoft Corporation, CSTB member emeritus DAVID LIDDLE, U.S. Venture Partners TERESA H. MENG, Stanford University TOM M. MITCHELL, Carnegie Mellon University DANIEL PIKE, GCI Cable and Entertainment ERIC SCHMIDT, Google Inc. FRED B. SCHNEIDER, Cornell University BURTON SMITH, Cray Inc. WILLIAM STEAD, Vanderbilt University ANDREW J. VITERBI, Viterbi Group, LLC JEANNETTE M. WING, Carnegie Mellon University ALAN S. INOUYE, Interim Director JON EISENBERG, Interim Deputy Director KRISTEN BATCH, Research Associate JENNIFER M. BISHOP, Senior Project Assistant JANET BRISCOE, Administrative Officer DAVID DRAKE, Senior Project Assistant RENEE HAWKINS, Financial Associate PHIL HILLIARD, Research Associate MARGARET MARSH HUYNH, Senior Project Assistant HERBERT S. LIN, Senior Scientist LYNETTE I. MILLETT, Program Officer DAVID PADGHAM, Research Associate CYNTHIA A. PATTERSON, Program Officer JANICE SABUDA, Senior Project Assistant BRANDYE WILLIAMS, Staff Assistant STEVEN WOO, Dissemination Officer For more information on CSTB, see its Web site at <http://www.cstb.org>, write to CSTB, National Research Council, 500 Fifth Street, N.W., Washington, DC 20001, call at (202) 334-2605, or e-mail the CSTB at cstb~nas.edu. v
Prepublication Copy Subject to Further Editorial Correction Vi
Preface High-performance computing is important in solving many kinds of complex problems in domains from weather science and biology to national security. U.S. government spending on supercomputing has been relatively flat over the last 10 years and has declined compared to industrial or commercial purchases of high-performance systems. Some observers associate the trends with the end of the Cold War, during which both national security and scientific research needs were believed to justify spending on a range of high-performance computing programs. Others point to the influence of the changing marketplace for computing systems overall, which has seen an increase in demand for less expensive systems of less than maximal performance. Several factors have led to the recent interest in reexamining the rationale for federal investment in research and development in support of high-performance computing, including (1) continuing changes in various component technologies and their markets, (2) the evolution of the computing market (and particularly the high-end supercomputing segment), (3) experience with several systems using the clustered processor architecture, and (4) the evolution of the problems, many of them mission-driven, for which supercomputers are used. The Department of Energy's (DOE's) Office of Science expressed an interest in sponsoring a study by the Computer Science and Telecommunications Board (CSTB) of the National Research Council that would assess the state of U.S. supercomputing capabilities and relevant research and development. Spurred by the development of the Japanese vector-based Earth Simulator supercomputer, the Senate's Energy and Water Development Appropriations Committee directed the Advanced Simulation and Computing (ASC) Program of the National Nuclear Security Administration (NNSA) at DOE to commission a study (in collaboration with DOE's Office of Science) by the National Research Council. Congress also commissioned a study by the JASONs2 to identify the distinct requirements of the stockpile stewardship program and its relation to the ASC acquisition strategy. CSTB convened the Committee on the Future of Supercomputing to assess prospects for supercomputing technology research and development in support of U.S. needs, to examine key elements of context the history of supercomputing, the erosion of research investment, the changing nature of problems demanding supercomputing, and the needs of government agencies for supercomputing capabilities and to assess options for progress. The committee has been tasked with preparing two iDebra Goldfarb, IDC's HPC Industry Analyst, presentation to the committee on May 23, 2003. 2The JASONs, formed in 1959, are a select group of scientific advisors who consult with the federal government chiefly on classified research issues. For more on the JASONs, see Ron Southwick, 2002, Elite Panel of Academics Wins Fight to Continue Advising Military, The Chronicle of Higher Education, June 7. Available at <www.globalsecurity.org/org/news/2002/020607 jason.htm>. . . V11
reports a brief interim report and a final in-depth report. This interim report is intended only to establish context the history and current state of supercomputing, application requirements, technology evolution, the socioeconomic context, and so on and to identify some of the issues that may be explored in more depth in the second phase of the study. In order to provide feedback as soon as possible, this interim report has been developed on a very tight time line. It is based on committee deliberations and briefings received from numerous experts at two committee meetings.3 It does not contain formal conclusions or recommendations. The committee expects that its understanding of some background issues will change as it collects more data and deepens its analysis in the final report, anticipated in late 2004. The committee thanks the many individuals who contributed to its work. The people who briefed the committee at one of the plenary meetings are listed in Appendix C. Their willingness to answer our questions was most helpful. Stanford University, with the able local support of Pamela Elliott and Charles M. Orgish, hosted the second plenary meeting of the committee. The sponsors of the report at the Department of Energy Daniel Hitchcock, Fred Johnson, Jose L. Munoz, Dimitri Kusnezov, and Hans Ruppel have been most supportive and responsive in helping the committee to do its work. The reviewers of the draft report provided insightful and constructive comments that contributed significantly to the clarity of the report. The work of the committee was made considerably easier because of the participation of excellent NRC staff members. Marjory Blumenthal, the outgoing director of CSTB, shared her wisdom and knowledge with the committee until she left the NRC in June. She will be greatly missed by all of us who have worked with her on this study and in earlier CSTB activities. Jon Eisenberg, Herb Lin, and Richard Rowberg have given valuable counsel to the chairs. Margaret Marsh Huynh is providing first-rate administrative and logistical support to the committee. Liz Fikre, the report editor, has taught us much about clear exposition. Phil Hilliard has willingly and competently given research support to the committee. Finally, Cynthia Patterson, the study director, has been an outstanding partner and mentor to the chairs. Her contributions have strengthened both the study and the interim report. Susan L. Graham and Marc Snir, Co-chairs Committee on the Future of Supercomputing 3The speakers are listed in Appendix C. . . . v'''
Acknowledgment of Reviewers This report has been reviewed in draft form by individuals chosen for their diverse perspectives and technical expertise, in accordance with procedures approved by the National Research Council's (NRC's) Report Review Committee. The purpose of this independent review is to provide candid and critical comments that will assist the institution in making the published report as sound as possible and to ensure that the report meets institutional standards for objectivity, evidence, and responsiveness to the study charge. The review comments and draft manuscript remain confidential to protect the integrity of the deliberative process. We wish to thank the following individuals for their review of this report: Roy Radner, New York University, Ahmed H. Sameh, Purdue University, Charles L. Seitz, Myricom, Inc., Alian Snavely, San Diego Supercomputing Center, Francis Sullivan, IDA Center for Computing Sciences, and Paul R. Woodward, University of Minnesota. Although the reviewers listed above have provided many constructive comments and suggestions, they were not asked to endorse the conclusions or recommendations, nor did they see the final draft of the report before its release. The review of this report was overseen by Samuel H. Fuller, Analog Devices, Inc. Appointed by the National Research Council, he was responsible for making certain that an independent examination of this report was carried out in accordance with institutional procedures and that all review comments were carefully considered. Responsibility for the final content of this report rests entirely with the authoring committee and the institution. IX
Contents EXECUTIVE SUMMARY INTRODUCTION Study Context, 6 About This Interim Report, 8 SUPERCOMPUTING PAST AND PRESENT Previous Reports and Recent Federal Initiatives, 9 Supercomputing Technology, 13 Vendors, 13 Architecture, 13 Products, 15 The NEC Earth Simulator, 16 Software, 17 Algorithms, 17 3 CONTINUITY AND PREDICTABILITY Important Work Is Getting Done, 18 No Near-Term Alternatives, 19 Older Architectures Coexist with New Ones, 19 The Importance and Continuing Value of Software Research and Algorithm Development, 20 Legacy Codes Cannot Be Abandoned Until They Are Replaced, 21 Uncertainty and Inconsistent Policies Can Be Expensive, 21 4 FUTURE SUPERCOMPUTING AND RESEARCH Innovation in High-End Computing, 22 Architecture Research, 23 Software Research, 24 Research on Applications and Algorithms, 25 5 9 18 22
. . X11 THE ROLE OF GOVERNMENT IN SUPERCOMPUTING Government As a Leading Customer, 28 National Security Implications, 29 Market Forces, 29 6 CONCLUSION APPENDIXES A COMMITTEE MEMBER AND STAFF BIOGRAPHIES B ACRONYMS C BRIEFERS TO THE COMMITTEE WHAT IS CSTB? CONTENTS 28 31 35 43 45 46