DOCSLIB.ORG

COMPUTING RESEARCH NEWS January 1997

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
COMPUTING RESEARCH NEWS January 1997 COMPUTING RESEARCH NEWS The News Journal of the Computing Research Association January 1997 Vol. 9/No. 1 How will election results affect science policy? BY Fred W. Weingarten political contributions. If basic CRA Staff research PACs exist, they have not The November elections are over, If science policy is to be dealt with substantively in the attracted much attention or been and the science policy community, very influential either inside or 105th Congress, the House Science Committee will gearing up for another year of intense outside Congress. An exception may be strong the industry support of NIH budget battles, is speculating about have to lead the way. the effects on science policy in the research, but NIH is not under the next Congress. Although last year’s purview of the Science Committee. final numbers turned out to better As a result, the committee usually than expected—at least for the dealt with substantively in the (D-KY), were first-term members; experiences substantial turnover in budgets of the National Institutes of 105th Congress, that committee Bill Baker (R-CA) was in his second the junior ranks and nearly always Health (NIH) and the National will have to lead the way. Whether term. Harold Volkmer (D-MO), the winds up with a higher than most senior incumbent to lose his Science Foundation (NSF)—the it will be interested and equipped to average proportion of first-term do so will depend on how the seat, had served for 20 years. Includ- coming year is likely to be much members. This year is likely to be elections affected its makeup, ing the retirement of committee chair tougher. no different. agenda and working style. Robert Walker (R-PA), only five out The political desire to balance Leadership at the top has already of 50 members of last year’s Science the budget continues, both in the Committee composition been settled. F. James Sensenbrenner Committee will be departing the administration and in Congress, and (R-WI), second in seniority on the The election did not create major House. there are not many places left to cut committee to Walker in the last changes, at least directly. As was true However, in reshuffling commit- in the discretionary portion of the Congress, has been selected to chair in general, the election was kind to tee assignments, more changes are budget. the committee. He is a fiscal conser- incumbents. But many races were likely to be made. For many members, Most initial attention is on the vative who has represented the 9th very close, and a few results could the Science Committee is not a House. The Senate has been District in Wisconsin, a district in the still change with recounts or other highly rewarding post, and many relatively indifferent to science challenges. At press time, one race suburbs of Milwaukee, for 18 years. leave as soon as they can. The Although the University of Wiscon- policy in recent years, and that involving a Science Committee committee has little influence beyond attitude is not likely to change. By member, Steve Stockman (R-TX), sin is not in his district, reports are a narrow set of special issues, and that he has visited the campus contrast, the House has a full was to go to a runoff. NSF is not one of the larger govern- committee—the Science Commit- several times, is interested in basic Only four incumbent members ment agencies. tee—exclusively devoted to science lost their seats. Two of them, Andrea science and is a strong supporter of Consequently, membership on the policy. If science policy is to be Seastrand (R-CA) and Mike Ward academic research. committee attracts little political The ranking Democrat on the full Inside CRN action committee (PAC) money, committee once again is George unlike major House committees such Brown (D-CA), who won another Expanding the Pipeline .................. 2-3 Policy Update ................................. 9 as Commerce and Ways and Means very close race in his Riverside Association News ........................... 4-5 Washington Update ........................ 10 (the committee that deals with district. Cliff-hangers have been the CRA Taulbee Survey ...................... 6-7 Awards and Honors ........................ 11 taxation). Furthermore, the scientific Research News .............................. 8 Professional Opportunities ............. 12-16 community is notoriously stingy with Continued on Page 10 Changes in copyright policy debated BY Louise Arnheim download and send original works to Future Coalition (DFC), a group of Special to CRN colleagues anywhere, anytime. At the 25 organizations, including the PAID same time, the administration and Warning. If you are reading this item Computer and Communications U.S. POSTAGE PERMIT NO. 993 NONPROFIT ORG. electronically, you may be in violation of Congress have been weighing those Industry Association; several of the WASHINGTON, DC federal copyright law. rights with a democratic society’s major library associations, including That—or a similar caution—is need to access information and the the American Library Association; what Internet users might see if, as ability of institutions (libraries, educational interests such as the certain groups suggest, proposed universities) to meet that need. National Education Association; changes to copyright law come to The ongoing copyright debate is consumer groups such as the Con- pass. one with a long list of players: the sumer Federation of America; and These groups—which include information industry, which sees the cyberspace rights groups such as the consumers, educators, librarians, digital environment as one which Electronic Frontier Foundation and scholars of intellectual property and significantly decreases its ability to the Electronic Privacy Information members of the computer and recoup investment in an author’s Center. communications industries—are work and substantially increase The current debate began in 1993 reacting to a proposal raised in instances of copyright piracy; with the Clinton administration’s Congress last year that would have educational institutions and businesses blueprint for an information infra- broadened copyright owners’ right to that distribute limited amounts of structure, Agenda for Action. Among control distribution of their work. copyrighted material to students and the nine principles set forth in the The proposal would have included colleagues, based on what is accept- agenda was this directive: “The “transmission” as part of that right, able under the fair use doctrine; administration will investigate how to making the act of browsing on the libraries, which depend on doctrines strengthen domestic copyright laws Internet a possible copyright such as first sale to lend material; and and international intellectual infringement. service providers that, to avoid property treaties to prevent piracy The “transmission right” is just liability, might be required to monitor and to protect the integrity of one of the changes being talked their subscribers’ Internet browsing intellectual property.” The agenda about in Washington. For the past habits. also established the IITF and, within three years, the Clinton administra- Two coalitions have been particu- that overall structure, several tion, through its national Information larly active in the copyright policy working groups. One of those groups Infrastructure Task Force (IITF), has debate. One is the Creative Incentive was the Working Group on Intellec- been reviewing how well a law Coalition (CIC), which represents tual Property Rights, headed by intended to protect the economic producers, publishers and distributors Bruce Lehman, Commissioner of rights of authors functions in an of copyrighted audio, music, software, CRA NW Ave. 1875 Connecticut Suite 718 DC 20009-5728 Washington, environment that enables readers to text and video. Another is the Digital Continued on Page 9 COMPUTING RESEARCH NEWS January 1997 Expanding the Pipeline Computing Research Association Recruiting on the net not done well Computing Research Association, or the site’s Web home page. Many Board Officers BY Margaret the IEEE Computer Society, the departments use physical mail to David A. Patterson M. Fleck Chair A few years ago, Association for Computing Machin- gather letters of recommendation, University of California, Berkeley recruiting for ery and the Chronicle of Higher confirm receipt of applications and Mary Jane Irwin computer science Education. The best organized and provide information about their Vice Chair jobs was done most comprehensive site is run by an status. It is not uncommon to receive Pennsylvania State University primarily using individual, Alan Kaplan [2]. a form rejection letter months after Nancy G. Leveson hard-copy These sites suffer from several the position has been filled. Commu- Secretary problems. They are incomplete, both nication through e-mail or Web pages University of Washington journal ads and mail. Over the past few years, individually (for faculty posts) and is faster and less expensive. Michael R. Garey collectively (for industrial and Finally, countless pages of hard- Treasurer Internet distribution has become Bell Labs, Lucent Technologies increasingly important, to where postdoctoral positions). Only copy listings are distributed to all many job candidates now view the Kaplan’s site has an organization that people who subscribe to Communica- Board Members Internet as their primary source for is acceptable by the standards of tions of the ACM and IEEE Computer, Alfred V. Aho job ads. Properly used, the Internet modern Web sites. Except at Kaplan’s most of whom are not seeking jobs. Columbia University could save the community substantial site, ads disappear after an unaccept- These listings are printed on expen- Frances E. Allen time and money and allow us to ably short period of time. sive archival journal paper. This IBM T.J. Watson Research Center better implement the intent of equal As a result, job seekers must wastes paper, mailing costs and shelf Gregory R. Andrews employment opportunity laws. monitor multiple listing services, space in university libraries. University of Arizona copy ads before they disappear and Currently, however, we have a Equal opportunities Sandra Johnson Baylor hybrid system.
Load more

Recommended publications
  • Paper in PDF Format
    Managing System Resources on ‘Teras’ Experiences in Managing System Resources on a Large SGI Origin 3800 Cluster The paper to the presentation given at the CUG SUMMIT 2002 held at the University of Manchester. Manchester (UK), May 20-24 2002. http://www.sara.nl Mark van de Sanden [email protected] Huub Stoffers [email protected] Introduction Outline of the Paper This is a paper about our experiences with managing system resources on ‘Teras’. Since November 2000 Teras, a large SGI Origin 3800 cluster, is the national supercomputer for the Dutch academic community. Teras is located at and maintained by SARA High Performance Computing (HPC). The outline of this presentation is as follows: Very briefly, we will tell you what SARA is, and what we presently do, particularly in high performance computing, besides maintaining Teras. Subsequently, to give you an idea, on the one hand which resources are available on Teras, and on the other hand what resources are demanded, we will give overviews of three particular aspects of the Teras: 1) Some data on the hardware – number of CPUs, memory, etc. - and the cluster configuration, i.e. description of the hosts that constitute the cluster. 2) Identification of the key software components that are used on the system; both, system software components, as well as software toolkits available to users to create their (parallel) applications. 3) A characterization of the ‘job mix’ that runs on the Teras cluster. Thus equipped with a clearer idea of both ‘supply’ and ‘demand’ of resources on Teras, we then state the resource allocation policy goals that we pursue in this context, and we review what resource management possibilities we have at hand and what they can accomplish for us, in principle.
    [Show full text]
  • Overview and History Nas Overview
    zjjvojvovo OVERVIEWNAS OVERVIEW AND HISTORY a Look at NAS’ 25 Years of Innovation and a Glimpse at the Future In the mid-1970s, a group of Ames aerospace researchers began to study a highly innovative concept: NASA could transform U.S. aerospace R&D from the costly and cghghmgm time-consuming wind tunnel-based process to simulation-centric design and engineer- ing by executing emerging computational fluid dynamics (CFD) models on supercom- puters at least 1,000 times more powerful than those commercially available at the time. In 1976, Ames Center Director Dr. Hans Mark tasked a group led by Dr. F. Ronald Bailey to explore this concept, leading to formation of the Numerical Aerodynamic Simulator (NAS) Projects Office in 1979. At the same time, a user interface group was formed consisting of CFD leaders from industry, government, and academia to help guide requirements for the NAS concept gmfgfgmfand provide feedback on evolving computer feasibility studies. At the conclusion of these activities in 1982, NASA management changed the NAS approach from a focus on purchasing a specially developed supercomputer to an on-going Numerical Aerody- namic Simulation Program to provide leading-edge computational capabilities based on an innovative network-centric environment. The NAS Program plan for implementing this new approach was signed on February 8, 1983. Grand Opening As the NAS Program got underway, a projects office to a full-fledged division its first supercomputers were installed at Ames. In January 1987, NAS staff and in Ames’ Central Computing Facility, equipment were relocated to the new starting with a Cray X-MP-12 in 1984.
    [Show full text]
  • Fast Synchronization on Shared-Memory Multiprocessors: an Architectural Approach
    Fast Synchronization on Shared-Memory Multiprocessors: An Architectural Approach Zhen Fang1, Lixin Zhang2, John B. Carter1, Liqun Cheng1, Michael Parker3 1 School of Computing University of Utah Salt Lake City, UT 84112, U.S.A. fzfang, retrac, [email protected] Telephone: 801.587.7910 Fax: 801.581.5843 2 IBM Austin Research Lab 3 Cray, Inc. 11400 Burnet Road, MS 904/6C019 1050 Lowater Road Austin, TX 78758, U.S.A. Chippewa Falls, WI 54729, U.S.A. [email protected] [email protected] Abstract Synchronization is a crucial operation in many parallel applications. Conventional synchronization mechanisms are failing to keep up with the increasing demand for efficient synchronization operations as systems grow larger and network latency increases. The contributions of this paper are threefold. First, we revisit some representative synchronization algorithms in light of recent architecture innovations and provide an example of how the simplifying assumptions made by typical analytical models of synchronization mechanisms can lead to significant performance estimate errors. Second, we present an architectural innovation called active memory that enables very fast atomic operations in a shared-memory multiprocessor. Third, we use execution-driven simulation to quantitatively compare the performance of a variety of synchronization mechanisms based on both existing hardware techniques and active memory operations. To the best of our knowledge, synchronization based on active memory outforms all existing spinlock and non-hardwired barrier implementations by a large margin. Keywords: distributed shared-memory, coherence protocol, synchronization, barrier, spinlock, memory controller This effort was supported by the National Security Agency (NSA), the Defense Advanced Research Projects Agency (DARPA) and Silicon Graphics Inc.
    [Show full text]
  • Remanufactured SGI™ Origin™ 2000 High-Performance Server
    Product Guide Remanufactured SGI™ Origin™ 2000 High-Performance Server Origin 2000 Rack Multirack System The rack-mounted Connect multiple racks as a shared- Origin 2000 system memory powerhouse or redeploy is the industry’s those assets into multiple, tightly most powerful and integrated departmental servers. flexible shared-memory Add IRIS FailSafe™ software to server. Easily expand provide failover between systems, the system by adding and Origin 2000 becomes the additional racks. ultimate high-availability server for critical computational and business solutions. Origin 2000 Deskside Revolutionary Server for Your Expanding Business SGI server family, can expand from dual-processor With its outstanding Today’s corporate environment is characterized by deskside systems to powerful 512-processor scalable price/performance, the a rapid growth in communications requirements servers without disruptive and expensive box-swaps deskside Origin 2000 and continual organizational change. A great deal common with other server solutions. System server is ideal for of energy is spent planning future system purchases resources can be deployed as tightly integrated evolving applications in an ongoing effort to keep system capabilities arrays of small Origin 2000 systems or as single requiring expansion capability. Seamless upgradability to larger ahead of application demands. Corporations require large shared-memory systems capable of meeting rack-based configurations supports scalable, flexible computer systems that can adapt the most demanding data processing and your growing requirements. to business change and satisfy growing server computing requirements. requirements. SGI offers you a proven, more adaptable way to purchase your data processing, SGI pioneered the development and delivery of computation, and communications capabilities. shared-memory parallel processing systems.
    [Show full text]
  • IRIX® Device Driver Programmer's Guide
    IRIX® Device Driver Programmer’s Guide Document Number 007-0911-210 CONTRIBUTORS Written by David Cortesi, John Raithel, Bill Tuthill, and Anita Manders UpdatedbyJulieBoneyandStevenLevine Illustrated by Dany Galgani, Cheri Brown, and Chrystie Danzer Production by Karen Jacobson COPYRIGHT © 1998-2003 Silicon Graphics, Inc. All rights reserved; provided portions may be copyright in third parties, as indicated elsewhere herein. No permission is granted to copy, distribute, or create derivative works from the contents of this electronic documentation in any manner, in whole or in part, without the prior written permission of Silicon Graphics, Inc. LIMITED RIGHTS LEGEND The electronic (software) version of this document was developed at private expense; if acquired under an agreement with the USA government or any contractor thereto, it is acquired as "commercial computer software" subject to the provisions of its applicable license agreement, as specified in (a) 48 CFR 12.212 of the FAR; or, if acquired for Department of Defense units, (b) 48 CFR 227-7202 of the DoD FAR Supplement; or sections succeeding thereto. Contractor/manufacturer is Silicon Graphics, Inc., 1600 Amphitheatre Pkwy 2E, Mountain View, CA 94043-1351. TRADEMARKS AND ATTRIBUTIONS Silicon Graphics, SGI, the SGI logo, Challenge, Indigo, IRIS, IRIX, O2, Octane, Onyx, Onyx2, and Origin are registered trademarks and Indigo2,Indigo2 Maximum Impact, IRIS InSight, Power Challenge, Power Channel, Power Indigo2,PowerOnyx,andREACT/proare trademarks of Silicon Graphics, Inc., in the United States and/or other countries worldwide. Indy is a registered trademark, used under license in the United States and owned by Silicon Graphics,Inc.inothercountriesworldwide. IBM is a trademark of International Business Machines Corporation.
    [Show full text]
  • SGI™ Origin 3000 Series Technical Configuration Owner's Guide
    SGI™ Origin 3000 Series Technical Configuration Owner’s Guide Document Number 007-4311-002 CONTRIBUTORS Written by Dick Brownell Illustrated by Dan Young Production by Diane Ciardelli Engineering contributions by SN1 Electrical, Mechanical, Manufacturing, Marketing, and Site Planning teams COPYRIGHT © 2000 Silicon Graphics, Inc. All rights reserved; provided portions may be copyright in third parties, as indicated elsewhere herein. No permission is granted to copy, distribute, or create derivative works from the contents of this electronic documentation. The contents of this document may not be copied or duplicated in any form, in whole or in part, without the prior written permission of Silicon Graphics, Inc. LIMITED RIGHTS LEGEND The electronic (software) version of this document was developed at private expense; if acquired under an agreement with the USA government or any contractor thereto, it is acquired as “commercial computer software” subject to the provisions of its applicable license agreement, as specified in (a) 48 CFR 12.212 of the FAR; or if acquired for Department of Defense units, (b) 48 CFR 227-7202 of the DoD FAR Supplement; or sections succeeding thereto. Contractor/manufacturer is Silicon Graphics, Inc., 1600 Amphitheatre Pkwy. 2E, Mountain View, CA 94043-1351. TRADEMARKS AND ATTRIBUTIONS Silicon Graphics is a registered trademark and SGI, the SGI logo, Origin, Onyx3, and IRIS InSight are trademarks of Silicon Graphics, Inc. PostScript is a trademark of Adobe Systems, Inc. MIPS is a registered trademark of MIPS Technologies, Inc. Windows and Windows NT are registered trademarks of Microsoft Corporation. DVCPRO is a trademark of Panasonic, Inc. Linux is a registered trademark of Linus Torvalds.
    [Show full text]
  • Sgrortgtn'" 3000 Sertes Modular, Hlgh-Performance Servers Sgiorigin 3000 Series Product Stanford and SGI .Q.Y..~Iyi~W
    ENCLOSURE 2 • l~ Product Guide 5 ...... SGrOrtgtn'" 3000 Sertes Modular, Hlgh-Performance Servers SGIOrigin 3000 Series Product Stanford and SGI .Q.y..~IYi~w. Announce New ~l,;?!.Q[igln.~2..Q.Q Partnership in §.l,;?LQ.r:isln Biomedical ~ Supercomputing SGI Origin 3400 [more] SGI Origin 3800 SARA Supercomputer System Facility Installation [more] Software Modular NUMAflex Supercomputing [view] Bricks ill Partijioning SAIC Upgrades FBI's The SGI Origin 3000 series takes system modularity to Datasheet & National Instant new heights, as NUMAflex™ allows you to scale CPU, Criminal Background White Papers storage, and I/O components independently within each Check System with Server system. Complete multidimensional flexibility allows SGI Origin Technology Solutions organizations to deploy. service, and expand system [more] components in every possible dimension to meet any Related News ShareAPhoto Unveils business demand. The only limitation is your imagination. Revolutionary Online Contact Us Digital Video Solution Using SGI Origin 3000 Configure Series Server [more] This System SGIOrigin Origin 3000:S ~ll.if.S 3200 SGIOrigin dem 3400 p';:2l;? i:t ~{!q(lj(~~ M'l\;iUlilM\;j SGIOrigin f·;ilS~".~ pi<~.hi 3800 Related Sites SGIOnyx 3000 Series Storage Networking Streaming Media privacy policy IQyestions/CQmments • Copyright © 1993-2001 Silicon Graphics, Inc. All rights reserved. ITrademark InfonDation °SGIOrigin Product Overview 3000 Series SGITM Origin™ 3000 Series of Servers Product Overview The SGITM Origin™ 3000 Series of servers sets the standard for high • S.G.LQ.r.igin.~2..QQ performance in today's market place by delivering flexibility, resiliency, • S.g.LQrisin investment protection, and performance in a new and innovative 3200c package.
    [Show full text]
  • Accelerating Scientific Discovery Through Computation and Visualization
    Volume 105, Number 6, November–December 2000 Journal of Research of the National Institute of Standards and Technology [J. Res. Natl. Inst. Stand. Technol. 105, 875-894 (2000)] Accelerating Scientific Discovery Through Computation and Visualization Available online http://www.nist.gov/jres Please note that although all figures in the printed version of this paper are in black and white, Figs. 1–3, 7–9, 12–14, and 17–19 in the online ver- sion are in color. Volume 105 Number 6 November–December 2000 James S. Sims, John G. Hagedorn, Peter The rate of scientific discovery can be condensate modeling, (2) fluid flow in M. Ketcham, Steven G. Satterfield, accelerated through computation and porous materials and in other complex Terence J. Griffin, William L. George, visualization. This acceleration results from geometries, (3) flows in suspensions, Howland A. Fowler, Barbara A. am Ende, the synergy of expertise, computing (4) x-ray absorption, (5) dielectric Howard K. Hung, Robert B. Bohn, John E. Koontz, Nicos S. Martys, Charles E. tools, and hardware for enabling high- breakdown modeling, and (6) dendritic Bouldin, James A. Warren, David L. performance computation, information growth in alloys. Feder, Charles W. Clark, B. James Filla, science, and visualization that is provided and Judith E. Devaney by a team of computation and visualiza- tion scientists collaborating in a peer-to- National Institute of Standards and peer effort with the research scientists. Technology, Gaithersburg, MD 20899-0001 In the context of this discussion, high Key words: discovery science; distri- [email protected] performance refers to capabilities beyond buted processing; immersive environments; [email protected] the current state of the art in desktop IMPI; interoperable MPI; message pass- [email protected] computing.
    [Show full text]
  • IRIX® 6.5.7 Update Guide 1600 Amphitheatre Pkwy
    IRIX® 6.5.7 Update Guide 1600 Amphitheatre Pkwy. Mountain View, CA 94043-1351 Telephone (650) 960-1980 FAX (650) 961-0595 February 2000 Dear Valued Customer, SGI is pleased to present the new IRIX 6.5.7 maintenance and feature release. Starting with IRIX 6.5, SGI created a new software upgrade release strategy, which delivers both the maintenance (6.5.7m) and feature (6.5.7f) streams. This upgrade is part of a family of releases that enhance IRIX 6.5. There are several benefits to this strategy: it provides periodic fixes to IRIX, it assists in managing upgrades, and it supports all platforms. Additional information on this strategy and how it affects you is included in the updated Installation Instructions manual contained in this package. If you need assistance, please visit the Supportfolio Online Web site at: http://support.sgi.com or contact your local support provider. 2 We thank you for your continued commitment to SGI. Sincerely, Jorge Helmer Vice President & General Manager Customer Support Division SGI 3 Welcome to your SGI IRIX 6.5.7 update. This booklet contains: • A list of key features in IRIX 6.5.7 • A list of CDs contained in the IRIX 6.5.7 update kit • A guide to SGI Web sites 4 IRIX 6.5.7 Key New Features The following features are in the core IRIX 6.5.7 overlays. Hardware Supported • Support for Fiber Channel Tape on a Storage Area Network (fabric) using the qLogic 2200 fibre channel controller Introduced in 6.5.6: • R12000S CPU on Origin 200, SGI 2100, SGI 2200, SGI 2400, and SGI 2800 systems Introduced in 6.5.5: • QLA2200
    [Show full text]
  • Numaflex Modular Computing: New Model for Scalable Computing Infrastructure
    NUMAflex Modular Computing: New Model for Scalable Computing Infrastructure Analysts: Earl C. Joseph II, Ph.D., Christopher G. Willard, Ph.D., and Debra Goldfarb Background Managers and users of high-performance computers (HPCs) require systems with the performance to address next-generation problems. They require the scalability to both manage large workloads and run large problems, the flexibility to configure systems to meet individual site requirements, the resiliency to be trusted with “bet-your-business” engineering and operations applications, and the investment protec- tion and longevity to incorporate Moore’s Law technology improve- ments without sacrificing effective components, software, and personnel skills. Computer vendors have responded to these requirements in part by developing scalable computer architectures. Although great strides in scalability have been made over the last decade, we believe that the full potential of computer scaling has yet to be fulfilled. In designing suc- cessful, scalable high-performance computers, manufacturers have to balance such factors as cost, performance, feature richness, form fac- tors, and scalability ranges. The most successful of today’s scalable sys- tem designs started with powerful microprocessors, industry standard DRAM or SDRAM memory components, and standard I/O products. The technology for connecting these components is the number-one challenge for computer designers, and it determines the overall effec- tiveness of the architecture. A major challenge has been to design systems that can grow in multiple dimensions, such as processing performance, memory and I/O, while maintaining consistent price/performance across all configurations and allowing each dimension to expand independently from the others (e.g., increasing I/O without increasing processors.) The capability to tightly integrate these HPC systems with graphics subsystems and large-scale data storage facilities has also become a requirement for many customers.
    [Show full text]
  • The SGI Origin: a Ccnuma Highly Scalable Server
    The SGI Origin: A ccNUMA Highly Scalable Server James Laudon and Daniel Lenoski Silicon Graphics, Inc. 2011 North Shoreline Boulevard Mountain View, California 94043 [email protected] [email protected] Abstract Node 0 The SGI Origin 2000 is a cache-coherent non-uniform memory access (ccNUMA) multiprocessor designed and manufactured by Proc A Proc B Silicon Graphics, Inc. The Origin system was designed from the ground up as a multiprocessor capable of scaling to both small Node Node 511 and large processor counts without any bandwidth, latency, or cost Mem 1 cliffs.The Origin system consists of up to 512 nodes interconnected & Hub IO by a scalable Craylink network. Each node consists of one or two Dir Chip Xbar R10000 processors, up to 4 GB of coherent memory, and a connec- tion to a portion of the XIO IO subsystem. This paper discusses the motivation for building the Origin 2000 and then describes its ar- IO Ctrls chitecture and implementation. In addition, performance results are presented for the NAS Parallel Benchmarks V2.2 and the Scalable Interconnect Network SPLASH2 applications. Finally, the Origin system is compared to other contemporary commercial ccNUMA systems. Figure 1 Origin block diagram 1 Background shared memory (DSM), with cache coherence maintained via a di- Silicon Graphics has offered multiple generations of symmetric rectory-based protocol. A DSM system has the potential for meet- multiprocessor (SMP) systems based on the MIPS microproces- ing all three goals: scalability, ease of programming, and cost. The sors. From the 8 processor R3000-based Power Series to the 36 directory-based coherence removes the broadcast bottleneck that processor R4000-based Challenge and R10000-based Power Chal- prevents scalability of the snoopy bus-based coherence.
    [Show full text]
  • SGI® Fibre Channel PCI Option Board and XIO™ Option Board User's
    SGI® Fibre Channel PCI Option Board and XIO™ Option Board User’s Guide 007-3633-006 CONTRIBUTORS Written by Carolyn Curtis Updated by Matt Hoy and Nancy Heller Illustrated by Dan Young Edited by Susan Wilkening Production by Karen Jacobson Contributions by William Mckevitt, Michael Raskie, Carl Rigg, Yin So, and Audy Watson. Cover Design By Sarah Bolles, Sarah Bolles Design, and Dany Galgani, SGI Technical Publications. COPYRIGHT © 2003, Silicon Graphics, Inc. All rights reserved; provided portions may be copyright in third parties, as indicated elsewhere herein. No permission is granted to copy, distribute, or create derivative works from the contents of this electronic documentation in any manner, in whole or in part, without the prior written permission of Silicon Graphics, Inc. LIMITED RIGHTS LEGEND The electronic (software) version of this document was developed at private expense; if acquired under an agreement with the USA government or any contractor thereto, it is acquired as "commercial computer software" subject to the provisions of its applicable license agreement, as specified in (a) 48 CFR 12.212 of the FAR; or, if acquired for Department of Defense units, (b) 48 CFR 227-7202 of the DoD FAR Supplement; or sections succeeding thereto. Contractor/manufacturer is Silicon Graphics, Inc., 1600 Amphitheatre Pkwy 2E, Mountain View, CA 94043-1351. TRADEMARKS AND ATTRIBUTIONS Silicon Graphics, SGI, the SGI logo, InfiniteReality, IRIX, O2, Octane, Onyx, Onyx2, and Origin are registered trademarks, and GIGAchannel, NUMAlink, and XIO are trademarks of Silicon Graphics, Inc., in the United States and/or other countries worldwide. R10000 is a registered trademark of MIPS Technologies, Inc., used under license by Silicon Graphics, Inc., in the United States and/or other countries worldwide.
    [Show full text]