Frontiers of Supercomputing II
  PREFACE
  ACKNOWLEDGMENTS

 collapse section1—  OPENING, BACKGROUND, AND QUESTIONS POSED FOR THIS CONFERENCE
 collapse sectionWelcome
 •Reference
 •Supercomputing As a National Critical Technologies Effort
 collapse sectionGoals for Frontiers of Supercomputing II and Review of Events since 1983
 •1983 Conference Summary
 •Events in Supercomputing since 1983
 •Conference Goals
 •Questions
 collapse sectionCurrent Status of Supercomputing in the United States
 •The Global Imperative
 •Importance of Computers—The Knowledge Economy
 •Computers—A Historic Perspective
 collapse sectionCorrective Action
 •Human Resources
 •R&D Investment
 •Technology Strategy
 •International Cooperation
 •Summary

 collapse section2—  TECHNOLOGY PERSPECTIVE
 •Overview
 •Supercomputing Tools and Technology
 •High-Performance Optical Memory Technology at MCC
 •Digital Superconductive Electronics
 collapse sectionEnabling Technology:  Photonics
 •Introduction
 •A Thousand Interconnections, Each at One Gigabit per Second
 •One Connection at One Terabit per Second
 •References

 collapse section3—  VECTOR PIPELINE ARCHITECTURE
 •Vector Architecture in the 1990s
 •In Defense of the Vector Computer
 •Market Trends in Supercomputing
 •Massively Parallel SIMD Computing on Vector Machines Using PASSWORK
 •Vectors Are Different

 collapse section4—  SCALABLE PARALLEL SYSTEMS
 collapse sectionSymbolic Supercomputing
 •References
 •Parallel Processing:  Moving into the Mainstream
 •It's Time to Face Facts
 •Large-Scale Systems and Their Limitations
 collapse sectionA Scalable, Shared-Memory, Parallel Computer
 •References
 •Looking at All of the Options

 collapse section5—  SYSTEMS SOFTWARE
 •Parallel Software
 collapse sectionSupercomputer Systems-Software Challenges
 •Abstract
 •Introduction
 •Distributed Computing
 •High-Speed Networks
 •Virtual Memory
 •Resource Management
 •Parallel Processing
 •Progress
 •Future Supercomputing Elements
 •Compiler Issues for TFLOPS Computing
 collapse sectionPerformance Studies and Problem-Solving Environments
 •References
 •Systems and Software

 collapse section6—  USER-INTERFACE SOFTWARE
 •Parallel Architecture and the User Interface
 collapse sectionObject-Oriented Programming, Visualization, and User-Interface Issues
 collapse sectionObject-Oriented Parallel Programming
 •Distributed Computing
 •Data Parallel Programming
 •Visualization Requirements
 •References
 collapse sectionSoftware Issues at the User Interface
 •Abstract
 •Introduction
 collapse sectionCompilers and Communication
 •Myrias SPS-2:  Virtual Memory on a Distributed System
 •Myrias SPS-2:  A Concrete Example
 •Myrias SPS-2:  Efficiency of Virtual Memory
 •The Connection Machine CM-2:  Overlapping Communication with Computation
 •Debugging Tools
 •High-Level Languages, Extensions, Libraries, and Graphics
 •Future Supercomputing Environments:  Heterogeneous Systems
 •An Application for a Heterogeneous System
 •Conclusions
 •References
 •What Can We Learn from Our Experience with Parallel Computation up to Now?

 collapse section7—  ALGORITHMS FOR HIGH-PERFORMANCE COMPUTING
 collapse sectionParallel Algorithms and Implementation Strategies on Massively Parallel Supercomputers
 •Introduction
 •Some Developments in Parallel Algorithms
 •Some Developments in Parallel Applications
 •Some Developments in Parallel Applications II
 •Closing Remarks
 •References
 •The Interplay between Algorithms and Architectures:  Two Examples
 collapse sectionLinear Algebra Library for High-Performance Computers
 •Introduction
 •LINPACK
 collapse sectionLINPACK Benchmark
 •Transfer Rate
 •Memory Latency
 •Development of Standards
 •LAPACK
 collapse sectionAlgorithm Design
 •Divide-and-Conquer Approach
 •Accuracy
 •Tools
 •Testing
 •Future Directions for Research
 •References
 •Design of Algorithms
 •Computing for Correctness

 collapse section8—  THE FUTURE COMPUTING ENVIRONMENT
 •Interactive Steering of Supercomputer Calculations
 •A Vision of the Future at Sun Microsystems
 •On the Future of the Centralized Computing Environment
 collapse sectionMolecular Nanotechnology
 •References
 collapse sectionSupercomputing Alternatives
 •Less Is More
 •Supersubstitutes Provide More Overall Capacity
 collapse sectionHow Supers Are Being Niched
 •Workstations
 •Minicomputers and Superminis
 •Mainframes
 •Massively Data-Parallel Computers
 •Minisupercomputers
 •Superworkstations
 •Why Supercomputers Are Becoming Less General Purpose
 •The Supercomputer Industry
 •Is the Supercomputer Industry Hastening Its Own Demise?
 •A Smaller, Healthier Supercomputer Industry
 collapse sectionPolicy Issues
 •Supporting Circuit and Packaging Technology
 •Supers and Security
 •Supers for Competitiveness
 •Conclusions
 •Epilogue, June 1992

 collapse section9—  INDUSTRIAL SUPERCOMPUTING
 collapse sectionOverview of Industrial Supercomputing
 •Abstract
 •Introduction
 •Why Use Supercomputing at All?
 •Impediments to Industrial Use of Supercomputers
 •Technology Transfer and Collaboration
 •Conclusion
 •References
 •Shell Oil Supercomputing
 •Government's High Performance Computing Initiative Interface with Industry
 collapse sectionAn Overview of Supercomputing at General Motors Corporation
 •Abstract
 •Introduction
 •People and the Machine Environment
 •History of Supercomputing at GM
 •Automotive Industry Interest in Supercomputers
 •Applications
 •Long-Term Benefits
 •Needs and Challenges
 •References
 •Barriers to Use of Supercomputers in the Industrial Environment

 collapse section10—  GOVERNMENT SUPERCOMPUTING
 •Planning for a Supercomputing Future
 collapse sectionHigh-Performance Computing at the National Security Agency
 •Introduction
 •Characterization of HPC
 •HPC Architecture
 •Software Environment
 •Mass-Storage Requirements
 •Summary of Issues
 collapse sectionThe High Performance Computing Initiative:  A Way to Meet NASA's Supercomputing Requirements for Aerospace
 •Reference
 •The Role of Computing in National Defense Technology
 •NSF Supercomputing Program

 collapse section11—  INTERNATIONAL ACTIVITY
 collapse sectionA Look at Worldwide High-Performance Computing and Its Economic Implications for the U.S.
 •Abstract
 collapse sectionA Brief Technical Overview of the Present-Day Landscape
 •The Soviet Union
 •Western Europe
 •Japan
 collapse sectionThe Japanese Challenge and "McAdams's Laws"
 collapse sectionIntroduction
 •Law 1—  That Which Is Currently Taking Place Is Not Impossible
 •Japan:  Vertical Integration, Keiretsu, and Government Coordination
 collapse sectionThe U.S.:  Rugged Individualism and Trade-War Losses
 •Law 2—  You Don't Catch up without Catching Up
 collapse sectionTrade:  "Successful" Negotiations and "Potato Chips"
 •Law 3—  When Two Countries Are in a Trade War and One Does Not Realize It, That Country Is Unlikely to Win
 collapse sectionRemedies
 •Law 4—  An Important Aspect of Change Is That Things Are Different Afterward
 •The Future
 •References and Bibliography
 collapse sectionEconomics, Revelation, Reality, and Computers
 •References

 collapse section12—  EXPERIENCE AND LESSONS LEARNED
 •Supercomputing since 1983
 collapse sectionLessons Learned
 •Abstract
 •Introduction
 •Parallel Processing:  1980 to 2000
 •The Attack of the Killer Micros
 •Programmer Productivity on Massively Parallel Systems
 •Front End/Back End Versus Native UNIX
 •Single User versus Multiple Users
 •Interconnect Performance, System Versatility, and Delivered Performance
 •Challenges and Directions for the Future
 •Summary
 •Appendix:  BBN Parallel-Processing Systems
 •References
 collapse sectionThe John von Neumann Computer Center:  An Analysis
 collapse sectionIntroduction
 •The "Pre-Lax Report" Period
 •The Lax Report
 •Establishment of the Centers
 collapse sectionThe John von Neumann Center
 •The Proposal
 •Consortium for Scientific Computing
 •The Universities
 •The State of New Jersey
 •The NSF
 •ETA
 •Zero One
 •JVNC
 collapse sectionWhat Went Wrong?
 •The Analysis
 •Location
 •Corporate Problems
 •NSF, Funding, and Funding Leverage
 •Governance
 •Conclusions
 •Project THOTH:  An NSA Adventure in Supercomputing, 1984–88
 collapse sectionThe Demise of ETA Systems
 •In the Beginning
 •Hardware
 •Software
 •Industry Observations
 •FPS Computing:  A History of Firsts

 collapse section13—  INDUSTRY PERSPECTIVE:  POLICY AND ECONOMICS FOR HIGH-PERFORMANCE COMPUTING
 collapse sectionWhy Supercomputing Matters:  An Analysis of the Economic Impact of the Proposed Federal High Performance Computing Initiative
 •Introduction
 collapse sectionPhase I Methodology
 •Scenario A
 •Scenario B
 •Phase II Methodology
 •Government As Buyer and Leader
 •Concerns about Policies and Economics for High-Performance Computing
 •High-Performance Computing in the 1990s
 •A High-Performance Computing Association to Help the Expanding Supercomputing Industry
 collapse sectionThe New Supercomputer Industry
 •Reference
 •The View from DEC
 •Industry Perspective:  Remarks on Policy and Economics for High-Performance Computing

 collapse section14—  WHAT NOW?
 collapse sectionConference Summary
 •Introduction
 •Skilled, Imaginative Users and a Broad Spectrum of Applications
 •Workstations and Visualization Engines
 •Mass Storage and Accessible Knowledge Bases
 •Heterogeneous High-Performance Computer Engines
 •Fast, Local, Wide-Area, and National Networks
 •Software Environment
 •Concluding Remarks
 •The High Performance Computing Initiative
 •Government Bodies As Investors
 •Realizing the Goals of the HPCC Initiative:  Changes Needed
 •The Importance of the Federal Government's Role in High-Performance Computing
 •Legislative and Congressional Actions on High-Performance Computing and Communications
 •The Federal Role As Early Customer
 •A View from the Quarter-Deck at the National Security Agency
 •Supercomputers and Three-Year-Olds
 •NASA's Use of High-Performance Computers:  Past, Present, and Future
 •A Leadership Role for the Department of Commerce
 •Farewell

  CONTRIBUTORS
collapse section Collapse All | Expand All expand section