Contents

• Contents
• Introduction
  • Overview
  • Objectives
    • Identify the Applications
    • Determine the Challenge
    • Reveal the Enabling Technologies
    • Derive the Research Issues
    • Set the Research Agenda
  • Approach
  • Background
  • Issues
  • Report Organization
• PetaFLOPS from Two Perspectives
• Summary of Working Group Reports
  • Applications
  • Device Technology
    • Semiconductor Technologies
    • Optical Technologies
    • Superconducting Technologies
  • Architecture
    • State of the Art
    • Barriers
    • Alternatives
    • Results
    • Recommendations
      • Architecture Models
      • SIA Projections
      • Applications Scaling
  • Software Technology Working Group
• Applications Working Group
  • Introduction
  • Applications Motivation of a PetaFLOPSComputer Program
  • Issues/Characteristics for Architecture(Hardware) and Software of PetaFLOPSMachines
  • Exemplar Applications
    • Porous Media
    • Computational Astrophysics
      • Need for a PetaFLOPS Computer
      • Architectural Issues
      • I/O and Memory Requirements
    • Lattice QCD
    • Computational Quantum Chemistry-HIV Protease Structure
    • PetaFLOPS or PetaOPS Requirements fromGenome Projects
    • Drug Design
    • Three-Dimensional Heart
    • Global Surface Database
      • Satellite down link characteristics.
      • Data processing requirements.
      • Database Queries.
    • Video Image Fusion with Virtual Environments:Generating Interactive CyberScenes
  • Algorithmic Issues for the PetaFLOPSComputer
• Device Technology Working Group
  • Introduction
  • Silicon Device Technology
    • Existing Silicon Devices
    • National Roadmap for Silicon Technology
    • Projected Year 2015 Devices
    • Technology Suitability for PetaFLOPS Machines
  • Optical Devices
    • Interconnects
      • Guided Optical Interconnects
      • Free-Space Interconnects
    • Memory
      • Planar Optical Storage
      • 3-D Optical Storage
    • Processors
  • Projections for Technology Development
  • Optical Technology Mapped onto Proposed PetaFLOPS Scenarios
    • Interconnects
    • Memory
  • R&DRecommendations
  • Superconductive Electronics for PetaFLOPS Computing
    • Logic
    • Memory
    • Interconnects
      • On-chip
      • Chip-to-Chip
      • MCM-to-Room Temperature Connection
    • Physical Limits
    • Barriers/Obstacles
      • R&Dto Overcome Barriers and Obstacles
  • Device Technology Summary
• Architecture Working Group
  • Metrics and Limitations
  • PetaFLOPS Architectures Design Points
  • Role of Device Technology
  • Obstacles and Uncertainties
  • Final Comments
  • Acknowledgment
• Software Technology Working Group
  • Introduction
  • The Challenge of Software Technology
  • Technology Trends and StrategicOpportunities
  • BLISS versus the Metasystem: TwoVisions of PetaFLOPS Computing
  • Discussion of Software Technology Areas
    • Programming Languages and Models
    • Programming Technology
    • Input/Output
    • Resource Management/Scheduling
    • The Training and Experience Base of ScientificProgram Developers
  • Recommendations
    • Machine-Effectiveness Recommendations
    • Human-Effectiveness Recommendations
    • Infrastructure Recommendations
  • Epilogue
• Major Findings
  • Summary Position
  • Feasibility
  • Broad Potential Use
  • Cost a Pacing Item
  • Reliability Manageable
  • MIMD Model
  • Latency Management and Parallelism
  • Riding the Semiconductor Technology Wave
  • Memory
    • Size
    • Bandwidth
    • Global Name Space
    • petaByte Computer
  • Software Paradigm Shift
  • Merging of Technologies
  • A Role for Superconductivity
  • Optical Logic Unlikely
• Issues and Implications
  • Why Consider PetaFLOPS Now?
  • Role of a PetaFLOPS Computer
  • Side-effect Products
  • Impact of Exotic Technologies
  • Performance Versus Efficiency
  • Programming Paradigms
  • U.S. Capabilities in Memory Fabrication
  • Special Widgets, Where to Invest
  • A Range of Architectures
  • Far-side Architectures
  • Latency Hiding Techniques May HelpSmaller Machines
  • Long versus Short Latency Machines
  • SIA Predictions
  • I/O Scaling
• Conclusions and Recommendations
  • Concluding Observations
    • When?
    • By whom?
    • Usage?
    • Long Lead to Research-derived Products
    • Leverage Mainstream Hardware and SoftwareTechnology
    • Value Beyond Applications
    • Intellectual Synergism
  • Recommendations for Near-TermInitiatives
    • Superconducting Technology
    • Scaled-up Applications Resource Requirements
    • A Future Application Scenario
    • Detailed Architectures
    • Extend SIA Projections
    • Programming Methodologies
    • Alternative Concepts and Approaches
    • Alternative Architectural Approaches
    • Cheap TeraFLOPS Machines
    • Review Progress
• Attendee List
• References