Industrial Strength Parallel Computing

Industrial Strength Parallel Computing

1st Edition - October 11, 1999

Write a review

  • Editor: Alice Koniges
  • eBook ISBN: 9780080495385

Purchase options

Purchase options
DRM-free (PDF)
Sales tax will be calculated at check-out

Institutional Subscription

Free Global Shipping
No minimum order


Today, parallel computing experts can solve problems previously deemed impossible and make the "merely difficult" problems economically feasible to solve. This book presents and synthesizes the recent experiences of reknown expert developers who design robust and complex parallel computing applications. They demonstrate how to adapt and implement today's most advanced, most effective parallel computing techniques. The book begins with a highly focused introductory course designed to provide a working knowledge of all the relevant architectures, programming models, and performance issues, as well as the basic approaches to assessment, optimization, scheduling, and debugging.Next comes a series of seventeen detailed case studies—all dealing with production-quality industrial and scientific applications, all presented firsthand by the actual code developers. Each chapter follows the same comparison-inviting format, presenting lessons learned and algorithms developed in the course of meeting real, non-academic challenges. A final section highlights the case studies' most important insights and turns an eye to the future of the discipline.

Key Features

* Provides in-depth case studies of seventeen parallel computing applications, some built from scratch, others developed through parallelizing existing applications.

* Explains elements critical to all parallel programming environments, including:
** Terminology and architectures
** Programming models and methods
** Performance analysis and debugging tools

* Teaches primarily by example, showing how scientists in many fields have solved daunting problems using parallel computing.

* Covers a wide range of application areas—biology, aerospace, semiconductor design, environmental modeling, data imaging and analysis, fluid dynamics, and more.

* Summarizes the state of the art while looking to the future of parallel computing.

Presents technical animations and visualizations from many of the applications detailed in the case studies via a companion web site.

Table of Contents

  • Contents


    Color Plates

    PART I - The Parallel Computing Environment

    Chapter 1 - Parallel Computing Architectures

    Alice E. Koniges, David C. Eder, Margaret Cahir

    1.1 Historical Parallel Computing Architectures

    1.2 Contemporary Parallel Computing Architectures

    1.2.1 MPP Processors

    1.2.2 MPP Memory

    1.2.3 MPP Interconnect Network


    Chapter 2 - Parallel Application Performance

    Alice E. Koniges

    2.1 Defining Performance

    2.2 Measuring Performance

    2.2.1 MPP Application Speedup


    Chapter 3 - Programming Models and Methods

    Margaret Cahir, Robert Moench, Alice E. Koniges

    3.1 Message-Passing Models

    3.1.1 PVM

    3.1.2 MPI

    3.1.3 SHMEM

    3.2 Data-Parallel Models

    3.2.1 High-Performance Fortran

    3.3 Parallel Programming Methods

    3.3.1 Nested- and Mixed-Model Methods

    3.3.2 POSIX Threads and Mixed Models

    3.3.3 Compiler Extensions for Explicit Parallelism with Distributed Objects

    3.3.4 Work-Sharing Models


    Chapter 4 - Parallel Programming Tools

    Margaret Cahir, Robert Moench, Alice E. Koniges

    4.1 The Apprentice Performance Analysis Tool

    4.2 Debuggers

    4.2.1 Process Control

    4.2.2 Data Viewing

    Chapter 5 - Optimizing for Single-Processor Performance

    Jeff Brooks, Sara Graffunder, Alice E. Koniges

    5.1 Using the Functional Units Effectively

    5.2 Hiding Latency with the Cache

    5.3 Stream Buffer Optimizations

    5.4 E-Register Operations

    5.5 How Much Performance Can Be Obtained on a Single Processor?


    Chapter 6 - Scheduling Issues

    Morris A. Jette

    6.1 Gang Scheduler Implementation

    6.2 Gang Scheduler Performance


    PART II - The Applications

    Chapter 7 - Ocean Modeling and Visualization

    Yi Chao, P. Peggy Li, Ping Wang, Daniel S. Katz, Benny N. Cheng, Scott Whitman

    7.1 Introduction

    7.2 Model Description

    7.3 Computational Considerations

    7.3.1 Parallel Software Tools

    7.3.2 Compiler Options

    7.3.3 Memory Optimization and Arithmetic Pipelines

    7.3.4 Optimized Libraries

    7.3.5 Replacement of If/Where Statements by Using Mask Arrays

    7.3.6 Computational Performance

    7.4 Visualization on MPP Machines

    7.5 Scientific Results

    7.6 Summary and Future Challenges



    Chapter 8 - Impact of Aircraft on Global Atmospheric Chemistry

    Douglas A. Rotman, John R. Tannahill, Steven L. Baughcum

    8.1 Introduction

    8.2 Industrial Considerations

    8.3 Project Objectives and Application Coder

    8.4 Computational Considerations

    8.4.1 Why Use an MPP?

    8.4.2 Programming Considerations

    8.4.3 Algorithm Considerations

    8.5 Computational Results

    8.5.1 Performance

    8.5.2 Subsidiary Technology

    8.6 Industrial Results

    8.7 Summary


    Chapter 9 - Petroleum Reservoir Management

    Michael DeLong, Allyson Gajraj, Wayne Joubert, Olaf Lubeck, James Sanderson, Robert E. Stephenson, Gautam S. Shiralkar, Bart van Bloemen Waanders

    9.1 Introduction

    9.2 The Need for Parallel Simulations

    9.3 Basic Features of the Falcon Simulator

    9.4 Parallel Programming Model and Implementation

    9.5 IMPES Linear Solver

    9.6 Fully Implicit Linear Solver

    9.7 Falcon Performance Results

    9.8 Amoco Field Study

    9.9 Summary



    Chapter 10 - An Architecture-Independent Navier-Stokes Code

    Johnson C. T. Wang, Stephen Taylor

    10.1 Introduction

    10.2 Basic Equations

    10.2.1 Nomenclature

    10.3 A Navier-Stokes Solver

    10.4 Parallelization of a Navier-Stokes Solver

    10.4.1 Domain Decomposition

    10.4.2 Parallel Algorithm

    10.5 Computational Results

    10.5.1 Supersonic Flow over Two Wedges

    10.5.2 Titan IV Launch Vehicle

    10.5.3 Delta II 7925 Vehicle

    10.6 Summary



    Chapter 11 - Gaining Insights into the Flow in a Static Mixer

    Olivier Byrde, Mark L. Sawley

    11.1 Introduction

    11.1.1 Overview

    11.1.2 Description of the Application

    11.2 Computational Aspects

    11.2.1 Why Use an MPP?

    11.2.2 Flow Computation

    11.2.3 Particle Tracking

    11.3 Performance Results

    11.3.1 Flow Computation

    11.3.2 Particle Tracking

    11.4 Industrial Results

    11.4.1 Numerical Solutions

    11.4.2 Optimization Results

    11.4.3 Future Work

    11.5 Summary



    Chapter 12 - Modeling Groundwater Flow and Contaminant Transport

    William J. Bosil, Steven F. Ashby, Chuck Baldwin, Robert D. Falgout, Steven G. Smith, Andrew F. B. Tompson

    12.1 Introduction

    12.2 Numerical Simulation of Groundwater Flow

    12.2.1 Flow and Transport Model

    12.2.2 Discrete Solution Approach

    12.3 Parallel Implementation

    12.3.1 Parallel Random Field Generation

    12.3.2 Preconditioned Conjugate Gradient Solver

    12.3.3 Gridding and Data Distribution

    12.3.4 Parallel Computations in ParFlow

    12.3.5 Scalability

    12.4 The MGCG Algorithm

    12.4.1 Heuristic Semicoarsening Strategy

    12.4.2 Operator-Induced Prolongation and Restriction

    12.4.3 Definition of Coarse Grid Operator

    12.4.4 Smoothers

    12.4.5 Coarsest Grid Solvers

    12.4.6 Stand-Alone Multigrid versus Multigrid As a Preconditioner

    12.5 Numerical Results

    12.5.1 The Effect of Coarsest Grid Solver Strategy

    12.5.2 Increasing the Spatial Resolution

    12.5.3 Enlarging the Size of the Domain

    12.5.4 Increasing the Degree of Heterogeneity

    12.5.5 Parallel Performance on the Cray T3D

    12.6 Summary



    Chapter 13 - Simulation of Plasma Reactors

    Stephen Taylor, Marc Rieffel, Jerrell Watts, Sadasivan Shankar

    13.1 Introduction

    13.2 Computational Considerations

    13.2.1 Grid Generation and Partitioning Techniques

    13.2.2 Concurrent DSMC Algorithm

    13.2.3 Grid Adaption Technique

    13.2.4 Library Technology

    13.3 Simulation Results

    13.4 Performance Results

    13.5 Summary



    Chapter 14 - Electron-Molecule Collisions for Plasma Modeling

    Carl Winstead, Chuo-Han Lee, Vincent McKoy

    14.1 Introduction

    14.2 Computing Electron-Molecule Cross Sections

    14.2.1 Theoretical Outline

    14.2.2 Implementation

    14.2.3 Parallel Organization

    14.3 Performance

    14.4 Summary



    Chapter 15 - Three-Dimensional Plasma Particle-in-Cell Calculations of Ion Thruster Backflow Contamination

    Robie I. Samanta Roy, Daniel E. Hastings, Stephen Taylor

    15.1 Introduction

    15.2 The Physical Model

    15.2.1 Beam Ions

    15.2.2 Neutral Efflux

    15.2.3 CEX Propellant Ions

    15.2.4 Electrons

    15.3 The Numerical Model

    15.4 Parallel Implementation

    15.4.1 Partitioning

    15.4.2 Parallel PIC Algorithm

    15.5 Results

    15.5.1 3D Plume Structure

    15.5.2 Comparison of 2D and 3D Results

    15.6 Parallel Study

    15.7 Summary



    Chapter 16 - Advanced Atomic-Level Materials Design

    Lin H. Yang

    16.1 Introduction

    16.2 Industrial Considerations

    16.3 Computational Considerations and Parallel Implementations

    16.4 Applications to Grain Boundaries in Polycrystalline Diamond

    16.5 Summary



    Chapter 17 - Solving Symmetric Eigenvalue Problems

    David C. O'Neal, Raghurama Reddy

    17.1 Introduction

    17.2 Jacobi's Method

    17.3 Classical Jacobi Method

    17.4 Serial Jacobi Method

    17.5 Tournament Orderings

    17.6 Parallel Jacobi Method

    17.7 Macro Jacobi Method

    17.8 Computational Experiments

    17.8.1 Test Problems

    17.8.2 Convergence

    17.8.3 Scaling

    17.9 Summary



    Chapter 18 - Nuclear Magnetic Resonance Simulations

    Alan J. Benesi, Kenneth M. Merz, James J. Vincent, Ravi Subramanya

    18.1 Introduction

    18.2 Scientific Considerations

    18.3 Description of the Application

    18.4 Computational Considerations

    18.4.1 Algorithmic Considerations

    18.4.2 Programming Considerations

    18.5 Computational Results

    18.6 Scientific Results

    18.6.1 Validation of Simulation

    18.6.2 Interesting Scientific Results

    18.7 Summary



    Chapter 19 - Molecular Dynamics Simulations Using Particle-Mesh Ewald Methods

    Michael F. Crowley, David W. Deerfield II, Tom A. Darden, Thomas E. Cheatham III

    19.1 Introduction: Industrial Considerations

    19.1.1 Overview

    19.1.2 Cutoff Problem for Long-Distance Forces

    19.1.3 Particle-Mesh Ewald Method

    19.2 Computational Considerations

    19.2.1 Why Use an MPP?

    19.2.2 Parallel PME

    19.2.3 Coarse-Grain Parallel PME

    19.3 Computational Results

    19.3.1 Performance

    19.3.2 Parallel 3D FFT and Groups

    19.4 Industrial Strength Results

    19.5 The Future

    19.6 Summary


    Chapter 20 - Radar Scattering and Antenna Modeling

    Tom Cwik, Cinzia Zuffada, Daniel S. Katz, Jay Parker

    20.1 Introduction

    20.2 Electromagnetic Scattering and Radiation

    20.2.1 Formulation of the Problem

    20.2.2 Why This Formulation Addresses the Problem

    20.3 Finite Element Modeling

    20.3.1 Discretization of the Problem

    20.3.2 Why Use a Scalable MPP?

    20.4 Computational Formulation and Results

    20.4.1 Constructing the Matrix Problem

    20.4.2 Beginning the Matrix Solution

    20.4.3 Completing the Solution of the Matrix Problem

    20.4.4 The Three Stages of the Application

    20.5 Results for Radar Scattering and Antenna Modeling

    20.5.1 Anistropic Scattering

    20.5.2 Patch Antennas-Modeling Conformal Antennas with PHOEBE

    20.6 Summary and Future Challenges



    Chapter 21 - Functional Magnetic Resonance Imaging Dataset Analysis

    Nigel H. Goddard, Greg Hood, Jonathan D. Cohen, Leigh E. Nystrom, William F. Eddy, Christopher R. Genovese, Douglas C. Noll

    21.1 Introduction

    21.2 Industrial Considerations

    21.2.1 Overview

    21.2.2 Description of the Application

    21.2.3 Parallelization and the Online Capability

    21.3 Computational Considerations

    21.3.1 Why Use an MPP?

    21.3.2 Programming Considerations

    21.3.3 Algorithm Considerations

    21.4 Computational Results

    21.4.1 Performance

    21.4.2 Subsidiary Technologies

    21.5 Clinical and Scientific Results

    21.5.1 Supercomputing '96 Demonstration

    21.5.2 Science Application

    21.5.3 What Are the Next Problems to Tackle?

    21.6 Summary



    Chapter 22 - Selective and Sensitive Comparison of Genetic Sequence Data

    Alexander J. Ropelewski, Hugh B. Nicholas, Jr., David W. Deerfield II

    22.1 Introduction

    22.2 Industrial Considerations

    22.2.1 Overview/Statement of the Problem

    22.3 Approaches Used to Compare Sequences

    22.3.1 Visualization of Sequence Comparison

    22.3.2 Basic Sequence-Sequence Comparison Algorithm

    22.3.3 Basic Sequence-Profile Comparison Algorithm

    22.3.4 Other Approaches to Sequence Comparison

    22.4 Computational Considerations

    22.4.1 Why Use an MPP?

    22.4.2 Programming Considerations

    22.4.3 Algorithm Considerations

    22.5 Computational Results

    22.5.1 Performance

    22.6 Industrial and Scientific Considerations

    22.7 The Next Problems to Tackle

    22.8 Summary



    Chapter 23 - Interactive Optimization of Video Compression Algorithms

    Henri Nicolas, Fred Jordan

    23.1 Introduction

    23.2 Industrial Considerations

    23.3 General Description of the System

    23.3.1 General Principle

    23.3.2 Main Advantages Offered by Direct View

    23.4 Parallel Implementation

    23.4.1 Remarks

    23.5 Description of the Compression Algorithm

    23.6 Experimental Results

    23.7 Summary



    PART III - Conclusions and Predictions

    Chapter 24 - Designing Industrial Parallel Applications

    Alice E. Koniges, David C. Eder, Michael A. Heroux

    24.1 Design Lessons from the Applications

    24.1.1 Meso- to Macroscale Environmental Modeling

    24.1.2 Applied Fluid Dynamics

    24.1.3 Applied Plasma Dynamics

    24.1.4 Material Design and Modeling

    24.1.5 Data Analysis

    24.2 Design Issues

    24.2.1 Code Conversion Issues

    24.2.2 The Degree of Parallelism in the Application

    24.3 Additional Design Issues

    Chapter 25 - The Future of Industrial Parallel Computing

    Michael A. Heroux, Horst Simon, Alice E. Koniges

    25.1 The Role of Parallel Computing in Industry

    25.2 Microarchitecture Issues

    25.2.1 Prediction

    25.2.2 Discussion

    25.3 Macroarchitecture Issues

    25.3.1 Prediction

    25.3.2 Discussion

    25.4 System Software Issues

    25.4.1 Prediction

    25.4.2 Discussion

    25.5 Programming Environment Issues

    25.5.1 Prediction

    25.5.2 Discussion

    25.6 Applications Issues

    25.6.1 Prediction

    25.6.2 Discussion

    25.7 Parallel Computing in Industry

    25.7.1 Area 1: Parallel Execution of a Single Analysis: Incompressible CFD Analysis

    25.7.2 Area 2: Design Optimization: Noise, Vibration, Harshness (NVH) Analysis

    25.7.3 Area 3: Design Studies: Crash Design Optimization

    25.7.4 Area 4: Interactive, Intuitive, Immersive Simulation Environments: Large- Scale Particle Tracing

    25.8 Looking Forward: The Role of Parallel Computing in the Digital Information Age

    25.8.1 Increasing the Demand for Parallel Computing

    25.8.2 The Importance of Advanced User Interfaces

    25.8.3 Highly Integrated Computing

    25.8.4 The Future


    Appendix: Mixed Models with Pthreads and MPI

    Vijay Sonnad, Chary G. Tamirisa, Gyan Bhanot




Product details

  • No. of pages: 597
  • Language: English
  • Copyright: © Morgan Kaufmann 1999
  • Published: October 11, 1999
  • Imprint: Morgan Kaufmann
  • eBook ISBN: 9780080495385

About the Editor

Alice Koniges

Alice E. Koniges is an internationally known authority on parallel application development. As leader of the Parallel Applications Technology Program at Lawrence Livermore National Laboratory, she directed researchers in the largest set of agreements between industries and national laboratories ever funded by the US Department of Energy. She has served as a consultant to the Max Planck Institutes of Germany on parallelization and high performance computing issues. She was the first woman to be awarded a Ph.D. in Applied Mathematics from Princeton University.

About the Contributors

Ratings and Reviews

Write a review

There are currently no reviews for "Industrial Strength Parallel Computing"