Parallel Programming in OpenMP

The rapid and widespread acceptance of shared-memory multiprocessor architectures has created a pressing demand for an efficient way to program these systems. At the same time, developers of technical and scientific applications in industry and in government laboratories find they need to parallelize huge volumes of code in a portable fashion. OpenMP, developed jointly by several parallel computing vendors to address these issues, is an industry-wide standard for programming shared-memory and distributed shared-memory multiprocessors. It consists of a set of compiler directives and library routines that extend FORTRAN, C, and C++ codes to express shared-memory parallelism. Parallel Programming in OpenMP is the first book to teach both the novice and expert parallel programmers how to program using this new standard. The authors, who helped design and implement OpenMP while at SGI, bring a depth and breadth to the book as compiler writers, application developers, and performance engineers. * Designed so that expert parallel programmers can skip the opening chapters, which introduce parallel programming to novices, and jump right into the essentials of OpenMP. * Presents all the basic OpenMP constructs in FORTRAN, C, and C++. * Emphasizes practical concepts to address the concerns of real application developers. * Includes high quality example programs that illustrate concepts of parallel programming as well as all the constructs of OpenMP. * Serves as both an effective teaching text and a compact reference. * Includes end-of-chapter programming exercises. Parallel Programming in OpenMP 4 Copyright Page 5 Contents 10 Foreword 8 Preface 14 Chapter 1. Introduction 18 1.1 Performance with OpenMP 19 1.2 A First Glimpse of OpenMP 23 1.3 The OpenMP Parallel Computer 25 1.4 Why OpenMP? 26 1.5 History of OpenMP 30 1.6 Navigating the Rest of the Book 31 Chapter 2. Getting Started with OpenMP 32 2.1 Introduction 32 2.2 OpenMP from 10,000 Meters 33 2.3 Parallelizing a Simple Loop 40 2.4 A More Complicated Loop 46 2.5 Explicit Synchronization 49 2.6 The reduction Clause 52 2.7 Expressing Parallelism with Parallel Regions 53 2.8 Concluding Remarks 56 2.9 Exercises 57 Chapter 3. Exploiting Loop-Level Parallelism 58 3.1 Introduction 58 3.2 Form and Usage of the parallel do Directive 59 3.3 Meaning of the parallel do Directive 63 3.4 Controlling Data Sharing 64 3.5 Removing Data Dependences 82 3.6 Enhancing Performance 99 3.7 Concluding Remarks 107 3.8 Exercises 107 Chapter 4. Beyond Loop-Level Parallelism: Parallel Regions 110 4.1 Introduction 110 4.2 Form and Usage of the parallel Directive 111 4.3 Meaning of the parallel Directive 114 4.4 threadprivate Variables and the copyin Clause 117 4.5 Work-Sharing in Parallel Regions 125 4.6 Restrictions on Work-Sharing Constructs 136 4.7 Orphaning of Work-Sharing Constructs 140 4.8 Nested Parallel Regions 143 4.9 Controlling Parallelism in an OpenMP Program 147 4.10 Concluding Remarks 154 4.11 Exercises 155 Chapter 5. Synchronization 158 5.1 Introduction 158 5.2 Data Conflicts and the Need for Synchronization 159 5.3 Mutual Exclusion Synchronization 164 5.4 Event Synchronization 174 5.5 Custom Synchronization: Rolling Your Own 179 5.6 Some Practical Considerations 182 5.7 Concluding Remarks 185 5.8 Exercises 185 Chapter 6. Performance 188 6.1 Introduction 188 6.2 Key Factors That Impact Performance 190 6.3 Performance-Tuning Methodology 215 6.4 Dynamic Threads 218 6.5 Bus-Based and NUMA Machines 221 6.6 Concluding Remarks 224 6.7 Exercises 224 Appendix A: A Quick Reference to OpenMP 228 References 234 Index 238

The rapid and widespread acceptance of shared-memory multiprocessor architectures has created a pressing demand for an efficient way to program these systems. At the same time, developers of technical and scientific applications in industry and in government laboratories find they need to parallelize huge volumes of code in a portable fashion. OpenMP, developed jointly by several parallel computing vendors to address these issues, is an industry-wide standard for programming shared-memory and distributed shared-memory multiprocessors. It consists of a set of compiler directives and library routines that extend FORTRAN, C, and C++ codes to express shared-memory parallelism.

Parallel Programming in OpenMP is the first book to teach both the novice and expert parallel programmers how to program using this new standard. The authors, who helped design and implement OpenMP while at SGI, bring a depth and breadth to the book as compiler writers, application developers, and performance engineers.

* Designed so that expert parallel programmers can skip the opening chapters, which introduce parallel programming to novices, and jump right into the essentials of OpenMP.
* Presents all the basic OpenMP constructs in FORTRAN, C, and C++.
* Emphasizes practical concepts to address the concerns of real application developers.
* Includes high quality example programs that illustrate concepts of parallel programming as well as all the constructs of OpenMP.
* Serves as both an effective teaching text and a compact reference.
* Includes end-of-chapter programming exercises. Metal cutting is an essential process throughout engineering design and manufacturing industries. To increase efficiency and reduce costs, it is necessary to improve understanding of the metal cutting process. This book presents a comprehensive treatment of the subject that focuses on the features of the behavior of tool and work materials that influence the efficiency of metal cutting operations. The fourth edition of this acclaimed book has been expanded and revised to include significant changes and additions to metal cutting theory, and to cover developments in tool materials and industrial practice. In particular, improvements in the understanding of the generation of heat and distribution of temperature in the cutting tool are described; a discussion of the structure, properties, and performance of newly developed ceramic tool materials and tool coatings is presented; new information of the machinability of alloys is given; and the introduction of calcium deoxidized steels and their improved machinability are assessed. Additionally, a material selection and design-based approach is expanded upon to improve industrial relevance.; "Metal Cutting" provides invaluable information for those engaged in machining, toolmaking, and related engineering activities, and it serves as a useful introduction to the subject for students of metallurgy and engineering. It presents a comprehensive treatment of the subject. It includes information on significant changes and additions to metal cutting theory. It offers industrial relevance through a materials selection and design-based approach "The rapid and widespread acceptance of shared memory multiprocessor architectures has created a pressing demand for an efficient way to program these systems. At the same time, developers of technical and scientific applications in industry and in government laboratories find they need to parallelize huge volumes of code in a portable fashion. OpenMP, developed jointly by several parallel computing vendors to address these issues, is an industry-wide standard for programming shared-memory and distributed shared-memory multiprocessors. It consists of a set of compiler directives and library routines that extend FORTRAN, C, and C++ codes to express shared-memory parallelism."