大规模并行处理器程序设计（英文版 原书第3版） 2025 chm pdf kindle rb azw3 下载 115盘

本书介绍并行编程和GPU架构的基本概念，详细探索了构建并行程序的各种技术，涵盖性能、浮点格式、并行模式和动态并行等主题，适合专业人士及学生阅读。书中通过案例研究展示了开发过程，从计算思维的细节着手，*终给出了高效的并行程序示例。新版更新了关于CUDA的讨论，包含CuDNN等新的库，同时将不再重要的内容移到附录中。新版还增加了关于并行模式的两个新章节，并更新了案例研究，以反映当前的行业实践。

Preface Acknowledgements

CHAPTER.1 Introduction.1

1.1 Heterogeneous Parallel Computing..2

1.2 Architecture of a Modern GPU.6

1.3 Why More Speed or Parallelism?.8

1.4 Speeding Up Real Applications..10

1.5 Challenges in Parallel Programming ..12

1.6 Parallel Programming Languages and Models.12

1.7 Overarching Goals..14

1.8 Organization of the Book..15

References ..18

CHAPTER.2 Data Parallel Computing.19

2.1 Data Parallelism..20

2.2 CUDA C Program Structure.22

2.3 A Vector Addition Kernel .25

2.4 Device Global Memory and Data Transfer.27

2.5 Kernel Functions and Threading.32

2.6 Kernel Launch.37

2.7 Summary.38

Function Declarations.38

Kernel Launch.38

Built-in (Predefined) Variables .39

Run-time API..39

2.8 Exercises.39

References ..41

CHAPTER.3 Scalable Parallel Execution..43

3.1 CUDA Thread Organization.43

3.2 Mapping Threads to Multidimensional Data..47

3.3 Image Blur: A More Complex Kernel ..54

3.4 Synchronization and Transparent Scalability .58

3.5 Resource Assignment..60

3.6 Querying Device Properties..61

3.7 Thread Scheduling and Latency Tolerance.64

3.8 Summary.67

3.9 Exercises.67

CHAPTER.4 Memory and Data Locality .71

4.1 Importance of Memory Access Efficiency..72

4.2 Matrix Multiplication..73

4.3 CUDA Memory Types..77

4.4 Tiling for Reduced Memory Traffic..84

4.5 A Tiled Matrix Multiplication Kernel.90

4.6 Boundary Checks..94

4.7 Memory as a Limiting Factor to Parallelism..97

4.8 Summary.99

4.9 Exercises.100

CHAPTER.5 Performance Considerations.103

5.1 Global Memory Bandwidth..104

5.2 More on Memory Parallelism.112

5.3 Warps and SIMD Hardware..117

5.4 Dynamic Partitioning of Resources..125

5.5 Thread Granularity..127

5.6 Summary.128

5.7 Exercises.128

References ..130

CHAPTER.6 Numerical Considerations .131

6.1 Floating-Point Data Representation..132

Normalized Representation of M..132

Excess Encoding of E .133

6.2 Representable Numbers..134

6.3 Special Bit Patterns and Precision in IEEE Format..138

6.4 Arithmetic Accuracy and Rounding .139

6.5 Algorithm Considerations.140

6.6 Linear Solvers and Numerical Stability..142

6.7 Summary.146

6.8 Exercises.147

References ..147

CHAPTER.7 Parallel Patterns: Convolution .149

7.1 Background.150

7.2 1D Parallel Convolution—A Basic Algorithm .153

7.3 Constant Memory and Caching..156

7.4 Tiled 1D Convolution with Halo Cells.160

7.5 A Simpler Tiled 1D Convolution—General Caching.165

7.6 Tiled 2D Convolution with Halo Cells.166

7.7 Summary.172

7.8 Exercises.173

CHAPTER.8 Parallel Patterns: Prefix Sum..175

8.1 Background.176

8.2 A Simple Parallel Scan.177

8.3 Speed and Work Efficiency.181

8.4 A More Work-Efficient Parallel Scan.183

8.5 An Even More Work-Efficient Parallel Scan..187

8.6 Hierarchical Parallel Scan for Arbitrary-Length Inputs..189

8.7 Single-Pass Scan for Memory Access Efficiency.192

8.8 Summary.195

8.9 Exercises.195

References ..196

CHAPTER.9 Parallel Patterns Parallel Histogram Computation .. 199

9.1 Background.200

9.2 Use of Atomic Operations ..202

9.3 Block versus Interleaved Partitioning.206

9.4 Latency versus Throughput of Atomic Operations.207

9.5 Atomic Operation in Cache Memory .210

9.6 Privatization..210

9.7 Aggregation ..211

9.8 Summary.213

9.9 Exercises.213

Reference..214

CHAPTER.10 Parallel Patterns: Sparse Matrix Computation .215

10.1 Background..216

10.2 Parallel SpMV Using CSR.219

10.3 Padding and Transposition.221

10.4 Using a Hybrid Approach to Regulate Padding.224

10.5 Sorting and Partitioning for Regularization.227

10.6 Summary..229

10.7 Exercises..229

References ..230

CHAPTER.11 Parallel Patterns: Merge Sort.231

11.1 Background..231

11.2 A Sequential Merge Algorithm..233

11.3 A Parallelization Approach..234

11.4 Co-Rank Function Implementation..236

Contents

11.5 A Basic Parallel Merge Kernel ..241

11.6 A Tiled Merge Kernel..242

11.7 A Circular-Buffer Merge Kernel..249

11.8 Summary..256

11.9 Exercises..256

Reference.

大卫·B. 柯克（David B. Kirk） 美国国家工程院院士，NVIDIA Fellow，曾任NVIDIA公司首席科学家。他领导了NVIDIA图形技术的开发，并且是CUDA技术的创始人之一。2002年，他荣获ACM SIGGRAPH计算机图形成就奖，以表彰其在把高性能计算机图形系统推向大众市场方面做出的杰出贡献。他拥有加州理工学院计算机科学博士学位。

胡文美（Wen-mei W. Hwu） 美国伊利诺伊大学厄巴纳-香槟分校电气与计算机工程系AMD Jerry Sanders讲席教授，并行计算研究中心首席科学家，领导IMPACT团队和CUDA卓越中心的研究工作。他在编译器设计、计算机体系结构、微体系结构和并行计算方面做出了卓越贡献，是IEEE Fellow、ACM Fellow，荣获了包括ACM SigArch Maurice Wilkes Award在内的众多奖项。他还是MulticoreWare公司的联合创始人兼CTO。他拥有加州大学伯克利分校计算机科学博士学位。

暂无出版社相关信息，正在全力查找中！

暂无相关书籍摘录，正在全力查找中！

在线阅读地址：大规模并行处理器程序设计（英文版 原书第3版）在线阅读

在线听书地址：大规模并行处理器程序设计（英文版 原书第3版）在线收听

在线购买地址：大规模并行处理器程序设计（英文版 原书第3版）在线购买

暂无原文赏析，正在全力查找中！

书籍介绍

本书介绍并行编程和GPU架构的基本概念，详细探索了构建并行程序的各种技术，涵盖性能、浮点格式、并行模式和动态并行等主题，适合专业人士及学生阅读。书中通过案例研究展示了开发过程，从计算思维的细节着手，最终给出了高效的并行程序示例。新版更新了关于CUDA的讨论，包含CuDNN等新的库，同时将不再重要的内容移到附录中。新版还增加了关于并行模式的两个新章节，并更新了案例研究，以反映当前的行业实践。