book

Heterogeneous Computing with OpenCL 2.0

by David R. Kaeli, Perhaad Mistry, Dana Schaa, Dong Ping Zhang

June 2015

Intermediate to advanced

330 pages

9h 6m

English

Morgan Kaufmann

Read now

Unlock full access

Abstract1.1 Introduction to Heterogeneous Computing1.2 The Goals of This Book1.3 Thinking Parallel1.4 Concurrency and Parallel Programming Models1.5 Threads and Shared Memory1.6 Message-Passing Communication1.7 Different Grains of Parallelism1.8 Heterogeneous Computing with OpenCL1.9 Book Structure
Abstract2.1 Introduction2.2 Hardware Trade-offs2.3 The Architectural Design Space2.4 Summary

Abstract3.1 Introduction3.2 The OpenCL Platform Model3.3 The OpenCL Execution Model3.4 Kernels and the OpenCL Programming Model3.5 OpenCL Memory Model3.6 The OpenCL Runtime with an Example3.7 Vector Addition Using an OpenCL C++ Wrapper3.8 OpenCL for CUDA Programmers3.9 Summary
Abstract4.1 OpenCL Examples4.2 Histogram4.3 Image Rotation4.4 Image Convolution4.5 Producer-Consumer4.6 Utility Functions4.7 Summary
Abstract5.1 Commands and the Queuing Model5.2 Multiple Command-Queues5.3 The Kernel Execution Domain: Work-Items, Work-Groups, and NDRanges5.4 Native and Built-In Kernels5.5 Device-Side Queuing5.6 Summary
Abstract6.1 Memory Objects6.2 Memory Management6.3 Shared Virtual Memory6.4 Summary
Abstract7.1 Synchronization and Communication7.2 Global Memory7.3 Constant Memory7.4 Local Memory7.5 Private Memory7.6 Generic Address Space7.7 Memory Ordering7.8 Summary
Abstract8.1 OpenCL on an AMD FX-8350 CPU8.2 OpenCL on the AMD Radeon R9 290X GPU8.3 Memory Performance Considerations in OpenCL8.4 Summary
Abstract9.1 Introduction9.2 The Feature Histogram on the CPU9.3 OpenCL Implementation9.4 Performance Analysis9.5 Conclusion
Abstract10.1 Introduction10.2 Profiling OpenCL Code Using Events10.3 AMD CodeXL10.4 Profiling Using CodeXL10.5 Analyzing Kernels Using CodeXL10.6 Debugging OpenCL Kernels Using CodeXL10.7 Debugging Using printf10.8 Summary
Abstract11.1 Introduction11.2 A Brief Introduction to C++ AMP11.3 OpenCL 2.0 as a Compiler Target11.4 Mapping Key C++ AMP Constructs to OpenCL11.5 C++ AMP Compilation Flow11.6 Compiled C++ AMP Code11.7 How Shared Virtual Memory in OpenCL 2.0 Fits in11.8 Compiler Support for Tiling in C++AMP11.9 Address Space Deduction11.10 Data Movement Optimization11.11 Binomial Options: A Full Example11.12 Preliminary Results11.13 Conclusion
Abstract12.1 Introduction12.2 Programming with WebCL12.3 Synchronization12.4 Interoperability with WebGL12.5 Example Application12.6 Security Enhancement12.7 WebCL on the Server12.8 Status and Future of WebCLWorks Cited
Abstract13.1 Introduction13.2 Beyond C and C+ +13.3 Haskell OpenCL13.4 Summary