Multi-core Programmiong with Cuda and Opencl

"Multi-Core Programming With Cuda And Opencl" covers multi-core GPU programming with some emphasis on CPU vs GPU hardware, with benchmark/performance tests to show improvements from purely a CPU approach to a GPU approach to processing data using several different libraries including CUDA, OpenCL, and DirectCompute. It includes topics such as differences and capabilities of hardware architectures, parallel API's and algorithm design for parallel machines. Emphasis will be on practical, applied use of these multi-core GPU libraries. In order for these new architectures to be utilized, programmers must educate themselves on the advantages, limitations, and algorithm design methods for parallel computing platforms. This is a very important but challenging subject that encompasses multi-threading, symmetric multi-processing (SMP), and parallel processing on the CPU side, and massively parallel processing on the GPU side. This book includes end-of-chapter exercises, discussions, and quizzes.