GPUs may have started life as graphics processors, but recently they've emerged as a fantastic numerical co-processor for high-performance general applications on the CPU. This book not only teaches you the fundamentals of parallel programming with GPUs, it helps you think in parallel. You learn best practices, algorithms, and designs for achieving greater application performance with these processors. Amazon recently added GPU supercomputing to its cloud-computing platform - a clear sign that parallel programming is becoming an essential skill. This book includes valuable input from major CPU and GPU manufacturers - Intel, NVIDIA and AMD - to help experienced programmers get a head start on programming GPU applications. * Understand the differences between parallel and sequential programming * Learn about GPU architecture, including the runtime environment, threads, and memory * Build and deploy GPU applications and libraries - and port existing applications * Use debugging and profiling tools and techniques * Write GPU programs for clusters and the cloud * Design programs that will take advantage of future enhancements to GPU technology - including the trend of putting CPU and GPU cores on a single chip