High Performance Computing (PDF)
High Performance Computing, written by Charles Severance and Kevin Dowd, is a comprehensive guide to the world of high-performance computing (HPC). The book delves into the concepts, tools, and techniques needed to build and operate HPC systems, and provides a comprehensive overview of the various aspects of HPC, including hardware, software, and networking.
The book is divided into three sections. The first section provides an introduction to the world of HPC, and covers topics such as the history of HPC, the various types of HPC systems, and the challenges of building and operating these systems. The second section covers the hardware components of HPC systems, including processors, memory, storage, and networking. The authors discuss the latest advances in HPC hardware, including the use of GPUs and other accelerators, and provide practical advice on how to choose and configure HPC hardware for specific applications.
The third section of the book covers software and applications for HPC systems. The authors provide an overview of the major HPC software tools and frameworks, including MPI, OpenMP, and CUDA, and explain how to use these tools to optimize HPC applications for maximum performance. They also discuss the challenges of porting existing applications to HPC systems and provide advice on how to overcome these challenges.
Throughout the book, the authors emphasize the importance of performance tuning and optimization for HPC applications. They provide numerous examples and case studies that illustrate the impact of optimization on application performance, and offer practical advice on how to optimize applications for different HPC architectures and workloads.
High Performance Computing is an essential resource for anyone interested in the world of HPC, from researchers and scientists to engineers and IT professionals. With its clear and concise explanations, practical advice, and comprehensive coverage of HPC concepts, tools, and techniques, this book is an invaluable guide for anyone looking to build, operate, or optimize HPC systems.