The book can serve as a tutorial, textbook, or research reference. It has three major parts. Part 1 discusses parallel models and methods. The authors point out, correctly, that parallel computing is usually applied to solving large problems, which requires debugging large data sets and achieving high performance. Achieving peak performance depends on optimizing performance for the single-processor case, as well as designing a program that fits the target multiprocessor architecture. This part concludes with a section on parallel programming tools, optimizing for single-processor performance, and scheduling.
In Part 2, the authors discuss 16 challenging applications. Each is presented carefully by subject experts who provide problem definitions, parallel implementations including algorithms and programming considerations, and computational results. Specific numerical results are assessed, with an emphasis on performance. Most applications are industrial-scale problems that require parallel computing.
Finally, Part 3 explores lessons learned from industrial parallel computation. The very last section of this part discusses the future of industrial parallel computing. It includes architectural issues; programming environments; and the exciting prospect of combining graphics, data management, and numerical simulation into a single high-performance system that will be available to future generations of scientists and engineers, providing them with an extremely powerful tool for solving the most challenging problems. I believe that this vision is right and achievable. The book is a small but important step in this direction. It contributes to educating a wide range of potential users and convincing them that parallel computing could help them to innovate and to successfully attack difficult problems that would be computationally intractable without this high-performance technology.
The book is an unqualified success in terms of its quality, contents, and achieved goals. Without exaggeration, it is one of the best sources of technical information on practical high-performance computing. I recommend it to every academic or industrial practitioner of parallel computing. I have benefited greatly from it myself.