........ published in NEWSLETTER # 50
[ NATO-PCO Home Page ] [ Table of Contents of NEWSLETTER # 50 ]
........ published in NEWSLETTER # 50
SOFTWARE FOR PARALLEL COMPUTING
by Dr. J.S. Kowalik, Boeing Computer Services, Seattle/WA (U.S.A.)
Very significant progress in computer system technology has been made over the past 40 years. The speed of computing as well as ease of use of computers has been greatly improved. But new challenges have arisen. One of them is an improved understanding of new architectural innovations such as the parallel computer system and associated software for parallel computation.
This book (NATO ASI SERIES F106) is focused on parallel software, programming tools, and practical experience of using parallel computers for solving practical problems. The largest section of the book covers: `Tools and Methods for Parallel Computing'. The ability to use software tools for developing and porting parallel computation is critical for the successful transition to high performance parallelism. Several experts from industry and academia discuss tools for heterogeneous network computing, and new parallel languages for massively parallel processes. Also, the timely subject of object oriented programming methods suitable for numerical computing on parallel machines is discussed.
Related to software development is the problem of graphical support for parallel debugging. This topic is presented in the graphics section of the book.
Construction of algorithms for multiprocessor systems may be difficult for programmers who are more familiar with uniprocessor systems. Several contributions illustrate the issues of and approaches to parallel algorithm design and implementation.
The final part of the book deals with a non_trivial problem of performance for parallel programs and systems. It offers a definition of parallel processing system scalability and shows examples of a good scalable performance. Other aspects of performance are discussed: performance prediction from data transport, dynamic load balancing and dynamic scheduling techniques for irregularly structured computation.
The book introduces readers to the key issues in high performance computing and can serve as a pointer to other sources of information. It should be of interest to high performance computing researchers and practitioners in industry and academia, including graduate students and information system managers.
Reference books: E9, F70, F103, F106