Browse Books

Reviewer: Jeff D. Parker

Michael J. Quinn has written an excellent introduction to parallel computation. His book is that rare combination that is accessible to the undergraduate, but is also a resource for the graduate student or scholar. For the undergraduate, there is a clear presentation of the fundamentals of each topic covered. The author accomplishes this through clear prose, good examples, well-executed graphics, and exercises that are within reach of the student. Nothing is too well known to escape definition. There is even a 10-page glossary of terms, from acyclic to iterative deepening. Issues are presented fully: after a discussion of the advantages of parallelism, he presents a sequence of arguments against parallelism. The implications of these arguments are put into context when he turns to computer architectures. The graduate student is given an excellent survey of each topic, bibliographic notes in each chapter, and a 20-page bibliography at the end. The book discusses the major models of computation and attempts throughout to review the performance of each class of algorithms under each model. The student gains a strong intuition about the bottlenecks in each architecture and the traits that make an algorithm amenable to parallelization. The topics covered include the usual suspects, such as sorting, graph algorithms, matrix multiplication, and numerical algorithms, but there are also sections on dictionary operations, combinational search, and logic programming. The best news is reserved for the instructor. With this text, even those who are not experts in the field can present a convincing course that combines a solid overview of the major issues with a selection of topics to suit the interests and abilities of a variety of audiences.