Parallel computing: theory and practice / Michael J. Quinn.

Rev. ed. of: Designing efficient algorithms for parallel computers. c1987.

Includes bibliographical references (p. 391-433) and index.

1. Introduction -- 2. PRAM Algorithms -- 3. Processor Arrays, Multiprocessors, and Multicomputers -- 4. Parallel Programming Languages -- 5. Mapping and Scheduling -- 6. Elementary Parallel Algorithms -- 7. Matrix Multiplication -- 8. The Fast Fourier Transform -- 9. Solving Linear Systems -- 10. Sorting -- 11. Dictionary Operations -- 12. Graph Algorithms -- 13. Combinatorial Search -- App. A Graph Theoretic Terminology -- App. B Review of Complex Numbers -- App. C Parallel Algorithm Design Strategies.

This text provides an exceptional introduction to parallel computing by balancing theory and practice. The emphasis is on designing, analyzing and implementing parallel algorithms suitable for execution on real parallel computers. Early chapters set the stage by introducing key concepts, illustrating fundamental parallel algorithms, and describing ways to incorporate high-level parallelism into hardware and software.

Later chapters explore the development of parallel algorithms for matrix multiplication, the fast Fourier transform, solving linear systems, sorting, searching, graph theoretic problems, and combinatorial search. Numerous graphs illustrate the speedups that can be achieved on actual parallel hardware by implementing the parallel algorithms developed in the text. As a result, students learn how to make efficient use of emerging parallel computer technology.

.

A number of additional features make this book distinctive: it surveys some of the most popular parallel computer architectures, including Thinking Machines' CM-5, Intel's Paragon XP/S, and the Sequent Symmetry; it covers some of the most popular parallel programming languages, including Fortran 90, C*, Linda and OCCAM; a glossary of parallel computing terminology contains all the terms defined in the text; more than 200 exercises cover the gamut from proofs to programming assignments; the exceptionally large bibliography provides ample opportunities for further study; and Library of Congress call numbers simplify access to frequently cited journals and proceedings.