Cellular computation in the form of specialized distributed functional structures is the focus of this book. In addition to presenting distributed functional structures designed to implement a variety of operations, the book discusses how these structures operate on arrays of data in parallel. This form of parallelism is applied to arithmetic processing, non-numerical processing, and data transfer with permutation. The book has a three-level table of contents. Following the preface and two forewords, the chapters are:
Introduction: The Need for High-Performance Computing
Parallel Computing
Universal and Specialized Cellular Arrays: Distributed Functional Structures
Functional Possibilities of Distributed Functional Structures
Solving Numerical Problems
Non-Numerical Data Processing
Interconnection Networks
Implementation of Cellular Microprocessors
The book concludes with four pages of references, a two-page acronyms list, a two-page author index, and a ten-page subject index.
Chapters 3 through 8 are broken down further into the following sections. Chapter 3: “Sequential and Parallel Automata,” “Iterative Networks,” “Parallel Substitution System,” “Homogeneous Computing Media,” and “Distributed Functional Structures.” Chapter 4: “Analysis of &agr;-Structure” and “Analysis of &lgr;-Structure.” Chapter 5: “Binary Arithmetic Based on Vertical Processing,” “Table Look-up Arithmetic Units,” “Arithmetic Computations in Unitary Codes,” and “Arithmetic Units in Residue Number Systems.” Chapter 6: “Features of Non-Numerical Problems,” “Database Machines,” “Labelled Array Method,” “Positional Sets,” “Combinatorial Memory,” “Modular Non-Numerical Computer,” and “Set-Intersection Processor (&ohgr;-structure).” Chapter 7: “Introduction,” “Definitions,” “Functional Possibilities of Interconnection Networks,” “Kautz’s Interconnection Network,” “Sorting Networks,” “Sorting in DF-Structures,” “Other Operations of Data Structure Transformation,” and “DF-Structures as Interconnection Networks.” Chapter 8: “Specialization,” “Autonomous Instruments,” “Functional Modules,” “Combined Architectures,” and “VLSI Implementation of DF-Structures.”
Unfortunately, along with the usual typos, there are significant errors and omissions. One typical example is in the presentation of the &agr;-cell, which is designed to support search for maximum values in a two-dimensional homogeneous structure. In the &agr;-cell description, a portion of the logic circuit (as well as the need for and use of it) is not discussed, and there is an error in the definition of one of the two primary functions of the circuit. Similar omissions of discussions or definitions, as well as errors, occur frequently throughout the book. This makes the book tedious at best for the careful reader, who must verify and often correct circuit diagrams, function definitions, and discussions.
The author makes a point of referring to specific parallel processors (STARAN, DAP, MPP, and CM), and states that “these architectures developed from the conceptions of cellular automata, distributed processing, and associative memories, have now taken a leading position in the supercomputer market.” It should be noted that long before the publication of this book, all but one of these machines were out of use, and the final holdout (the Connection Machine, from Thinking Machines Corp.) was having financial difficulties; Thinking Machines has since gone out of business. The history of these machines does not bode well for the future use of the author’s approach to parallel processing.
