Customizable Embedded Processors: Design Technologies and Applications
by Paolo Ienne and Ranier Leupers (Eds), Published by Morgan Kaufmann (Elsevier)
ISBN 0-12-369526-0, hardback, 497 pp, $59.95, 2007
EN-Genius Reviewer: Dennis L. Feucht

This book is a compilation of contributions from 33 authors scattered about the globe on the approximate topic of where to go with processing in an era of billion-transistor ICs. To some significant extent it provides perspectives on the next level of digital integration beyond the register-transfer level, that of treating processors themselves as building blocks.

Of what possible significance is this topic to analog engineers? Apart from the growing trend of mixed-signal systems, which mix digital computing and analog circuits on a single chip, we are faced with the same questions in the form of analog VLSI. Usually analog is more intensive and digital more extensive. That is analog has fewer components with more complicated behavior and analysis. In digital, the complexity is in the vast numbers of simpler components. Yet the management of IC design complexity is a common problem.

The book title points to a large one, that of standardizing components that can be programmed for specific applications, much like programmable logic devices (PLDs) and field-programmable gate arrays (FPGAs). It has become too costly to design large ASICs, and by customization of existing standard designs, development expense of the standard blocks can be spread over several applications.

Part 2 of the book narrows the scope from overview to specific aspects of processor customization. These include architecture description languages, C compiler retargeting, automated instruction extension and configuring, other topics in architecture extension, design verification, and some architectural details such as data-path synthesis. The final part (3) is case studies, including wireless modem design and "soft" processors for FPGAs.

In discussing such a lofty topic, the book authors cannot afford to delve too deeply into nuts and bolts, though a few figures with logic gates appear, many more with block diagrams, and a few with sequential flow graphs (bubbles and arcs). Concepts are presented in the concrete context of specific projects in industry and the academy.

This is the kind of book analog-circuits people are not likely to be drawn to. Yet it is a book that, I think, has some value in providing us a wider appreciation of how our fellow electronickers in the digital world next door are coping with many of the same problems we have. The forte of the book is in bringing the reader close to these project development problems from a technical standpoint. The analog of digital-to-analog electronics design is revealed and is somewhat intriguing.

If you have a long task list involving analog circuits, you might not put this book at the top of your reading pile. Yet some analog engineers, especially those who value maintaining a wider view of the discipline, could benefit from this book. It is not mechanical or civil engineering and is consequently quite readable by us. The parallel problems in digital design and how they are addressed by computer engineers can provide useful insights that, like stepping-stones, might offer analog engineers involved in large IC design new insight in solving analogous problems.

Postscript: Did you notice the publication date above? That is not in error. I don't know how they publish books in 2007 when it is still 2006! The marketing departments of car companies seem to know how to transcend causality. Apparently the book publishers have learned the secret too!