As a graduate student learning about parallel processing, I continually came across papers on the Cm* system, developed at Carnegie-Mellon University. The Cm* system was designed in 1975, and by 1979 a 50-processor system was fully operational. During the course of the project, a wealth of systems software was developed, a number of parallel languages and environments were designed, and numerous performance-oriented experiments were conducted. It is hard to overemphasize the value of the contribution made by this project to the understanding of issues in parallel processing.

Parallel processing: the Cm* experience extensively covers the results of the Cm* project in 12 chapters and four appendices. Operating systems, programming environments, and algorithm performance are the three major topics covered. The sections on operating systems give fairly detailed descriptions of the operating system mechanisms and their rationale. Likewise, the sections on languages and environments define syntax and illustrate constructs with some examples (although more examples would be beneficial). The sections on algorithm performance describe numerous parallel algorithms and describe measured and predicted performance by plotting speedup as a function of number of processors. The algorithms covered tend to be quite simple (e.g., Quicksort, Jacobi iteration, railway network simulations) but do focus on specific performance problems facing designers of parallel algorithms.

The point of parallel processing is increased performance, a fact the Cm* researchers never lost sight of. Performance considerations and tradeoffs rightly permeate the discussions in this book. I recommend it as a practical introduction to performance issues in parallel processing and anticipate its use in graduate courses on parallel processing and computer architecture.

I find it quite valuable to have the highlights (and original references) of such an extensive project coalesced into one book. It should serve as a model for organizing the results of other, large research efforts. One can take issue with many of the ideas explored by Cm*; the conclusions expressed by the authors of this book are not the last word on the subject. But Cm* should be viewed as a pioneering effort in parallel processing, and this book effectively chronicles that effort.