• The order in which the topics are covered is a bit unclear. The text would be clearer if the authors had described all the shared-memory models and standards together, followed by a discussion of standards based on message passing, for example.
• The depth of coverage of each topic is highly variable. If the authors' intent is simply to provide an overview of the most common parallel computing paradigms, it might be desirable to condense many of the chapters.
• The authors illustrate the programming methodologies with concise code segments, but they don't come in a ready-to-use form. The authors mention a Web site on the back cover where readers can download code, but no other references to it can be found anywhere inside the book.
• Many of the problems discussed are fairly standard and already have several popular libraries to solve them, so it might be useful if the authors pointed these out where appropriate. For instance, ScaLapack 3 could be mentioned in the context of solving dense linear systems.
• Given that most of the chapters have code fragments, it would be very useful to be able to actually compile and run full examples via an accompanying Web site as discussed in the previous section. In the same regard, a lot of the presented examples use IP-Fortran, but I wasn't able to download a compiler for the IPlanguages.