Customer Reviews

Handbook of Parallel Computing: Models, Algorithms and Applications (Chapman & Hall/CRC Computer & Information Science Series)

5 star:		(0)
4 star:		(1)
3 star:		(0)
2 star:		(1)
1 star:		(0)

 

 

This book is not a handbook in any traditional sense. Unfortunately, it feels more like a collection of articles by people who wanted to publish their own research in an overly pompous context. There is lots of hand-waving in the book, the authors write in the book things like "our initial [and only] experiment shows Y" and "we plan to do X". The book should have been titled something like "Contemporary Issues in Parallel Computing" instead. And the text is sloppy. If you want to read about what those guys the editors invited to contribute were researching in 2008, fine, but if you search for a good account of the best practices, established theories and academic consensus in parallel computing today, do not buy this book. Compared e.g. to the Handbook of Theoretical Computer Science, this book is a joke.

This CRC handbook gives us an impressive summary of the ideas in parallel computing. It describes the numerous hardware architectures that are currently in use. There is a wide variety of ways to bring many CPUs to bear on a large computational problem. The text also explains why parallel computing may be more widely used in future. It appears that the maximum performance of a single CPU is maxing out. Due in no small part to heat problems at gigahertz frequencies of operation.

So for future performance gains, being able to efficiently use several CPUs to decompose a problem might be necessary. For each architecture, you can get some idea of how to do this decomposition. From a reading, it is clear that not every problem will yield to parallel computing. But several classes of problems do.