Customer Reviews

Parallel Programming: for Multicore and Cluster Systems

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

 

 

This book is great academic quality survey of modern parallel programming. It starts with a brief and yet thorough overview of architecture and recent innovations of the multicore processors. To name a few, the book picks a glance at Intel, AMD, Sun T-series, and IBM Power processors. Followed by that, a short introduction into performance analysis as well as parallel design patterns are presented. Numerous references (often academic journals) are also provided throughout the book.

The book then quickly reviews MPI-1 and MPI-2, which are the most popular Distributed Memory models. I'd taken courses in parallel programming back in the graduate school and used both MPI and OpenMP; however this brief chapter on message passing was simply incredible. Shared memory using POSIX threads and a hint at how to implement different parallel patterns (such as consumer-producer, pipelining, thread pools, and task pools) are followed. The book goes further and addresses Java threads and OpenMP. The final chapter is algorithms for linear systems and works out a few examples like Gaussian elimination...

I found it an invaluable up to date book. Yet it's a survey of parallel programming, aiming at the more experienced programmers, either coming from industry or scientific computing world.