Overview

The computer industry is undergoing, if not another revolution, certainly a vigorous shaking-up. The major chip manufacturers have, for the time being at least, given up trying to make processors run faster. Moore's Lawhas not been repealed: each year, more and more transistors fit into the same space, but their clock speed cannot be increased without overheating. Instead, manufacturers are turning to "multicore" architectures, in which multiple processors (cores) communicatedirectly through shared hardware caches. Multiprocessor chips make computing more effective by exploiting parallelism: harnessing multiple processors to work on a single task.

The spread of multiprocessor architectures will have a pervasive effect on how we develop software. Until recently, advances in technology meant advances in clock speed, sosoftware would effectively "speed up" by itself over time. Now, however, thisfree ride is over. Advances in technology will mean increased parallelism and not increased clock speed, and exploiting such parallelism is one of the outstanding challenges of modern Computer Science.

This book focuses on how to program multiprocessors that communicate via a shared memory. Such systems are often called shared-memory multiprocessors or, more recently, multicores. Programming challenges arise at all scales of multiprocessor systems at a very small scale, processors within a single chip need to coordinate accessto a shared memory location, and on a large scale, processors in a supercomputer need to coordinate the routing of data. Multiprocessor programming is challenging because modern computer systems are inherently asynchronous: activities can be halted or delayed without warning by...