Focus: Multiparadigm Programming Focus: Multiparadigm Programming Focus: Multiparadigm Programming

THE COMPUTER INDUSTRY IS EXPERIENCING a major shift: improved single-processor performance via higher clock rates has reached its technical limits due to overheating. Fortunately, Moore’s law still holds, so chip makers use transistors to boost performance through parallelism. Modern chips consist of multiple microprocessors (also called cores), buses, and cache memory on the same chip. As of this writing, desktop processors already have up to six cores, but this number will likely increase. Server processors such as Intel’s Single Chip Cloud Computer demonstrate that it’s possible to integrate 48 generalpurpose processors on just 567 mm2. The record, however, is held by manycore graphics cards with processors that have hundreds of specialized cores, simultaneously executing tens of thousands of hardware threads. In fact, multicore general-purpose processors and manycore graphic processors are the de facto standard for every modern desktop or laptop computer. Exploiting the full hardware potential of these processors requires parallel programming. Thus, a large number of developers need to parallelize desktop applications, including browsers, business applications, media processing, and other domain-specifi c applications. This is likely to result in the largest rewrite of software in the history of the desktop. To be successful, systematic engineering principles must be applied to parallelize these applications and environments. In light of these developments, we’re pleased to present this special issue on programming methods, tools, and libraries for parallelizing desktop applications.