With the advent of parallel processing, the raw computing power available to solve many problems is now immense, well beyond what was available using conventional uniprocessor computing techniques. This is especially significant in solving the so-called Grand Challenge problems of computing. A Grand Challenge problem (such as accurate weather forecasting) is a fundamental problem of science or engineering, having a broad social or economic impact, whose solution requires the use of very high-end computing power.

Many problem domains are also naturally suited to parallel processing. For instance, applications such as weather prediction, biosphere modeling, population analyses, and the like are modeled by assuming a grid on the domain being modeled. The segments (squares or cubes) of the grid are analyzed for their influences on surrounding segments, and vice versa. This often requires solutions to large systems of differential equations, with the granularity of the grid chosen...