Traditionally, high performance computing was carried out by specialized hardware and software working together to produce a highly optimized environment. Grids are beginning to change that, by removing the need for the specialized elements while enabling the spreading of the work across a wider range of hardware, software, and networks. A new article takes a closer look at the two solutions, and how they are converging.From the article introduction:
Parallel computing is a methodology for designing applications designed to work within large parallel computing environments. In a grid system, the execution of an operation or equation is geared toward providing a structure for executing tasks in parallel, where the order in which the work units are executed and received is not sequential and not reliant on previous code iterations. While, overall, the models of these two systems — parallel computing and grids — are converging, the traditional programming models of these two systems remain different. In this article, we will look at parallel computing, grids, and their convergence.