Game Engineering for a Multiprocessor Architecture

Abstract

This paper explores the idea that future game consoles and computers may no longer be single processor units, but instead symmetrical multiprocessor units. If this were to occur games would need to be programmed with concurrency in mind so that they could take advantage of the additional processing units. We explore past research and works in the field of parallel computing to find principles applicable to computer game programming. Concepts such as the Flynn’s classification, task, task-dependency graphs, dependency analysis, and Bernstein’s conditions to concurrency are applied to computer game programming to develop a new model for computer games that is meant to replace the standard sequential game loop.