Jon HasselgrenLund University
Tomas Akenine-MöllerLund University
Lennart OhlssonLund University
Over the past few years, general-purpose computation using GPUs has received much attention in the research community. The stream-based rasterization architecture provides for much faster performance growth than that of CPUs, and therein lies the attraction in implementing an algorithm on the GPU: if not now, then at some point in time, your algorithm is likely to run faster on the GPU than on a CPU.
However, some algorithms that discretize their continuous problem domain do not return exact results when the GPU’s standard rasterization is used. Examples include algorithms for collision detection (Myzskowski et al. ...