Impulse-based dynamic simulation on the GPU

In this paper a new, efficient method for dynamic simulation on the GPU is presented. The method is based on an impulse-based approach which is an ideal candidate to simulate on limited hardware due to its simplicity. The proposed method shows how the impulse-based dynamic simulation can benefit from the highly parallel structure of the GPU without suffering too much losses by its limitations. This is achieved by the use of a new way to solve constraints. Most parts of the actual computation can be done in parallel, using only a few number of operations. This allows the implementation to run on a wide range of graphics boards.