high performance computing on graphics processing units: hgpu.org

Nowadays, physics simulation is a relevant topic in several domains, from scientific areas like medicine to entertainment purposes such as movie’s effects, computer animation and games. To make easier the production of faster simulations, developers are using physics engines because they provide a variety of features like rigid and deformable body simulation, fluids dynamics and collision detection. Computer game and film industries use increasingly more physics engines in order to introduce realism in their products. In these areas, speed is more important than accuracy and efforts have been made to achieve high performance simulations. Besides faster physical simulation algorithms, GPUs’ performance improvement in the past few years have lead developers to transfers heavy calculation work to these devices instead of doing it in the Central Processing Unit (CPU). Some engines already provide GPU implementations of several key features, particularly on rigid body collision detection. In this work we want to accelerate a feature present in most of the current physics engines: cloth simulation. Since collision detection is one of the major bottlenecks in this kind of simulation,we will focus specifically in improving this phase. To achieve a considerably speed-up we plan to exploit the massive parallelism of the Graphics Processing Unit (GPU) by designing an efficient algorithm using the Open Computing Language (OpenCL) framework. Finally,a study will be made to compare the performance of a sequential CPU approach against the parallel GPU proposed solution.