high performance computing on graphics processing units: hgpu.org

In the last 15 years, Video games have become a dominate form of entertainment. The popularity of video games means children are spending more of their free time play video games. Usually, the time spent on homework or studying is decreased to allow for the extended time spent on video games. In an effort to solve the problem, researchers have begun creating educational video games. Some studies have shown a significant increase in learning ability from video games or other interactive instruction. Educational games can be used in conjunction with formal educational methods to improve the retention among students. To facilitate the creation of games for science education, the RTPS library was created by Ian Johnson to simulate fluid dynamics in real-time. This thesis seeks to extend the RTPS library, to provide more realistic simulations. Rigid body dynamics have been added to the simulation framework. In addition, a two-way coupling between the rigid bodies and fluids have been implemented. Another contribution to the library, was the addition of fluid surface rendering to provide a more realistic looking simulation. Finally, a Qt interface was added to allow for modification of simulation parameters in real-time. In order to perform these simulations in real-time one must have a significant amount of computational power. Though processing power has seen consistent growth for many years, the demands for higher performance desktops grew faster than CPUs could satisfy. In 2006, general purpose graphics processing(GPGPU) was introduced with the CUDA programming language. This new language allowed developers access to an incredible amount of processing power. Some researchers were reporting up to 10 times speed-ups over a CPU. With this power, one can perform simulations on their desktop computers that were previously only feasible on super computers. GPGPU technology is utilized in this thesis to enable real-time simulations.