high performance computing on graphics processing units: hgpu.org

Wave based simulation methods have been utilized to numerically estimate wave propagation in domains where low-frequency wave effects dominate the response. Finite-difference time-domain (FDTD) methods are increasingly useful for such problems, but they require massive spatial oversampling to increase the bandwidth of the simulation, which leads to significant computational expense. The advantage of explicit time-stepping in FDTD over many other types of methods is that the computing can be carried out in completely data parallel manner. In such cases, graphics processing units can be used to accelerate the simulation substantially. This work presents an open source, 3-D FDTD solver for room acoustics that utilizes Nvidia CUDA to accelerate the simulation. The implementation includes two different frequency independent boundary formulations and several commonly-used source types. The system incorporates an Open GL visualization of the sound field. The solver can be called from Matlab scripts, making it suitable for batch jobs. Examples are presented which validate and demonstrate the capabilities of the solver.