This paper investigates some fourth-order accurate explicit finite difference schemes for the 2-D wave equation obtained using 13-, 17-, 21-, and 25-point discrete Laplacians. Optimisation is conducted in order to minimise numerical dispersion and computational costs. New schemes are presented that are more computationally efficient than nine-point explicit schemes at maintaining less than one percent wave speed error up to some critical frequency. Simulation results are presented.

Download

Implicit finite difference schemes for the 3-D wave equation using a 27-point stencil on the cubic grid are presented, for use in room acoustics modelling and artificial reverberation. The system of equations that arises from the implicit formulation is solved using the Jacobi iterative method. Numerical dispersion is analysed and computational efficiency is compared to second-order accurate 27-point explicit schemes. Timing results from GPU implementations demonstrate that the proposed algorithms scale over their explicit counterparts as expected: by a factor of M + 2, where M is a fixed number of Jacobi iterations (eight can be sufficient in single precision). Thus, the accuracy of the approximation can be improved over explicit counterparts with only a linear increase in computational costs, rather than the quartic (in operations) and cubic (in memory) increases incurred when oversampling the grid. These implicit schemes are advantageous in situations where less than 1% dispersion error is desired.

Download

In this paper, a physics-based model for a snare drum will be discussed, along with its finite difference simulation. The interactions between a mallet and the membrane and between the snares and the membrane will be described as perfectly elastic collisions. A novel numerical scheme for the implementation of collisions will be presented, which allows a complete energy analysis for the whole system. Viscothermal losses will be added to the equation for the 3D wave propagation. Results from simulations and sound examples will be presented.

Download

Stencil operations are often a key component when performing acoustics simulations, for which the specific choice of implementation can have a significant effect on both accuracy and computational performance. This paper presents a detailed investigation of computational performance for GPU-based stencil operations in two-step finite difference schemes, using stencils of varying shape and size (ranging from seven to more than 450 points in size). Using an Nvidia K20 GPU, it is found that as the stencil size increases, compute times increase less than that naively expected by considering only the number of computational operations involved, because performance is instead determined by data transfer times throughout the GPU memory architecture. With regards to the effects of stencil shape, performance obtained with stencils that are compact in space is mainly due to efficient use of the read-only data (texture) cache on the K20, and performance obtained with standard high-order stencils is due to increased memory bandwidth usage, compensating for lower cache hit rates. Also in this study, a brief comparison is made with performance results from a related, recent study that used a shared memory approach on a GTX 670 GPU device. It is found that by making efficient use of a GTX 660Ti GPU—whose computational performance is generally lower than that of a GTX 670—similar or better performance to those results can be achieved without the use of shared memory.

Download