High-Precision Numerical Simulations of Rotating Black Holes Accelerated by CUDA

Abstract: Hardware accelerators (such as Nvidia's CUDA GPUs) have tremendous promise for computational science, because they can deliver large gains in performance at relatively low cost. In this work, we focus on the use of Nvidia's Tesla GPU for high-precision (double, quadruple and octal precision) numerical simulations in the area of black hole physics -- more specifically, solving a partial-differential-equation using finite-differencing. We describe our approach in detail and present the final performance results as compared with a single-core desktop processor and also the Cell BE. We obtain mixed results -- order-of-magnitude gains in overall performance in some cases and negligible gains in others.