College of Arts and Sciences
Department of Mathematics
Dr. Donna Calhoun
The GPU performance of the adaptive wave propagation algorithm is critical to its effectiveness in simulating wave propagation in complex media. This algorithm employs adaptive mesh refinement to improve resolution in areas where the wavefield is changing rapidly. The algorithm's performance is significantly improved by the use of graphics processing units (GPUs), which offer faster computation times than traditional central processing units (CPUs). According to the studies in this poster, GPU acceleration of the adaptive wave propagation algorithm provides significant improvements in simulation speed and scalability, as seen in the simulated examples: scalar advection, shallow water equations, euler, and acoustics. When compared to traditional CPU-based algorithms, the algorithm can handle larger models and produce higher resolution results at a faster rate. The algorithm's efficiency and effectiveness are determined by the specific hardware and software configuration of the GPU used; for this study, we used INL Borah.
This project is funded by NASA, 2019, award number 80NSSC20K0495.
Kyanjo, Brian; Calhoun, Donna; Burstedde, C.; Aiton, S.; Snively, J.; and Shih, M., "GPU Accelerated Adaptive Wave Propagation Algorithm" (2023). Research Computing Days 2023. 13.