Acceleration of Complex Terrain Wind Predictions Using Many-Core Computing Hardware

Advances in software and hardware technologies in many-core graphics processing units (GPU) create new opportunities to accelerate computational turn-around time of computational fluid dynamics (CFD) simulations for wind engineering applications, where rapid simulations can make a tremendous impact on the current practice. We investigate the computational performance of a GPU-accelerated incompressible Navier-Stokes solver for 3D flows (GIN3D) for complex terrain simulations on multi-GPU desktop platforms and on the recently deployed NCSA Lincoln Tesla Cluster. Our results demonstrate the potential of GPU computing to accelerate complex terrain wind forecasting applications