GPU-Accelerated WENO Schemes for DNS of Compressible Turbulent Flows

This paper explores strategies to transform an existing CPU-based high-performance computational fluid dynamics (CFD) solver, HyPar, for compressible flow simulations on emerging exascale heterogeneous, CPU+GPU, computing platforms. The scientific motivation for developing a GPU enhanced version of HyPar is to simulate canonical flows, such as homogeneous isotropic turbulence (HIT) of compressible flow in a triply periodic box. We show that optimizing memory operations and thread blocks results in a code that is more than 200x faster on GPUs than on CPUs. Using multiple GPUs and MPI communication, we demonstrate both strong and weak scaling of our GPU-based HyPar implementation on the Summit supercomputer at Oak Ridge National Laboratory.