EFFICIENT GPU PARALLELIZATION OF ADAPTIVE MESH REFINEMENT TECHNIQUE FOR HIGH-ORDER COMPRESSIBLE SOLVER WITH IMMERSED BOUNDARY

An efficient GPU parallelization of a compressible flow solver is presented: the platform supports dynamic Adaptive Mesh Refinement (AMR) to improve the efficiency of the solution of Navier-Stokes equations by means of high-order Weighted Essentially Non Oscillatory (WENO) reconstruction schemes. Moreover, an accurate Immersed Boundary (IB) method is used to represent complex boundary conditions. The code structure has been designed to support a number of computational backends (e.g., CPU, NVIDIA GPU) while keeping optimal parallel performances and good readability. The results in terms of solution accuracy and execution performances are encouraging. In particular, the results of a gas dynamic supersonic test case are presented as validation.