Using overset grid can effectively handle problems with large-scale motions.However,as the scale of the grid and the complexity of the flow problem increase,traditional CPU-based overset grid assembly methods are becoming increasingly difficult to meet current computational demands.To address these issues,this paper develops a GPU-based k-d tree nested grid assembly method based on the CUDA platform,and optimizes the k-d tree construction process and search process,which greatly improves the contribution unit search efficiency and object-surface distance calculation efficiency,thereby accelerating the nested grid assembly speed.
graphics processornested gridk-d treeassembly methodflow field computing domain
万方数据
维普
