Fig. 8From: Accelerating CFD simulation with high order finite difference method on curvilinear coordinates for modern GPU clustersOrganization of CUDA threads for NVIDIA’s GPU and its mapping to hardware structureBack to article page