Fig. 14From: Accelerating CFD simulation with high order finite difference method on curvilinear coordinates for modern GPU clustersPerformance of main parts of the solver. Generally, the parts that contain large amounts of computation get apparent acceleration. In the sections of time advance and viscous flux, due to the high occupancies of kernels, both GPUs get higher speedups. For evaluation of time step and inviscid flux, Titan V gets higher speedups than RTX 2080 Ti, because Titan V has more double precision operation processing units, which means higher double precision operation performance, see Table 1Back to article page