“…The question, historically, has always been whether a high-order scheme can be made to converge rapidly enough to be competitive with second-order schemes on a computational cost basis. While there are some results that suggest this is the case for structured meshes [41,21,46,14], the situation for unstructured meshes is less clear, because little work has been done on optimizing convergence rates for high-order unstructured mesh solvers. The main contribution of much of our recent research [30,31,29,33] has been to fill this gap by developing an efficient implicit time advance scheme for our high-order accurate solver.…”