Recent advances in manufacturing technologies, such as additive manufacturing (AM), have raised the potential of choosing surface finish pattern as a design parameter. Hence, understanding and prediction of aerothermal effects of machined microstructures (machined roughness) would be of great interest. So far, however, roughness has been largely considered as a stochastic attribute and empirically modeled. A relevant question is: if and how would shape of the machined roughness elements matter at such fine scales? In this paper, a systematic computational study has been carried out on the aerothermal impact of some discrete microstructures. Two shapes of configurations are considered: hemispherical and rectangular elements for a Reynolds number range typical for such structures (Re < 5000). Several validation cases are studied as well as the turbulence modeling and grid sensitivities are examined to ensure the consistency of the results. Furthermore, large eddy simulation (LES) analyses are performed to contrast the behavior in a well-established turbulent to a transitional flow regime. The results reveal a distinctive common flow pattern change (from an “open separation” to a “reattached separation”) associated with a drastic change of drag correlation from a low to a high loss regime. The results indicate a clear dependence of drag and heat transfer characteristics on the element pattern and orientation relative to the flow. The distinctive performance correlations with Reynolds number can be affected considerably by the element shape, for both a transitional and a turbulent flow regime. The results also consistently illustrate that conventional empirical stochastic roughness parameters would be unable to predict these trends.