In this paper we present a turbomachinery density-based CFD solver optimized for CPUs as well as GPUs, which accounts for complex thermodynamics including non-equilibrium condensation and two-phase flow, making extensive use of tabulation techniques. The two-phase flow is treated by means of the mono-dispersed Source-Term Euler-Euler model. The non-equilibrium wet-steam model is validated in classical nozzle test cases and its application in turbomachinery configuration is demonstrated in a well-documented steam turbine cascade in the context of classic RANS modeling. Finally, the LES-solver performance and scalability, together with its accuracy, are assessed and discussed on the basis of the well-known and theoretically relevant experiment by Comte-Bellot and Corrsin. For both, standard RANS computations, where an upwind schemes has been adopted, as well as for the LES computations, where a central scheme in skew-symmetric form has been employed, the solver shows remarkable computational speed and accuracy for non-ideal gas applications, rendering it suitable for more complex LES computations in steam turbine flows.