Classical train simulation is the domain of low Degree-of-Freedom simulators such as Longitudinal Train Simulators. These systems execute quickly, at faster than real-time rates and provide accurate results for a limited application space. For the computationally fast computer reason they are not only used in engineering analysis but have also been integrated into onboard vehicle computers for live train state information. With computer power ever increasing, most recently from the explosion of multi-core computing, the constraints of the past are no longer as restrictive. Through the use of parallel computing, multibody vehicle simulation modelling of wheel-rail contact has been introduced into the train simulation field. Current implementations of this hybridized approach still run significantly slower than real-time. This paper investigates the issues surrounding the use of this simulation methodology in hard real-time systems such as those required for real-time train simulation in onboard vehicle computers and provides some approaches for further computational enhancement. The initial findings demonstrate the benefit of a parallel scheme for multi-body train simulation.