Biomolecular machines tranaduce between different forms of energy. These machines make directed progress and Increase their speed by consuming free energy, typically in the form of nonequilibrium chemical concentrations. Machine dynathics are often modeled by transitions between a set of discrete metastable conformational states. In general, the free-energy change associated with each transition can increase the forward rate constant, decrease the reverse rate constant, :or both. In contrast to previous optimizations, we find that in general flax is maximized neither by devoting all free-energy changes to increasing forward rate constants nor by solely decreasing reverse fate constants. Instead, the optimal free-energy splitting depends on the detailed dynamics. Extending our analysis to machines with vulnerable states (from which corresponding to in vivo cellular conditions, processivity is maximized by reducing the they can break down), in the strong driving occupation of the vulnerable state.