This work investigates the reservoir computing capabilities of single-crystal yttrium manganese oxide (YMnO3) with different device configurations. Two crystallite orientations, perpendicular (YMO⊥) and parallel (YMO//) to the c-axis of the YMnO3 crystal, were studied with electrodes on the same side or opposite sides. Electrical measurements revealed pronounced nonlinear characteristics and phase shifts across all configurations. Wave-form generation tasks benchmarking the in-material reservoir computing performance showed the YMO⊥ devices with the same side electrodes achieved the highest accuracy. The results highlight YMnO3 as a promising material for energy-efficient neuromorphic computing platforms exploiting its rich nonlinear dynamics.