Monte Carlo simulation is important to get efficiencies for cases where the experimental efficiencies are difficult to get such as for samples with nonstandard geometries and for large samples. In this paper, efficiency of the HPGe detector, routinely used in our lab for a variety of samples, has been computed for point source geometry and its parameters has been optimized to match MCNP and experimental efficiencies within 5% at different sample to detector distances. This optimized geometry was then validated by efficiency transfer to other geometries.