Developing a theory for the long time dynamics of polypeptides requires not only a proper choice of the relevant dynamic variables, but also a meaningful definition of friction coefficients for the individual atoms or groups of atoms in the reduced system. We test various aspects of the optimized Rouse-Zimm model for describing the long time rotational dynamics of a peptide fragment. The necessary equilibrium input information is constructed from a 1 ns molecular dynamics simulation for the solvated peptide by using a parallel Gray version of CHARMm, whose new features are described here. The simulations also provide time autocorrelation functions for comparisons with both theoretical predictions and with experiment. Two atomic friction models (van der Waals radii and accessible surface area) are chosen, and tests are made of the applicability of two combining rules for calculating the group friction coefficients. While the rotational dynamics of the peptide is insensitive to the friction models used, the combining rules are found to impact profoundly upon the theoretical descriptions for the behavior of the peptide dynamics for the reduced descriptions with fewer variables. The calculations study the role of the memory functions, neglected in this dynamical theory, and the interatomic hydrodynamic interactions in constructing the group friction coefficients. While the I ns trajectory is longer than customarily used for very complex systems, there are nontrivial influences of the duration of the molecular dynamics trajectory on the description of the dynamics.