The DIMPLE S06 critical benchmark has been calculated using a direct modeling and a two-step modeling approaches. A direct modeling is performed with MCNP6, SERPENT2, CASM04E, and TRITON/NEWT. A two-step modeling is performed by using the nodal code PARCS3.0 employing the homogenized two-group cross section which are generated through SERPENT2, CASM04E and TRITON/NEWT. Detailed calculation models are developed in this paper and the parameters for a two-step modeling generated by each code are compared to each other. Finally, the eigenvalue for DIMPLE S06 are compared to each a direct and a two-step modeling. In this paper, the calculation result of two-step modeling agrees well with that of direct modeling. The result shows that the use of the assembly discontinuity factor in two-step calculation is essential to improve the relative error to the result of a direct modeling. A necessity of the assembly discontinuity factor is clearly proved by calculating the flux distribution for the homogenized assembly.