This report presents numerically predicted Radar Cross Section (RCS) results for the ElectroMagnetic Code Consortium (EMCC) cone-sphere benchmark. The variants with and without an electrically small circumferential square groove are compared. RCS predicted with the Multi-Level Fast Multipole Algorithm (MLFMA) is compared to RCS computed with a high frequency asymptotic Geometric Optics assisted Physical Optics method (GO+PO), and alternatively, with a Physical Theory of Diffraction assisted Physical Optics method (PTD+PO). The groove in the EMCC-benchmark is used as a stand-in model for different types of gaps present on many aeronautic vehicles. The investigation shows that the presence of the small groove on the cone sphere has a tremendous influence on the RCS. It also corroborates the known fact that it may act as a separator (cf. waterbreak) changing the surface current fields and consequently the scattering. The presence of the groove is not captured satisfactory by the asymptotic high frequency methods. The MLFMA results are shown to be convergent using mesh refinement. Their robustness is investigated with respect to variation in the linear blending parameter of the Combined Field Integral Equation (CFIE) approach, and their sensitivity with respect to the residual tolerance for the interative equation solution is studied and presented. Mitigation of the negative influence of gaps and grooves on the radar cross section is of great interest for low observable vehicles.