This paper presents significant improvements to the nominal control architecture of the open-access Solo-12 quadruped that were done to implement and compare different centroidal Model Predictive Controllers (MPC). This work was motivated by our previous study in which various MPC schemes of increasing complexity were tested in simulation. They range from a simplified linearized model with contact points fixed by a heuristic, to a nonlinear one that also optimizes the contact points locations. We describe the developments that were necessary to implement such control schemes on the real robot while doubling its maximum velocity. Notably, to synthesize a stable whole-body controller, the inverse dynamics resolution was replaced by a mixed inverse kinematics and quadratic programming scheme. These developments enabled the successful deployments of the various centroidal MPCs on Solo-12. Experimental results show that all MPCs lead to quite similar performances with the proposed whole-body controller and, as a consequence, do not confirm the result of previous simulation study that concluded on the preeminence of the nonlinear centroidal MPC optimizing both the center of mass trajectory and the foot placements.