In this paper, we propose a comparison of the three major dynamic control strategies for a parallel robot, namely the linear control, the nonlinear feedforward control and the Computed Torque Control. These control schemes are first transposed in the Cartesian space. This control space is shown to be perfectly relevant for a parallel robot. Then, the control equations are analysed to exhibit the ordinary differential equation of the controlled error. This allows for obtaining a characteristic of the disturbances on this error. Finally, simulation and experiment are provided to verify the theoretical approach.