Reflectometry is foreseen in the coming generation of tokamaks such as DTT, ITER and DEMO, creating a need to predict the behavior and capabilities of these new reflectometry systems through the used of synthetic diagnostics. The FDTD time-dependent codes use to implement synthetic diagnostics are computational demanding, the reason why 2 dimensional codes (as REFMUL or REFMULF) are used. REFMUL3, a newly developed performing parallel code gives access to 3D simulations, although at a much higher cost than the 2D ones. With this work we begin a benchmark e ort to assess the main di erences and compromises done when using 2D versus 3D.