Quasimonoenergetic GeV-scale protons are predicted to be efficiently generated via radiation pressure acceleration (RPA) when the foil thickness is matched with the laser intensity, e.g., L_{mat} of several nm to 100 nm for 10^{19}-10^{22}Wcm^{-2} available in laboratory. However, nonmonoenergetic protons with much lower energies than predicted were usually observed in RPA experiments because of too small foil thickness which cannot support insufficient laser contrast and foil surface roughness. Besides the technical problems, we here find that there is an upper-limit thickness L_{up} derived from the requirement that the laser energy should dominate over the ion source energy in the effective laser-proton interaction zone, and L_{up} is lower than L_{mat} with the intensity below 10^{22}Wcm^{-2}, which causes inefficient or unsteady RPA. As the intensity is enhanced to ≥10^{23}Wcm^{-2} provided by 10-100 PW laser facilities, L_{up} can significantly exceed L_{mat}, and therefore RPA becomes efficient. In this regime, L_{mat} acts as a lower-limit thickness for efficient RPA, so the matching thickness can be extended to a continuous range from L_{mat} to L_{up}; the range can reach micrometers, within which foil thickness is adjustable. This makes RPA steady and meanwhile the above technical problems can be overcome. Particle-in-cell simulation shows that multi-GeV quasimonoenergetic proton beams can be steadily generated and the fluctuation of the energy peaks and the energy conversation efficiency remains stable although the thickness is taken in a larger range with increasing intensity. This work predicts that near future RPA experiments with 10-100 PW facilities will enter a new regime with a large range of usable foil thicknesses that can be adjusted to the interaction conditions for steady acceleration.