Ibliss, gravity is non-renormalizable only perturbatively, i.e. if you expand the metric in "series" about some background metric. At least, this is 't Hooft argument. However, this is not the case if you attempt to quantize it non-perturbatively, like in LQG. In LQG, at hte very least at the kinematical level, since the area and volume are quantized, this provides a natural UV regulator. This is the major difference between lattice field theory and LQG (or lattice LQG if you like it better).