My understanding was that renormalization only needs "parameters" when the computational methods are approximate.

I thought the issue was more that we have no idea how to perform appropriate calculations for SM+QG? Maybe I'm wrong?