Without hydrodynamics, nothing fundamental sets the mass scale. If you assume that c = G = 1, then there is a rescaling freedom for M, the black hole mass. Apart from this, it is up to the designer of the parameter file what they call M. I think that the ADM mass is a good choice, but others might use a different convention. In practice, one examines the black hole masses and the ADM mass in a simulation, which is then returned in terms of the grid spacing, which is related to the time step size, thus you see what M is used in this particular case.

For example, in <http://einsteintoolkit.org/gallery/bbh/index.html> I assume that M_ADM = 1, but I don't know whether this is taken from the initial conditions or whether this is measured after the initial junk radiation has left.

I hope my answer isn't too trivial.

-erik