Classical calculators cannot be used for electronic structure, they only provide forces (and stress, sometimes) for geometry optimization. The Brenner potential in ATK-Classical can be used to very quickly - in a matter of seconds, in fact - optimize the structure of even a large, complicated graphene or other hydrocarbon structure.
GPAW can in theory be used, but is probably too time-consuming unless you have a large cluster.
As Nordland wrote, experiences with the Huckel method (provided the Cerda basis set is used) show excellent results for all shapes of graphene, but do note that if you want to study specific things like adsorption of molecules on the graphene, you may need to resort to DFT for a proper description of the chemical interactions.
