Fuck. GAMESS. Use the same damn basis sets as everyone else.
--> Back story:
Lab asked me to learn GAMESS so we can mess with it, which will get you blacklisted (for serious) from our current quantum program Gaussian. GAMESS and Gaussian MP2 energies for a single water-chloride dimer don't match... I find it's the basis sets that differ. GAMESS compensates with simply more primitive functions. Can't just "plug-in" Gaussian numbers because Gaussian = Jerks... so *insert initial statement*
We need GAMESS for further work using a fancy program that hacks... well, GAMESS... to perform some symmetry-adapted perturbation theory calcs. You might imagine the program is called SAPT... and you'd be right.