The system I'm running those on is NWChem 6.6 on a CentOS 7 cluster, one node has two Intel Xeon E5-2650v2, 2.6 GHz, 8 cores processors and 64GB RAM. Those calculations run on 6 or 8 nodes using MPI.
I've now tried using the full direct MP2 method and for now it's running fine, but will still take a while to finish so I don't know yet if that solves the problem.