It shows that the parallel implementation of the program is successive, a speedup 6.33 performed on 8 nodes was obtained, and a very large MRCI calculation with 5.8×10^8 CSF has been carried out on 12 nodes.