The new version is mpt.184.108.40.206 and mainly contains improved performance on the MPI routines. The latency has been much reduced, see below for a more detailed description.
To take advantage of these improvements you need to relink your application.
Besides bug fixes, MPT 220.127.116.11 offers a much reduced MPI latency on T3E systems. The optimization is due to work done at Edinburgh Parallel Computing Centre (EPCC) to use hardware message queues in the critical MPI communication path (the minimum latency decreases from 15 usec to approximately 7 usec on T3E-600 systems.) In addition, MPI_Barrier() performance has been improved for synchronization using MPI_COMM_WORLD.