Unless otherwise noted benchmarks run for 20 ps (10000 steps * 2 fs), Langevin thermostat, Berendsen barostat.

Lonepeak (intel OpenMpi) March 27, 2015

pmemd.mpi

Atoms1 node2 nodes4 nodes8 nodes
129756.8411.7918.8530.97
155997.1710.1917.2637.36
194344.528.1613.9022.38
294532.985.168.5814.86
323742.754.807.9617.20
330222.474.547.6812.12

Kingspeak (intel mvapich2) March 30, 2015

MPI: /uufs/ember.arches/sys/pkg/mvapich2/std_intel/etc/mvapich2.sh (X)

MPI: /uufs/kingspeak.peaks/sys/pkg/mvapich2/1.9i/etc/mvapich2.sh (O)

Note the performance drop for small systems (< 30k atoms). Using ember-specific mvapich2 appears to help somewhat at higher processor counts. Of course, this could also be because the benchmark is too short to be very accurate. May need to re-run these numbers.

pmemd.mpi