I am performing several benchmarks with Intel MPI and SGI MPT. Intel MPI has been 1.2 times faster them MPT. However, i have three specifications who is need to follow:
1 - Use the faster MPI - is this case is Intel MPI
2 - Prevent the user to access to the processing node through ssh
3 - Use Torque queue system
With MPT i was able to prevent the SSH access and use torque queue system, but, MPT is not the faster MPI.
Have Intel MPI any way to integrate with torque queue system as to a be abble to block SSH access (like OpenMPI has --with-tm ) ?
Thanks for getting in touch and glad to hear Intel MPI is performing well for you! Which version of Intel MPI Library are you using? Our latest is Intel MPI 2017 and I would encourage you to update to it.
Intel MPI has supported tight integration with Torque for a couple of releases. If you're using our 'mpirun' command, it should automatically pick it up. If you really want to make sure, you can run with the "-bootstrap pbsdsh" flag for either mpirun or mpiexec.hydra:
$ mpiexec.hydra -bootstrap pbsdsh -n 12 ...
This will use Torque's pbsdsh command instead of ssh to do the shell connectivity.
We have more details on our bootstrap option in the online Reference Manual.
Hope this helps. Let me know how it goes.