Running QM-MM tutorial on a cluster

From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Sat Nov 17 2018 - 13:30:02 CST

Hi all
Back to namd qm-mm for a project, after a survey of the tutorial one year
ago. At that time Example1 was run completely with mopac and up to the LINE
MINIMIZER BRACKET step 7 with orca (did not take notice of the time). All
that on a desktop with 4 core, on /dev/shm/ with orca.

Now I am trying the tutorial with orca on a cluster with namd2/12 mpi, in
order to establish how the simulations for the project can be best carried
out. On 4 nodes, 144 core (144 tasks, 1 cpu per task) along 1/2 hour the
simulation was still carrying out qm calculations:

TCL: Minimizing for 100 steps
Info: List of ranks running QM simulations: 2.

It is now running from scratch on a single node, 36 core 110GB mem, 24hr
limit. After12hr, the namd log output is as above.
Is there a way to verify the progress done by orca? Should the allowed
24hr be not enough to arrive at step 1 of the qm calculation, or a minimum
is not reached, how to restart the simulation?

Is the scalability of Example1 known? I.e. how many tasks should be best
used?

Grateful for advice.
francesco pietra

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:21:32 CST