DL_MONTE

Overview

DL_MONTE provides Monte Carlo simulation of condensed phases, via DL_POLY.

The primary objective of DL_Monte is to provide an alternate methodology for users of the
DL_POLY program. As far as possible the input and commands are the same as those used in
DL_POLY.

At present DL_MONTE is capable of Metropolis sampling only.

Version 1.00 is installed and it was compiled with Intel Fortran 12.0.5 and openmpi 1.6.

Restrictions on use

Whilst the software is free for academic usage there are limitations within the license agreement which must be strictly adhered to by users. All users who wish to use the software must request access to the ‘dlmonte’ unix group. A copy of the full license is available on the CSF in $DLMONTE_HOME/DL_MONTE_licence.pdf. Important points to note are:

• No industrially-funded work must be undertaken using the software. See clauses 2.1.3 and 2.2 of the license.
• The software is only available to Staff and Students of the University of Manchester. Users are reminded that they must not share their password with anyone, or allow anyone else to utlise their account.
• Citation of the software must appear in any published work. See clause 4.2 for the required text.

There is no access to the source code on the CSF.

Set up procedure

DL_MONTE cannot be run in serial (i.e. on a single core).

To run DL_MONTE on 2 to 24 cores – a single compute node, you need to load the DL_MONTE modulefile:

module load apps/intel-12.0.5/dl_monte/1.00

If you wish to run jobs of 48 cores or more (using multuples of 24 cores) then please load the DL_MONTE infiniband modulefile:

module load apps/intel-12.0.5/dl_monte/1.00-ib

Note that the modulefiles load some Intel compilers and MPI. When you unload the DL_MONTE modulefile it will also unload the others, and these may be required by other applications you are using from the same login session.

Running the application

DL_MONTE is an MPI application – there is no serial version.

Parallel batch job submission – non-Infinband, 2 to 24 cores

Make sure you have the correct modulefile loaded. Change to the directory where you have the required input files, and where the application will run.

You can then write a batch submission script eg

#!/bin/bash
#$ -cwd
#$ -V
#$ -pe smp.pe 2                    # Number of cores (max is 24)

## $NSLOTS is automatically set to the number of cores given above

mpiexec -np $NSLOTS DLMONTE.X

and submit it to the batch queues using qsub batchscript (replacing batchscript with the name of your file).

You should experiment with the number of cores required to determine the optimum number for your data set. Simply requesting more cores may not may the code run any faster but may increase your wait time in the batch queue.

Parallel batch job submission – Infinband, 48+ cores

Large parallel jobs can be run by requesting multiple 24-core nodes. The minimum number of nodes is 2 (i.e., 48 cores). The number of cores requested must be a multiple of 24.

Make sure you have the InfiniBand version of the modulefile loaded (see above)

Change to the directory where you have the required input files e.g. CONTROL, FIELD, and where the application will run.

You can then write a batch submission script eg:

#!/bin/bash
#$ -cwd
#$ -V
#$ -pe orte-24-ib 48             # Minimum is 48 cores, must be in multiples of 24.

## $NSLOTS is automatically set to the number of cores given above

mpiexec -np $NSLOTS DLMONTE.X

Jobs in the orte-24-ib.pe must be 48 cores or more and a multiple of 24.

Further Info

• User doc in $DLMONTE_HOME/doc/DL_Monte_user_guide_v1.pdf which can be read using evince.
• DL_MONTE website

Updates

Version 1.00 installed 07 Feb 2014.

Access to unknown version removed 07 Feb 2014.