5.6.2. Executing Runs Using the Command Line Interface

This section describes how to run a prepared job (whether that job was prepared from the Forte User Interface or Command Line Interface) from the command line

If you want to run the job on a different compute resource to the one on which you prepared the job by the Forte User Interface or Command Line Interface, see section Running On a Remote Server.

Once you are ready to run the job, cd to the run directory and run the job using

Windows

call run_mpi.bat

Linux

run_mpi.sh

and then monitor the progress of your run (see Monitoring Runs for details). These scripts will exit when the run finishes.

5.6.2.1. Running On a Remote Server

Sometimes you may wish to run the job on a different server or location than where you prepared the job, for example:

  • Prepare the job using the Forte User Interface and then run on a different compute resource.

  • Prepare the job using the Forte Command Line Interface and then run on a different compute resource.

  • Prepare the job using the Forte User Interface or Forte Command Line Interface on one disk and then run on a different disk.

Since Ansys Forte generates both Windows and Linux job scripts, it is very easy to prepare your jobs on your local computer and then transfer them to a remote computer for execution. The line endings for the scripts are generated correctly for their target platforms; but individually FTP’ing those files could result in undesired line-ending conversions, so some care transferring files is needed to ensure that line endings are preserved appropriately.


Tip:  Ansys recommends that you compress the entire run directories into a single zip or tarball and then transfer that to the remote system.

You should transfer the entire prepared run folder and its files to the remote machine.


The following two sections (Distributed Parallel on Linux Using a Hostfile and Distributed Parallel Using a Queuing System ) describe how to run your job on a Linux distributed parallel cluster (for this case, you may need to transfer the run folder from the machine it was prepared on to a machine on the distributed cluster).