Siemens Nastran

Installation

Siemens Nastran is generally installed under /opt/drz/apps/nastran<VERSION>. A modulefile with configuration and MPI settings is placed in /opt/drz/modulefiles/nastran.

An example Nastran modulefile is below:

#%Module 1.0
#
# Siemens Nastran module file
#


set  SCNAS_HOME /opt/drz/apps/nastran
set  SCNAS_VERSION 14.02.012

setenv  SCNAS_HOME $SCNAS_HOME
setenv  SCNAS_VERSION $SCNAS_VERSION

prepend-path  PATH  $SCNAS_HOME/$SCNAS_VERSION/bin

# Set Intel MPI configuration

setenv   DRZ_SCNAS_IMPI_I_MPI_FABRICS  shm:ofi
setenv   DRZ_SCNAS_IMPI_PATH  /opt/intel/oneapi/mpi/latest/bin
setenv   DRZ_SCNAS_IMPI_FI_PROVIDER_PATH /opt/intel/oneapi/mpi/latest/libfabric/lib/prov
setenv   DRZ_SCNAS_IMPI_LD_LIBRARY_PATH /opt/intel/oneapi/mpi/latest/lib

Workflow

In the Nastran component you’ll find the following options:

Run Type: Select either SMP, DMP or DMP+SMP

The options below Run Type will change depending on the selection and you can specify the number of SMP and/or DMP tasks for your job run.

Arguments: Specify any additional options to Nastran. For example, you can try overriding the default smem and memory settings here? (To be verified)

Copy Input to Local Disk: This option allows the user to either use the Network mounted HOME directory as the start point for the Nastran run, or, use the local disk of the compute worker as the start point. Turning on this option (default) will direct HPCBOX to copy the .dat Nastran file from its current location to a temporary run directory on the compute worker. Once the job is complete, the results are copied back to the job’s working directory and placed under a sub-directory which is prefixed with results_. The picture below shows where the results would be placed.

Additionally, since the job is started on the worker’s local disk, intermediate files like .log aren’t available in the job’s working directory during the run. However, it is possible to browse the local directory of the compute-worker using the File Manager within your desktop.

The image below shows a pre-created Bookmark for the one compute worker. Selecting that Bookmark will display the local disk of the worker.

There’s a run and a scratch directory under /mnt on the compute node. The run directory is from where the job gets started and .log etc can be found. Scratch directory is what Nastran uses as scratch.

tip

There will be a sub-directory named with the job-id under run and scratch.