Templates for simplified submission

The submission script is a self-executable template; each scheduler gets its own version of the template. The user is guided to fill in the TO_BE_DEFINED fields.

The templates can be found at the following links:

• Slurm template ($AME/scripting/python/hpc/templates/SLURM.template.sh)

• OpenLava/LSF template ($AME/scripting/python/hpc/templates/LSF.template.sh)

• PBS template ($AME/scripting/python/hpc/templates/PBS.template.sh)

• UGE template ($AME/scripting/python/hpc/templates/UGE.template.sh)

• SGE template ($AME/scripting/python/hpc/templates/SGE.template.sh)

The file is composed of 3 parts:

1. Description of the study: model and simulation settings

2. Scheduler settings: request resources, accounting, ...

3. internal section to apply the job submission. You should not edit this part.

In the following part, specific settings of each part are described in details.

Simulation and model settings

• MODEL is the name of .ame or .pck file containing your model.

• DIRMODEL is the directory where the model is stored.

• SIMTYPE corresponds to the type of the Simcenter Amesim study: single run, batch run or design exploration.

• Note that for a design exploration study, the sub type and the name declared in the Study Manager dialog box has to be supplied respectively in STUDYTYPE and STUDYNAME Figure 25: Study Manager dialog box

• The working directory WDIR will be used during job execution, this directory must be accessible to all nodes involved in the computation. Please consult your HPC platform IT for more details.

• COMPILER defines the compiler: select  gcc64 or intel64. Please consult the HPC platform IT for more details.

• The option --unique_sub_dir creates a unique subdirectory to replace the working directory. Example with --unique-sub_dir activated: the WDIR defined is CopyWDIR=/home/user/hpcdemo/wdir The directory created and used during the study would be Copy/home/user/hpcdemo/wdir/042625a/

Model compilation bypass

It is possible to avoid compilation of the model with --no_compilation, considering the input model and embedded libraries are already compiled for the chosen compiler.

Saving strategy options

The extra options --save_as defines where to save the study with results at the end of the study. It is possible to declare a path or a new model name.

• A canonical path using the same initial system name Copy$HOME/my_result_dir/

• A canonical path to a new ame file Copy$HOME/my_result_dir/my_simulated_system.ame

The option --no_save deactivates model saving, the model remains exploded in the working directory at the end of simulation.

X server

The option --xserver defines the virtual X server environment command.

Pre and post-processing scripts

Pre and post-processing scripts can be executed before the compilation, between compilation and simulation, and after the simulation. Such scripts must be Python scripts and can be zipped with the other necessary files.

Note

Pre and Post processing Python scripts are executed remotely within the AMEPython context. No extra Python modules or external dependencies are supported unless they are available in the Simcenter Amesim setup directory or model.

If the script is included in an archive, the script path must be relative to the archive root.

• Path to the archive containing all the pre-compilation pre-processing material (must contain the script itself): --precomp_preproc_archive=my_precompile_preproc_archive.zip

• Path to the pre-compilation pre-processing script: --precomp_preproc_script=my_precompile_preproc_script.py

• Path to the archive containing all the post-compilation pre-processing material (must contain the script itself): --postcomp_preproc_archive=my_postcompile_preproc_archive.zip

• Path to the post-compilation pre-processing script: --postcomp_preproc_script=my_postcompile_preproc_script.py

• Path to the archive containing all the post-processing material (must contain the script itself): --postproc_archive=my_postproc_archive.zip

• Path to the post-processing script: --postproc_script=my_postproc_script.py

Combining options

The field OTHER_OPTS is null by default, but it is possible to declare several options in a row as follows:

Combining additional options:

Copy

OTHER_OPT="" 
OTHER_OPT=" --no_compilation --xserver=xvfb-run"
OTHER_OPT=" --save_as=/tmp/model_out.ame --no_compilation"

Field and options	Role
MODEL	.ame or .pck file
DIRMODEL	input directory
SIMTYPE	simulation type ("single", "batch" or "DE")
STUDYTYPE	subtype for DE , select "Doe", "Optim" or "DE"
STUDYNAME	name of the design exploration
WDIR	working directory
COMPILER	"gcc64" or "intel64"
PORT	(optional, leave blank if not used) port used for STDSIMManager
--unique_sub_dir	create unique working directory
--no_compilation	disable model and libraries compilation
--save_as=	define the saved model with simulation results
--no_save	disable autosave
--precomp_preproc_archive=	pre-compilation pre-processing archive
--precomp_preproc_script=	pre-compilation pre-processing script
--postcomp_preproc_archive=	post-compilation pre-processing archive
--postcomp_preproc_script=	post-compilation pre-processing script
--postproc_archive=	post-processing archive
--postproc_script=	post-processing script

Scheduler settings

The batch command options must be defined in the AME_SCHED_ARGS environment variable: to define the job name, the partition and the resource requirements (nodes, memory, ...).

The options are specific to all job schedulers supported and they depend on the local HPC platform usage and policy.

For more details, visit the job scheduler website:

Topic	Link
Slurm
Slurm batch command
LSF
LSF batch command
PBS
PBS batch command
UGE
UGE batch command
SGE
SGE batch command