Hprc banner tamu.png

Difference between revisions of "SW:Zebulon"

From TAMU HPRC
Jump to: navigation, search
(Loading the Zebulon modules)
Line 30: Line 30:
 
  #SBATCH --mem=360G                      # Requests 360GB of memory per node
 
  #SBATCH --mem=360G                      # Requests 360GB of memory per node
 
  #SBATCH --error=ZebulonJob.job.e%J      # Sends stderr to ZebulonJob.job.e[jobID]
 
  #SBATCH --error=ZebulonJob.job.e%J      # Sends stderr to ZebulonJob.job.e[jobID]
  #SBATCH --output=ZebulonJob.job.o%J    # Sends stdout to ZebulonJob.job.o[jobID]<br \>
+
  #SBATCH --output=ZebulonJob.job.o%J    # Sends stdout to ZebulonJob.job.o[jobID]<br \>  
cd  $SLURM_SUBMIT_DIR                  # Change to the directory where the job was submitted from<br />
 
 
  ml purge                                                            # purge all module
 
  ml purge                                                            # purge all module
 
  ml ZEBULON/z904                                                    # load the modules for ZEBULON/z904  
 
  ml ZEBULON/z904                                                    # load the modules for ZEBULON/z904  
 
  ml list                                                            # list the modules loaded <br />
 
  ml list                                                            # list the modules loaded <br />
  export OMP_NUM_THREADS = $SLURM_NPROCS                              # set OMP_NUM_THREADS to the number of cores requested above
+
  export OMP_NUM_THREADS = $SLURM_NPROCS                              # set OMP_NUM_THREADS to the number of cores requested above<br />
  Zrun -smp $SLURM_NPROCS $SLURM_SUBMIT_DIR/calcul.inp > $SLURM_SUBMIT_DIR/calcul.out                    # zrun<br />
+
  Zrun -smp $SLURM_NPROCS calcul.inp > calcul.out                    # zrun<br />
 
  exit                                                                # exit when the job is done <br />
 
  exit                                                                # exit when the job is done <br />
  

Revision as of 12:50, 29 June 2021

Zebulon

Description

Zebulon is a non-linear finite element solver of the Z-set suite which addresses the whole range of problems arising in structural mechanics. It is jointly developed by École des Mines ParisTech (France), Onera – the French Aerospace Lab, and NW Numerics & Moldeling, Inc (USA). Zebulon may use multi-threading to accelerate material integration and linear system resolution on a single shared memory machine.

For more information, visit the Zebulon Official Website.

Loading the Zebulon modules

Grace Instructions:

List the versions of Zebulon installed:

mla zebulon

Load a specific ZebulonVersion to setup your environment to run Zebulon:

ml ZebulonVersion

Grace Example Job file

A multicore (48 cores per node) example: (Updated Jun. 5, 2021)

#!/bin/bash
##NECESSARY JOB SPECIFICATIONS
#SBATCH --job-name=ZebulonJob           # Sets the job name to ZebulonJob
#SBATCH --time=2:00:00                  # Sets the runtime limit to 2 hr
#SBATCH --ntasks=48                     # Requests 48 cores
#SBATCH --ntasks-per-node=48            # Requests 48 cores per node
#SBATCH --mem=360G                      # Requests 360GB of memory per node
#SBATCH --error=ZebulonJob.job.e%J      # Sends stderr to ZebulonJob.job.e[jobID]
#SBATCH --output=ZebulonJob.job.o%J     # Sends stdout to ZebulonJob.job.o[jobID]
ml purge # purge all module ml ZEBULON/z904 # load the modules for ZEBULON/z904 ml list # list the modules loaded
export OMP_NUM_THREADS = $SLURM_NPROCS # set OMP_NUM_THREADS to the number of cores requested above
Zrun -smp $SLURM_NPROCS calcul.inp > calcul.out # zrun
exit # exit when the job is done

To submit the job to the queue, use the following command:

[ NetID@Grace ~]$ sbatch jobscript