Free Courses Fluids Structures 3D Design Electronics Materials Photonics Student Teams STEM Python Missions Indo Kursus My Courses

[FAQ]

AUTODYN reads a text file named parallel.cfg which defines compute nodes and cores (tasks) to run in parallel on single or multiple Linux machines. A PBS Pro job script should write compute nodes and cores information from job environment variable PBS_NODEFILE to parallel.cfg file, to start AUTODYN solver in parallel. Below is an example PBS Pro job script for AUTODYN. Run qsub to submit the job. #!/bin/bash # job name #PBS -N autodyn-190 # number of compute nodes and number of CPU cores per node #PBS -l nodes=2:ppn=4 # AUTODYN version REVN=190 # change to job submission directory cd $PBS_O_WORKDIR # create head lines of parallel.cfg echo “#@EPDEF=/apps/ansys_inc/v$REVN/autodyn/bin/linx64” > parallel.cfg echo “#@PPDEF office” >> parallel.cfg echo “#@PPCFG office” >> parallel.cfg # get MachinesCores from PBS_NODEFILE in format of host1:N:host2:N MachinesCores=`uniq -c ${PBS_NODEFILE} | awk ‘{print $2 “:” $1}’ | paste -s -d ‘:’` # print hostname and relative speed in first line, then memory size, number of CPU cores used and number of tasks (usually equal to CPU cores used) in seconds line for each allotted compute node OIFS=$IFS; IFS=”:”; MachinesCoresArray=($MachinesCores); for ((i=0; i<${#MachinesCoresArray[@]}; ++i)); do if [[ $((i % 2)) -eq 0 ]]; then echo "${MachinesCoresArray[i]} sp=1000" >> parallel.cfg echo “#@ mem=100000 cpu=${MachinesCoresArray[i+1]} task=${MachinesCoresArray[i+1]}” >> parallel.cfg fi done FS=$OIFS; # Start AUTODYN solver by specifying input file. It reads parallel.cfg automatically to run in parallel. /apps/ansys_inc/v$REVN/autodyn/bin/autodyn$REVN -I impact-analysis.