Difference between revisions of "Palm.sge"
From HPC users
Jump to navigationJump to search
Albensoeder (talk | contribs) |
Albensoeder (talk | contribs) |
||
(One intermediate revision by the same user not shown) | |||
Line 43: | Line 43: | ||
#$ -l h_rt=0:10:0 | #$ -l h_rt=0:10:0 | ||
### memory per job slot (maximum: | ### memory per job slot (maximum: 1850M for low-memory nodes, ### | ||
### | ### 3850M for high-memory nodes): ### | ||
### Usuallay on FLOW you can use per | ### Usuallay on FLOW you can use per default 1850M because the ### | ||
### the node will be used exclusively by the job ### | ### the node will be used exclusively by the job ### | ||
#$ -l h_vmem= | #$ -l h_vmem=1850M | ||
### disk space - default: 50 GB (if too low, job aborts!) ### | ### disk space - default: 50 GB (if too low, job aborts!) ### | ||
Line 62: | Line 61: | ||
### which parallel environment to use, and number of slots ### | ### which parallel environment to use, and number of slots ### | ||
#$ -pe | #$ -pe impi 8 | ||
### enable resource reservation (to prevent starving of parallel jobs) ### | ### enable resource reservation (to prevent starving of parallel jobs) ### | ||
Line 79: | Line 78: | ||
cat $PE_HOSTFILE | cat $PE_HOSTFILE | ||
echo | echo | ||
### call script which starts palm ### | ### call script which starts palm ### | ||
Line 94: | Line 84: | ||
exit | exit | ||
</nowiki> | </nowiki> | ||
[[PALM | back to PALM page]] | [[PALM | (back to PALM page)]] |
Latest revision as of 11:50, 12 February 2015
#!/bin/bash ############################################################################## # Sun Grid Engine (SGE) Batch Job Script # # Example for submitting a PALM job to FLOW # # (BW, Unversity of Oldenburg, January 2013) # ############################################################################## ############################################################################## # SGE directives (options to the qsub command) # ############################################################################## ### NOTE: All lines containing SGE directives have to start with "#$" ### ################### # General options # ################### ### Write stdout and stderr in only one file: ### #$ -j y ### PALM requires the Korn-shell: ### #$ -S /bin/ksh ### name of the job ### #$ -N palm ### change to directory where job was submitted from: ### #$ -cwd ### extend PATH-variable to use PALM scripts ### ### and set environment variable PALM_BIN ### export PALM_BIN=$HOME/palm/current_version/trunk/SCRIPTS export PATH=$PALM_BIN:$PATH #################################### # Resource requirements of the job # #################################### ### maximum walltime of the job (hh:mm:ss): ### #$ -l h_rt=0:10:0 ### memory per job slot (maximum: 1850M for low-memory nodes, ### ### 3850M for high-memory nodes): ### ### Usuallay on FLOW you can use per default 1850M because the ### ### the node will be used exclusively by the job ### #$ -l h_vmem=1850M ### disk space - default: 50 GB (if too low, job aborts!) ### ### if job generates data larger than 50G --> increase value ### #$ -l h_fsize=50G ###################################### # Specific options for parallel jobs # ###################################### ### which parallel environment to use, and number of slots ### #$ -pe impi 8 ### enable resource reservation (to prevent starving of parallel jobs) ### #$ -R y ############################################################################## # Commands to be executed by the job # ############################################################################## echo `date` echo ### it's useful (and nice) to see on which machines the job is running ### echo "Contents of PE_HOSTFILE:" cat $PE_HOSTFILE echo ### call script which starts palm ### mrun -z -d example_cbl -h lcflow -K parallel -X 8 -t 600 -r "d3# pr#" exit