Difference between revisions of "Complex submissions"

From IBERS Bioinformatics and HPC Wiki
Jump to: navigation, search
Line 10: Line 10:
 
#run on a node that has 20GB free, specify that the job will be killed if it uses over 40G or total memory or run for more than 24hours.
 
#run on a node that has 20GB free, specify that the job will be killed if it uses over 40G or total memory or run for more than 24hours.
 
#indicate that the program should run for 1 hour and use 20G of total memory
 
#indicate that the program should run for 1 hour and use 20G of total memory
#$ -l memfree=20G
+
#$ -l mem_free=20G
 
#$ -l h_vmem=40G
 
#$ -l h_vmem=40G
 
#$ -l s_rt=360
 
#$ -l s_rt=360

Revision as of 12:56, 20 February 2014

In Submitting your job using SGE we looked at the following script;

   
#specify the shell type
#$ -S /bin/sh

#run in the current working directory
#$ -cwd

#run on a node that has 20GB free, specify that the job will be killed if it uses over 40G or total memory or run for more than 24hours.
#indicate that the program should run for 1 hour and use 20G of total memory
#$ -l mem_free=20G
#$ -l h_vmem=40G
#$ -l s_rt=360
#$ -l h_rt=3600

#specify exactly which node you wish to use (this isn't really recommended)
#the only time you may wish to use this is if there is specific hardware/software
#available on that node. In the case of Bert and Ernie, we don't have that.
#$ -l h=node006

#specify the number of cores required
#$ -pe mpich 1

#specify which queue you wish to use
#$ -q amd.q

#run a program command to print hostname and uptime
hostname && uptime
    

Most of this was not really required for the actual job (e.g. getting the hostname and uptime from each machine). However it is good practice to include limits for sun grid engine to work well.

Sun Grid Engine and Limits