Difference between revisions of "Torque"

From HPC Wiki
Jump to navigation Jump to search
Line 78: Line 78:
  
 
This serial job will run a given executable, in this case "myapp.exe".
 
This serial job will run a given executable, in this case "myapp.exe".
<syntaxhighlight lang="zsh">
+
<syntaxhighlight lang="bash">
#!/usr/bin/env zsh
+
#!/usr/bin/bash
  
 
### Job name
 
### Job name

Revision as of 15:14, 20 April 2018

General

Torque is a job scheduler. To get an overview of the functionality of a scheduler, go here.

Job Submission

This command submits the job you defined in your jobscript to the batch system:

$ qsub jobscript.sh

Just like any other incoming job, your job will first be queued. Then, the scheduler decides when your job will be run. The more resources your job requires, the longer it may be waiting to execute.

You can check the current status of your submitted jobs and their job ids with the following shell command. The most common states for a job are queued Q (job waits for free nodes), running R (the jobscript is currently being executed) or on hold H (job is currently stopped, but does not wait for resources). The command also shows the elapsed time since your job has started running and the time limit.

$ qstat -u <user_id>

In case you submitted a job on accident or realised that your job might not be running correctly, you can always remove it from the queue or terminate it when running by typing:

$ qdel <job_id>

#PBS Usage

If you are writing a jobscript for a Torque batch system, the magic cookie is "#PBS". To use it, start a new line in your script with "#PBS". Following that, you can put one of the parameters shown below, where the word written in <...> should be replaced with a value.

Basic settings:

Parameter Function
-N <name> job name
-o <path> file to write stdout to
-e <path> file to write error output to
-j oe both the output and error log will be written to the same log file called <job_name>.o<job_id>

Requesting resources:

Parameter Function
-l walltime=<total_time_limit> time limit (including waiting time in the queue!) in the format hours:minutes:seconds; once the time specified is up, the job will be killed by the scheduler
-cput=<runlimit> maximum execution time, specify as above
-l mem=<memlimit> memory limit per process as an integer number, followed by a unit, e. g. 400MB
-l nodes=1:ppn=1 ask for a single processor for a sequential application

Email notifications:

Parameter Function
-M <address> set email address
-m a receive a mail if your job gets aborted
-m b get notified when your job starts running
-m e receive a mail when your job has finished
-m abe enable all mail options above

Parallel programming (read more here):

Parameter Function
-l nodes=1:ppn=<threads> specify the number of threads to use for an OpenMP application; set OMP_NUM_THREADS accordingly
-l nodes=<num_nodes>:ppn=<num_cores> specify number of processes to start (one per core), which is num_nodes*num_procs; be careful about your system's architecture and do not request more nodes or cores than available on the machine

For hybrid programs, make sure to disable processor affinity by adding export MV2_ENABLE_AFFINITY=0 to your script. Otherwise, all threads will be using the same core.

Jobscript Examples

This serial job will run a given executable, in this case "myapp.exe".

#!/usr/bin/bash

### Job name
#PBS -N MYJOB

### File where the output should be written
#PBS -o MYJOB_OUTPUT.txt

### Time your job needs to execute, e. g. 1 h 20 min
#PBS -l cput=1:20:00

### Memory your job needs, e. g. 1000 MB 
#PBS -l mem=1000MB

### The last part consists of regular shell commands:
### Change to working directory
cd /home/user/mywork

### Execute your application
myapp.exe

References

Overview of how to write a jobscript for Torque

Job submission on Torque

Guide to the Torque scheduler