The CSF2 has been replaced by the CSF3 - please use that system! This documentation may be out of date. Please read the CSF3 documentation instead. To display this old CSF2 page click here. |
Time Limits and Short Jobs
Default Maximum Wallclock Time (Intel nodes)
The wallclock time is currently set to a maximum of 7 days (unless otherwise noted in the
parallel environment (PE) table).
A 7 day runtime limit also affects how long your jobs may have to queue for. In general, the longer the permitted wallclock, the longer you potentially have to wait for your job to run. A 7 day wallclock limit is longer than most HPC systems offer. We find that this suits our users’ workloads. But it does mean when the system is busy you could be waiting for up to 24 hours for your jobs to run (jobs are finishing all the time on the CSF and hence new jobs are continually being selected to run – but please be patient if your job does not run immediately).
FAQ for jobs that require more than 7 days runtime.
Specifying a Shorter Wallclock Time
It is recommended, if you know the wallclock time you require, to state this during submission as this helps the SGE scheduler make decisions. This is, however, optional (unlike some HPC systems). This is done by adding to your jobscript:
#$ -l s_rt=hh:mm:ss
where hh is the number of hours, mm is the number of minutes, ss is the number of seconds.
For example to give a limit of 10 minutes add the following line to your batch script:
#$ -l s_rt=00:10:00
An example of 6 hours:
#$ -l s_rt=06:00:00
Note that s_rt
causes the batch system to send a soft kill signal which some applications will detect and then shutdown cleanly – for example saving current state and results before exiting but this depends on your application’s capabilities.
Use h_rt
for a hard time limit (meaning the batch system will simply kill the job when the time limit is reached).
If a job terminates at the time limit it may not save any data – this is dependent on the software or code you are running. Some applications can checkpoint and then be restarted from a known status point. Please consult the manual for your software for more information.
Short Jobs
Intel
Two 12-core Intel westmere nodes (Gigabit ethernet only, no infiniband, no sandybridge, no ivybridge, no haswell, no AMD) have been reserved for short test jobs or work such as post-processing. To access it please add the following option to your batch script:
#$ -l short
If submitting from the command line you do not need the #$
just include -l short
.
The maximum wallclock time with this option is currently 1 hour and the maximum number of cores is 12.
Please do not fill ‘short’ with production compute work as it means people who genuinely need to test cannot access it. Job arrays should not be run in the short environment because they can prevent any other user getting a job to run due to the small number of cores assigned to the short environment.
Please note that these nodes have only 2GB of RAM per core.
AMD Bulldozer
One bulldozer node has been reserved for compiling code and short test jobs. Details on how to use it are on the AMD compiler page.
Jobs of Up to Two Days Wall Clock Time
As of the 14th January 2016 upgrade two day jobs cannot be run. Please do not use this flag in your jobscripts.
#$ -l twoday
If you wish to run a job that you think will not require the full 7 day runtime it is possible to specify a shorter runtime (see above) which the batch system will take into account and potentially give your job a slightly higher priority.