LILAC
|
Nersc has published some good examples of jobs to run, mainly <a href="https://www.nersc.gov/users/computational-systems/genepool/running-jobs/submitting-jobs/" here >
In addition, one can find the script that describes nersc jobs in rundir/nersc_submit_script. The script that runs LILAC is rundir/tf_script.sh. It is a fairly simple script. It creates a filename that is attached to the index, called run#.out. LILAC is then piped to the scratch folder, where the data is stored
To run with varying numbers of cores, one must edit the proper parameters in the script. These are:
In addition, the queue type can be set to regular or debug. The debug queue can only run smaller jobs and should only be used for testing.
To submit a jub, enter the rundir directory of LILAC and perform
This saves the data currently in the ouptut folder, copies the running directory to a compute node and then submits a job. You can check the status of the job with
and delete a job with
The id can be found from the status command
There is also a script called save_data. This will take everything in the folder $SCRATCH/lilac_output and puts it into a .tgz file. It also prints the location of the file, which will be in $SCRATCH/final_data_outs/lilac-out.xxx.tgz. xxx consistof of two things. The first parameter passed to the script, and the second is the data.
This is automatically run upon submitting a script, and all unsaved data will be moved to a tarfile in the save directory called lilac-out.no-name-given-to-save-data.<date>.tgz