In order to see your jobs only, run qstat -u <username>
Any program that is installed can be run interactively (qsub -I [requirements...]. Remember to use the "module load <program>" command to access it after the interactive job has started.
Programs which open a display can also be run interactively. Just use the -v DISPLAY option to qsub.
You can see which nodes you are using on one of the cluster machines with the command: cat $PBS_NODEFILE. This can either be used when you have an interactive job running or it can be added to the job script.
If there is a problem and stdout/stderr is not written to file, that can instead be done directly in the submit-file. Example (MPI - remember to load an MPI module first!) stdout/stderr in same file:
mpirun ... >output 2>&1
stdout/stderr in separate files:
mpirun ... >output 2>error
When running OpenMP programs, you need the processors to be on the same node to get the advantage of shared memory. Remember, check how many processor a node has and don't ask for more than that.
If you run several jobs from one script, you do not want the output-files/error-files for the later ones to overwrite the earlier ones. This can be solved by using names of the following type for the output/error files:
The file names will then contain the job-id, and so have a separate name for each job.
If your parallel program prints out something, the order of the print statements will be as default be random.
Remember that the number of processor you ask for on a node (ppn) can not be larger than the number of processors on each node on the machine in question.
If you do not define how much memory you want, you will be given the default amount, which is 1900mb on Akka (pmem=1900mb and pvmem=2000mb). In order to determine how much memory your program needs (if you do not know), you can look at https://www.hpc2n.umu.se/node/226 to see how much memory your job used (if you have already tried running it).
The maximum amount of memory you can ask for on Akka is pmem=15700mb and pvmem=16000mb. pvmem must always be greater than pmem.
The batch system will allocate a job to a combination of nodes where each node has at least pmem * requested-number-of-processors-per-node (default pmem is 1900mb on Akka)
Note: if you specify a combination (mem/procs) that requires more memory per processor then the nodes have, the job will not start.
