STOPOS 2022.08
It is common practice to submit many jobs: each job with about the same content, but with different parameters for the program to run. Examples are parameter scans and Monte-Carlo simulations. There are a number of problems with this kind of jobs: - The number of jobs is so large that a script or program has to be developed to generate these jobs. - In order to make maximum use of the capacity of the system, care should be taken, that the maximum number of programs are running simultaneously on a node. - In a compute cluster, it is possible that nodes contain different numbers of cores, which complicates the task to generate efficient jobs. - It happens that the running time of the programs are different, so that at the start of a job the optimal amount of programs is running, but that later on some programs are already finished, resulting in a sub-optimal filled node. All these topics can be taken care of using conventional scripting, but in practice this is so complicated that almost nobody does that. Therefore we developed a tool to facilitate the construction of efficient jobs: Stopos.
Tags | batch slurm tasks |
---|---|
License | Apache |
State | mature |