The SIP High Productivity Toolset for Parameter Sweeps and Monte Carlo Runs

Many high performance computing tasks take the form of parameter sweeps. The same task is run many times with different sets of input parameters. For example, a researcher may run thousands of radar signature jobs over various azimuth and elevation angles. Keeping track of the status of the overall situation can be daunting and tedious, leading the user to manually track the state of all the jobs. Time that could otherwise be spent on analysis is spent managing the parameter sweep itself. The signal and image processing high productivity toolset (SIPHPT) automates much of the parameter sweep management activity. With the user supplying a single configuration (setup) file, the SIPHPT provides utilities to submit jobs, verify proper job setup, and monitor overall status. The SIPHPT creates underlying master scripts to manage the jobs. In the event jobs do not complete successfully, the exact same commands (and setup file) can be used to submit and monitor the jobs again. The master script will only run jobs that have not yet been successful. The SIPHPT thus provides utilities to reduce manual job management activity associated with large parameter sweep jobs, availing users more time to concentrate on analysis. We describe the SIPHPT and provide information on where to find it on Department of Defense (DoD) high performance computing (HPC) Aeronautical Systems Center (ASC) and Army Research Laboratory (ARL) Major Shred Resource Center (MSRC) systems.