next up previous contents
Next: Up: Utility Scripts Previous:   Contents

A wrapper script for running multiple copies of experiments (in copied directories). You should make sure that there are no large files in the experiment directories (use links for large data files) if you cannot afford the disk space.

On a single processor, this script will run jobs in serial. If you have a cluster, or a multi-processor machine, you can use the -queue option but you must modify the script to work with your local queueing system.

You run the script from the parent directory of the experiment directory:

  usage: -num 20 -hours 4 expt_dir_name

The main options are:

-number X
- makes X copies of the experiment and runs them
-hours H
- runs the jobs for H hours
-mins M
- or use this if you count in minutes
- use a queueing system (see above)
- pass the -loop option to

Other options should be self-explanatory in the script itself.

Bob MacCallum 2003-02-03