Example for a Batch Script for Seaborg

#@ job_name = lss
#@ account_no = mp345
#@ output = lss.log
#@ error = lss.err
#@ job_type = parallel
#@ environment = COPY_ALL
#@ notification = complete
#@ network.MPI = csss,not_shared,us
#@ node_usage = not_shared
#@ class = regular
#
#
#@ tasks_per_node = 16
#@ node = 4
#@ wall_clock_limit= 8:00:00
#
#@ queue

poe+ ./inexe ./indat

This is an example for a batch script for Seaborg. The class can be changed to premium, which increases the priority of the job in the queue, but also charges double the amount of time to the account. Under premium and regular jobs can run for max. 8 hrs on max. 128 nodes. Another option is regular_long, the wall_clock_limit then extends to 24 hrs but the maximum number of available nodes goes down to 32. The maximum number of tasks per node is 16, independent of what one specifies here, one will always be charged 16 processors per node. In this example inexe is the name of the executable and indat is the name of the input data-file for the code. The batch-script in this form should be in the same directory as the executable and the input-file.

Back to the supercomputing page
Katrin Heitmann / LANL / heitmann@lanl.gov / revised January 20
 Valid HTML 4.0!