Last update: 05/14/2008
On NCAR supercomputers, you will run most of your production jobs in one of the batch queues. You may run short jobs interactively, but the session must not exceed 30 minutes of wallclock time, and the job's memory usage should not preclude others from using the interactive nodes. If your job does not meet these conditions, it will likely be dropped. Long-duration jobs must be submitted to the LSF batch job scheduler using the queuing system on the computer where the job will run. Please run any heavy interactive jobs in the share queue. Please consider running any data analysis or visualization jobs on the data analysis cluster as per the information provided below.
Production computing resources at NCAR are shared by a large community of users. To serve everyone contending for these resources, production work must be scheduled or coordinated. On the supercomputers, major production computing work and all computing resources are controlled by the Load Sharing Facility (LSF) batch job scheduler from Platform Computing Inc. The vendor provides LSF manuals for browsing at the Platform Knowledge Centre. Also, please see these documents for machine-specific batch job information:
Each production supercomputer has a specific set of job queues that classify jobs for the batch scheduler. The job queues are specified on the queues page for each computer:
Tasks such as code development and short checkout runs are appropriate for interactive runs. Please note that interactive use of the supercomputers for individual users has been limited to short sessions -- 30 minutes wallclock or less. Longer-duration work must be run in one of the queues or performed on another platform that accepts longer interactive work. All of the common UNIX shells are available on the interactive nodes.
The data analysis cluster is a platform where users run jobs for data analysis and visualization. These jobs are run interactively rather than in a batch system. In fact, the nodes do not support a batch system. Please see this FAQ for information on running long jobs on the Data Analysis Cluster. Beyond this FAQ, please see this link for collected information on the Data Analysis Cluster
.
Overview of computing at NCAR - Table of contents
If you have questions about this document, please contact us via any of the methods (phone, email, ticket, or in person) described here: CISL Customer Support.
© Copyright 2003-2008. University Corporation for Atmospheric Research (UCAR). All Rights Reserved.
Address of this page: http://www.cisl.ucar.edu/docs/access/run.html