Optimizing Resource Use in Scheduled Jobs

11/20/2020 - 10:00am

Trying to estimate and specify computational resource needs for batch jobs and interactive jobs can be daunting, particularly when the types of resources available are growing. In addition to CPU cores and walltime, users need to consider node memory, GPUs, X11 support, temporary storage, and more. Poorly constructed resource requests can lead to long queue times, poor application performance, and job failure.

To help users make the most efficient use of their time and effort, Brian Vanderwende of the CISL Consulting Services Group will present a 60-minute online tutorial at 10 a.m. MST on Friday, November 20. Topics to be discussed will include:

  • Node resources on Cheyenne and Casper
  • How to specify resource requests in PBS and Slurm
  • How to query resource usage for existing jobs
  • Identifying resource limitations when jobs fail

Please use this link to register. You will receive a link and your personal login details the morning of the tutorial. The tutorial will be recorded and published.