Skip to main content

Upcoming Training Workshops

March 06, 2019

Eagle uses the workload manager, Slurm. The HPC Operations team will be conducting a series of Slurm workshops to assist users in migrating from Peregrine's Torque/Moab to Eagle's Slurm. In addition to new system capabilities with Eagle, Slurm brings a host of new features and functionality that are worth taking some time to investigate. These workshops will assume basic familiarity with Eagle and Slurm as documented on the User Basics and Running Jobs pages of our website.

Eagle Workshop - Slurm: New NREL capabilities

In the first of our series, Eagle Workshop - Slurm: New NREL capabilities, the following topics will be addressed:

  • Basic Slurm core functionality overview
  • Slurm Partitions - Request by features
    • Effective queue partition requests
    • Request by resource needs
      • GPU compute nodes
      • Local Scratch
      • Memory requirements
  • Job Dependencies and Job Arrays
  • Job Steps
  • Job monitoring and basic troubleshooting
  • Followed by Q&A including extended one-on-one sessions if needed.

Date/Time/Location
Friday, March 8th starting at 1:00PM in the ESIF B308 Visualization Room
Add event to calendar

Eagle Workshop - Slurm: Advanced

In the second of our series, Eagle Workshop - Advanced Slurm Techniques, we will be covering topics that will be beneficial for job management.

  • Job monitoring and forensics: usage examples on sreport, sacct, sinfo and sview (FastX).
  • Advanced srun and sbatch functions (flags).
  • Parallelizing with SLURM.
  • Remote exclusive GPU usage, requesting GPU nodes.
  • Using "srun" in place of mpirun.

Please bring your questions and working examples.

Date/Time/Location
Wednesday, March 20th from 10 a.m. to 11 a.m. in the ESIF B208 Maxwell Conference Room

Add event to calendar