Loading Events

Advanced Slurm Training

13 July - 16 July 2021
9:00am - 12:00pm

This training is targeted at users who have already used SLURM but whose needs go beyond simple batch files or small interactive jobs.

The training outline follows:

  1. Slurm Refresher
    1. How Slurm actually works.
    2. How Slurm schedules jobs.
    3. How long to wait; how to better schedule jobs.
    4. Slurm and priorities; how is it done?
  2. Key features
  3. Resource Management
  4. Running a job; job/step allocation
    1. Examples – GPUs
    2. Examples – Job Arrays
  5. Advanced Features
    1. Topology Aware Scheduling
    2. Job Sanity Check
    3. Job profiling
    4. Multithreading (SMT)
    5. Heterogeneous j obs
  6. Job Dependencies
    1. Chain Jobs
    2. Staging input before running, and storing outputs
    3. Master/Slave programs
    4. Submitting collections of programs (multi-prog)
  7. System Information Job monitoring
  8. Checkpointing & Restart
  9. Use of SLURM API (plans to support this in the future on Pawsey systems)

To register, please complete the following form:

NOTE: This course is capped at 16 attendees. If you cannot attend all 4 sessions, please do not register.

  • For example, PhD, Masters, Researcher, etc.
  • A Pawsey Friend receives occasional newsletters with Pawsey-related updates ranging from events, job opportunities, training, news and more.