Skip to content
Snippets Groups Projects
Select Git revision
  • 72a88823919901e9bc576c3e73a800e1ecc862c0
  • main default protected
2 results

workshop-2023.11-matlab4hpc

Training Course 2023, MATLAB for HPC at Jülich Supercomputing Center (JSC)

Document Updates:

  • 21.10.23, 12:00 : initial commit
  • 06.11.23, 18:00 : recordings added

General Information:

MATLAB for HPC at Jülich Supercomputing Center (JSC) is a hybrid workshop on parallel computing using large compute resources. The workshop includes hands-on exercises where you learn how to effectively use MATLAB and MATLAB Parallel Server to speed up your computations on the cluster at the Jülich Supercomputing Center.

Attendees will learn the following:

  • How to submit MATLAB jobs to the HPC cluster, including hands-on submission of example jobs
  • Ways to tune job submissions, including accessing GPUs on the HPC cluster
  • How to optimize job submissions
  • Troubleshooting job submission techniques
  • Best practices for rehosting code
  • Opportunity to discuss their MATLAB code in-person with experts from MathWorks

Target Audience

Beginner/Intermediate: Students, postdocs, and other academic users with a beginner/intermediate knowledge of MATLAB and/or using MATLAB for parallel computing on an HPC cluster. Users from commercial backgrounds are welcome to attend, but cannot access the cluster hands-on.

Notes

For every Participant, please log in to Zoom during the course under the full name (as you provided in your registration). This is important for checking the registration. Participants who do not provide their full name will be excluded from the course.

Main Data:

date
November 6th, 2023 09:00-12:30 (CET)
instructor:
Raymond Norris, MATHWORKS

Raymond Norris has worked at MathWorks for over 25 years as a Quality Engineer, a Technical Consultant, and now, most recently, as an Application Engineer. He has written training courses and workshops at MATHWORKS on MATLAB Compiler, MEX, and Parallel Computing. Raymond currently leads a small team of engineers to integrate MATHWORKS parallel tools with end-users' HPC cluster and cloud environments.

Raymond holds a B.S. in computer science and applied mathematics from the University of Massachusetts Boston and an M.S. in software engineering from Brandeis University.

contents level:
Beginner's contents: 20 % Intermediate contents: 80 %
venue
On-Site, FZJ, Building 16.4, Room Rotunde Online, https://fz-juelich-de.zoom.us/j/69592445601?pwd=bHpydUFoQ053cmpCOFVQNEQvTk1adz09
workshop materials
Slack Channel (for the workshop and beyond): https://join.slack.com/t/matlabforhpc/shared_invite/zt-26e2ubbis-xHYoxSj0nsp6cKcOwjvNIA Slides + Files: https://gitlab.jsc.fz-juelich.de/goebbert1/workshop-2023.11-matlab4hpc Shared notes: https://gitlab.jsc.fz-juelich.de/hedgedoc/krClbkrfRre-UaOX5V090g#
recordings
1. Scientific Use Case, Andreas Kleefeld (JSC) 2. Parallel Computing with MATLAB, Raymond Norris (MATHWORKS) - Session 1 3. Parallel Computing with MATLAB, Raymond Norris (MATHWORKS) - Session 2
preparation
A) recommended workshop B) pre-workshop todos (click here)
  • ensure you can
  • if you have trouble with these pre-workshop todos we offer an onboarding session on Fri, 30th of October

Agenda:

Title
Monday, 6th
9:00 - 9:20 Welcome & Introduction
9:20 - 9:30 Teaser - MATLAB Use Case
9:30 - 10:30 MATLAB for HPC (part 1)
10:30 - 10:45 Break
10:45 - 12:30 MATLAB for HPC (part 2)
12:30 - 13:30 Lunch + connect to MathWorks
13:30 - 15:00 Open End (back in the Rotunde)

Pre-Workshop Todos

  • Register at https://judoor.fz-juelich.de - Video-Howto (1:10)

    • Click on "Register"
    • Enter the email address you used to register for this event
    • Click on "Send confirmation mail..."
    • Open the confirmation email in your email program
    • Click the confirmation link
  • Join the training project

  • Please be patient

    • This takes as long as the PI needs to respond!
    • The PI has to accept your join-request first
    • After acceptance
      • project will be listed in section "Projects"
      • systems will be listed in section "Systems"
  • Sign usage agreement - Video-Howto (1:13)

    • Login to https://judoor.fz-juelich.de
    • Scroll to section "Systems"
    • Click on "You need to sign the usage agreement to access this system"
    • Sign the usage agreement
  • Please be patient

    • This may take up to 15 minutes!
    • Your HPC account will now be
      • created on the desired systems (~5 minutes).
      • registered to the job scheduler SLURM (~15 minutes).
    • Shortly after your HPC accounts are prepared you will see
      • Section "Software" and
      • "jupyter-jsc" in the section "Connected Services"
  • Enable access to Software MATLAB

    • Login to https://judoor.fz-juelich.de
    • Scroll to section "Software"
    • Click on "Request access to restricted software"
    • Check the box "MATLAB"
    • Check the box "[..] noncommercial academic research [..]"
    • Click on "Send request"
  • Login at https://jupyter-jsc.fz-juelich.de - Video-Howto (1:34)

    • Click on "Login"
    • Login with your new JSC account
    • If this is your first login
      • click on "Register"
      • agree to the terms of service
      • click the link in the confirmation mail
  • CONGRATULATIONS - YOU ARE READY FOR THE TRAINING COURSE

FAQ: