Introduction to GPU programming II: Open MP

JupyterHub

Get access to a JupyterHub instance with access to an A40 GPU on Alex:

• Go to https://hub.nhr.fau.de/jupyter-training-nhr-grad-school-access.php
• Enter your e-mail address and submit
• Click on the link in the e-mail you received
• Accept terms and conditions
• Choose profile "Alex - 1x A40, 8 hours (Container - NHR Course Week 2025)
• Click on "Start my server"
• Work ("Try Yourself" section from lecture)
• Stop the server:
  • Go to "File -> Hub Control Panel"
  • Click on "Stop my server"
  • Wait
  • If you don't stop the server, resources will be blocked until the cluster job runs out (up to eight hours)

In JupyterHub, use the Git function or open a shell to clone the notebook repository via

git clone https://git.zib.de/bzbchris/nhrgrads2025.git

easyWave

If you want to try out OpenMP on a real work application code, clone the following repositories: 

• Source Code https://github.com/christgau/easywave-sycl.git
  Take look in the README of this repository to see how to launch easywave.
• Data (needed for running the code): https://git.zib.de/bzbchris/easywave_bm_data.git

Slides

Please look at the SYCL part next day.