Skip to content

Conversation

@brownbaerchen
Copy link
Contributor

When running the code on the cluster, you want to use modules as much as possible because they are installed in an optimised way. Installing MPI with conda, for instance, will give significantly lower performance compared to MPI modules.
However, there is not a module for everything and so you have to install some custom stuff using pip. In Jülich there are quotas for the number of files you can have in ~/home, which is quickly exceeded when installing large packages like numpy.
The support team recommended to use python venvs instead, which can be stored wherever you want. There is a handy template for this that let's you choose you own modules and packages and then easily activate the venv by running source activate.sh. By adding this to the projects, others can see how to install on Jülich machines and adapt to their own machines if needed. So this is also good for reproducibility. I do recommend this to anyone running python on clusters!

@codecov
Copy link

codecov bot commented Jun 28, 2024

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 77.83%. Comparing base (18989d3) to head (9b4892e).
Report is 31 commits behind head on master.

Additional details and impacted files
@@            Coverage Diff             @@
##           master     #451      +/-   ##
==========================================
+ Coverage   77.38%   77.83%   +0.44%     
==========================================
  Files         327      332       +5     
  Lines       26085    25843     -242     
==========================================
- Hits        20187    20114      -73     
+ Misses       5898     5729     -169     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

@pancetta pancetta merged commit a20c787 into Parallel-in-Time:master Jun 28, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants