Cotainr
The Cotainr software tool makes it easier for users to utilize supercomputer resources without having to deal with complex technical setups.
Cotainr is designed for researchers working with AI, machine learning (ML), and data science in Python. The tool packages the researcher's Python environments and necessary libraries in a container, making it possible to run calculations directly on the HPC facility without first having to install software on local machines.
The container-based setup means that researchers can easily transfer their existing Python environments to the LUMI supercomputer, whether they use tools like Conda and Pip. Cotainr eliminates the need to have in-depth knowledge of Linux or command-line tools and simplifies the process by allowing researchers to focus on their research rather than technical challenges.
The first version of Cotainr focuses on Python applications in AI and ML, where it can be used for tasks like deep learning training and data analysis. The tool enables researchers to define their environment in a single file and then easily create and run that environment on LUMI's powerful GPUs.
Cotainr is part of DeiC's broader vision to make supercomputing more accessible and user-friendly. The tool is open source and freely available on Github, enabling everyone to benefit from its features and contribute to its further development. The goal is to extend Cotainr to more HPC facilities, allowing researchers to move their code and simulations between different systems with minimal effort.
Cotainr is developed by DeiC's HPC team. In collaboration with LUMI-LUST, an integration to LUMI has been developed, making the tool easy to use and directly accessible on LUMI.
Who can use Cotainr?
The target audience for Cotainr is researchers and developers who want to run Python AI software on HPC facilities such as LUMI. LUMI.
What does it cost to use Cotainr?
Cotainr is open source and freely available on GitHub, enabling everyone to benefit from its features and contribute to its further development.
How do I get started with Cotainr?
DeiC has developed a guide for those who want to get started with Cotainr: Getting Started - cotainr
A detailed user guide can be found on the LUMI website: Python packages - Documentation
- Social Science
- DeiC Interactive HPC
Where can I get help?
Questions regarding the use of Cotainr in connection with LUMI can be directed to LUMI's user support team LUMI LUST.
If you are interested in learning more about Cotainr in relation to other HPC systems, you can contact the DeiC service desk.
On Gibhub there is an issue tracker where users can report bugs in the code or suggestions for improvements.