Runtime limits for GPU Jobs
On the 06.03.2026 we changed how many GPU nodes from the c23g partition are allowed to run user GPU Jobs with runtimes longer than 24 hours.
The change has the following main goals:
- Increase the throughput and reduce the waiting times for GPU jobs with runtimes lower than 24 hours.
- Encourage users to submit shorter jobs and make use of resilience methods like checkpointing if necessary.
- Reduce the maintenance downtime of GPU nodes.
The change effectively limits the amount of long running GPU jobs to only half the GPU nodes in the c23g partition. Long running GPU jobs are user GPU jobs with runtimes longer than 24 hours. GPU jobs with less than 24 hours of runtime are considered short and will be scheduled to all available GPU nodes of the c23g partition.
It is understood that the waiting times of long running GPU jobs will increase and we therefore encourage users to change their workflows to accommodate shorter running GPU jobs.
This change is necessary to improve QoS for users of the c23g partition and to allow for faster maintenance works on the GPU nodes.
Release Notes Version 2.48.1
Improvements and bug fixes:
- Various bug fixes
Support for Embedding Models with API Keys and Chat Model Switching (v1.13.0)
Features:
- Support for Embedding Models with API Key Usage
- Model switching within an active chat is now possible
- Expanded and updated model catalog
Bugfixes and Improvements:
- Improved Azure OpenAI support and compatibility (more consistent behavior across providers)
- Refined Sidebar & User Menu UX (dropdown menus, modals, improved usability)
- Corrected default deployment selection
- Improved syntax highlighting in code output
- Logos now display correctly after tenant switching
- Fixed API key inactivity/expiration logic
- Fixed file upload via keyboard
- Various minor UI and stability fixes
Support for Embedding Models, Iterative Image Generation, and Chat Model Switching (v1.13.0)
Features:
- Added support for embedding models:
- e5-mistral-7b-instruct, multilingual-e5-large-instruct, qwen3-embedding-8b, text-embedding-3-large , text-embedding-3-small , text-embedding-ada-002
- Added image input for image models; image models now use the most recently generated image as the default input
- Model switching within an active chat is now possible
- Expanded and updated model catalog
Bugfixes and Improvements:
- Improved Azure OpenAI support and compatibility (more consistent behavior across providers)
- Refined Sidebar & User Menu UX (dropdown menus, modals, improved usability)
- Corrected default deployment selection
- Improved syntax highlighting in code output
- Logos now display correctly after tenant switching
- Fixed API key inactivity/expiration logic
- Fixed file upload via keyboard
- Various minor UI and stability fixes
Call for participation
For an analysis of the joint use of Coscine and the RWTH HPC system, we are looking for people who actively use both services. The aim is to gain a better understanding of typical use cases, workflows, and data flows. Your experiences will help us to further develop support structures and tools in a targeted manner and align them with the actual needs of users. By sharing your approaches and perspectives with us, you will make an important contribution to the optimization and further development of both systems. If you are interested in participating, we look forward to hearing from you at servicedesk@rwth-aachen.de. Thank you for your support!
Interactive Workflow Graphic
An additional visual change concerns the Coscine website. The familiar workflow graphic is now interactive and provides a good starting point for new users.
Display of Available Storage Space
The display of storage space blocked by versions in the web interface will be available on March 3, 2026. This will give researchers a more transparent overview of storage space already used and still available.
Resource Creation
The “Create” button in the second step of resource creation was temporarily not working. The development team has now fixed the bug. The button will be functional again with the next live deployment on March 3, 2026.
Release Notes Version 2.48.0
Improvements and bug fixes:
- RWTHmoodle course list can be sorted by name and last access
- The barcode scanner in the university library (UB) has been improved
- Various bug fixes
Update to Moodle 4.5.10
RWTHmoodle has been updated to version 4.5.10.
