IT Center Changes

As you know, we operate our cluster in 1-cluster-concept. This also means that the system is supported by multiple funding sources with different budgets. The NHR share (Tier-2) is the largest. Unfortunately, the project quota for RWTH-s, RWTH-thesis and RWTH-lecture (Tier-3) in the machine learning segment (GPU partition) is currently overbooked. Therefore, the following changes will take effect immediately for all new project applications (already granted projects are not affected):

• RWTH-s projects: reduction of the maximum project size to 4000 GPU-h per project per year
• RWTH-lecture and -thesis projects: reduction of the maximum project size to 1500 GPU-h per project

In order to ensure that all users can still successfully carry out their research projects, we propose the following approach for future proposals:

• If you do not need more than 4000 GPU-h, please apply for RWTH-s as before.
• If you do not need more than 10000 GPU-h and would like to conduct a project in the field of machine learning or artificial intelligence, please apply for a WestAI project.
• If you do not need more than 10000 GPU-h, you are welcome to apply once (!) for an NHR Starter project. Please note that this category is intended as preparation for an NHR Normal or Large project. An extension is therefore only possible with a full proposal and should be for more than 10000 GPU-h. NHR Starter projects are allocated centrally by the NHR office; if you choose this route, please select RWTH as your preferred center.
• If you need more than 10000 GPU-h, please apply for resources within an NHR Normal or Large project. To ensure appropriate use of the system, NHR projects undergo a science-led peer-review process.

We are convinced that this step is in the interest of all users in order to avoid overloading the GPU partition (ML segment).

Furthermore, please note that in future, NHR Normal projects may request up to 20 million core-h.

Please refer to the project catalogue for a complete overview (including a decision flow chart),

Please contact servicedesk@itc.rwth-aachen.de for any further questions.

Yesterday we updated half of our available compute nodes to Rocky Linux 9. This reduced the available Rocky Linux 8 nodes by half and might lead to increase waiting times on the older OS nodes. We advice all users to continue migrating their workflows to Rocky Linux 9.

If you experience any further issues, please contact servicedesk@itc.rwth-aachen.de.

You can track any disruptions or security advisories that may occur due to the aforementioned change in the Email category on our status reporting portal.

Yesterday we undertook a modernization of the RegApp infrastructure to enhance performance and maintainability. This included implementing containerization and continuous integration/continuous deployment (CI/CD) processes. While the software of the RegApp remained unchanged, we reorganized the configuration distribution and service setups for better efficiency. We apologize for any inconvenience caused and appreciate your understanding as we work to enhance our systems. If you experience any further issues, please contact servicedesk@itc.rwth-aachen.de.

You can track any disruptions or security advisories that may occur due to the aforementioned change in the Email category on our status reporting portal.

For all partitions we had to slightly decrease the maximum available memory. Please consider our documentation for all details.

In order to unify the resource allocation process and to avoid confusion, we will apply the following changes on October 2nd, 2024:

• If you want to apply for GPU resources, you have to use “k GPU-h”  (i.e., thousand GPU hours) instead of “Mio core-h”. This includes the detailed project description as well as the JARDS online form. For convenience reasons, the JARDS system will still calculate the core hour equivalent. Until the end of the year we will accept the old metric in your detailed project descriptions, of course.
• The views in JARDS.project (RWTH projects, NHR projects) will show the used contingents in “k GPU-h” instead of “Mio. core-h” for all GPU resources.
• The command line tool r_wlm_usage will show the used contingents in “k GPU-h” instead of “Mio. core-h” for all GPU resources (i.e. “ML partition”).
• If you act as scientific reviewer, you should recommend GPU resources in “k GPU-h” instead of “Mio. core-h”.

There will be no changes for the HPC (i.e. CPU resources) partition. Please note: The internal billing mechanisms will not change at all. Our Slurm configuration will still use a billing which respects the used memory and core equivalents of a node. Basically, the factor 24 between GPU-h and Core-h will be used.

You can find all limits for the different project categories on our website.

You can track any disruptions or security advisories that may occur due to the aforementioned change in the Email category on our status reporting portal.

Dear CLAIX Users

As you may have noticed, access to the “default” Anaconda repositories (repo.anaconda.com and others hosted at anaconda.com) has been blocked by the firewall on the HPC cluster. This action is necessary because RWTH is not permitted to use the these repositories due to licensing issues.

We understand that blocking the Anaconda domain may disrupt your current workflows. To mitigate this, we set conda-forge as the default channel in /etc/conda/condarc. If you still encounter issues, please check the .condarc file in your home directory and make sure to remove “defaults”, “r”, and “main” from the channel list.

For new Conda users, we suggest using Miniforge, as this distribution uses conda-forge as its default repository.

We apologize for any inconvenience this may cause. If you require assistance, please reach out to servicedesk@itc.rwth-aachen.de.

PS: For a brief time, we accidentally blocked anaconda.org. We have corrected this issue.