Apply for a Compute Project¶
The NHR@KIT Center provides level 2 high-performance computing resources. The resources are allocated on a project basis according to a joint science-led procedure.
For detailed information on eligibility, categories, templates, and deadlines, see the linked guides:
If you're already familiar with the process and only need to create a new application or review submitted ones in the project application system (NHR-JARDS), please follow the link:
Who is eligible to apply?¶
Scientists with a doctoral degree who belong to a German accredited university are eligible to apply. Within the scope of the approved projects, non-doctoral scientists can also use the computers.
Use of HoreKa is free of charge, however the utilization of HoreKa has to be acknowledged on all publications and annual status reports have to be submitted to the NHR@KIT.
When you are submitting your project, please use your formal e-mail of the German University you are affiliated with.
Project Categories¶
NHR@KIT offers five categories of projects depending on compute resources needed which implies different call and review types.
Normal and Large Projects¶
| Project category | Duration | Extendable | Min. Requested Resources | Max. Requested Resources | Review | Call |
|---|---|---|---|---|---|---|
| NHR Normal | 1 year | Yes | 2.000.000 CPUh / 30.000 GPUh | 14.999.999 CPUh / 199.999 GPUh | 2x Scientific | Rolling |
| NHR Large | 1 year | Yes | 15.000.000 CPUh / 200.000 GPUh | 70.000.000 CPUh / 1.000.000 GPUh | 2x Scientific + NHR Steering Committee | Quarterly |
Test, Starter, and AI Fast Track Projects¶
| Project category | Duration | Extendable | Effectively Granted Resources | Review | Call |
|---|---|---|---|---|---|
| NHR Test | 6 months | No | 500.000 CPUh / 5.000 GPUh | Technical | Rolling |
| NHR Starter | 1 year | No | 360.000 CPUh / 10.000 GPUh | Technical | Rolling |
| AI Fast Track | 1 year | No | 380.000 CPUh / 20.000 GPUh | None | Rolling |
Note on effectively granted resources
We define the resources that you applied for in your application as "requested CPU" and "requested GPU". However, when we approve your application we grant you the so called "effectively granted CPU" and "effectively granted GPU". The effectively granted CPU = approved CPU + 19 * (approved GPU). So when you request resources in your application please be aware that you could sometimes be granted more. The reason for this is that every GPU is binded to 19 CPU.
Tip for estimating resources
It can be hard to estimate how many resources (CPU core hours / GPU hours) your project will need. The easiest way to find out is to apply for a NHR test project before handing in the proposal for a NHR normal or large project.
Resource calculation
Please keep in mind that individual compute nodes on HoreKa are always assigned to a single job exclusively. Please calculate your resource consumption with 76 CPU cores per node, and not 152 CPU threads per node (see the Project accounting page).
AI Fast Track currently suspended
Note: AI_FAST_TRACK types of project are suspended for the moment. Thus we don't accept any application of type AI_FAST_TRACK at the moment.
Recommendation for smaller categories
We recommend and encourage PIs to apply for Normal or Larger category of projects after they have submitted any project in the following categories: Test, Starter, or AI_FAST_TRACK.
Deadlines¶
NHR Normal projects¶
Application for normal projects can be done any time of the year (rolling call) and can be submitted on an ongoing basis. Allowed activities prior to the official project start can be found here (exceptions: NHR@Göttingen and NHR@ZIB - please note the deadlines for major projects for all projects).
NHR Large projects¶
Application for large projects can be only done quarterly and start time is scheduled 3 months after the submission deadline. The current deadline is displayed bold in the following table.
| Call no. | Submission deadline | Deadline technical reviews | Deadline scientific reviews | NHR steering committee meeting | Project start |
|---|---|---|---|---|---|
| 1 | 01.10. | 15.10. | 20.11. | Mid Dec | 01.01. |
| 2 | 01.01. | 15.01. | 20.02. | Mid Mar | 01.04. |
| 3 | 01.04. | 15.04. | 20.05. | Mid Jun | 01.07. |
| 4 | 01.07. | 15.07. | 20.08. | Mid Sep | 01.10. |
Application for computing time via NHR-JARDS¶
All the following steps should be done by the PI.
Step 1: Apply for resources via NHR-JARDS
Please apply for resources for your project via NHR-JARDS.
Note
This step is not needed if you are applying for "AI Fast Track".
Simplification for Well-Known Codes
Projects relying on well-known and established codes at NHR@KIT are not required to provide a detailed scaling analysis in the project description. This simplification applies at NHR@KIT for the following codes or applications:
Earth and Climate Science
- ICON
- PALM
- PALM-4U
- EMAC
Engineering
- OpenFOAM
- OpenLB
- Abaqus
- ANSYS CFX/Fluent
- StarCCM+
- COMSOL
Chemistry/solid-state-physics/atomistic simulations/Materials Science
- CP2K
- NWChem / NWChemEx
- Gromacs
- LAMMPS
- NAMD
- Quantum Espresso
- Turbomole
- VASP
Particle and Astroparticle Physics
- CORSIKA
- COBalD/TARDIS
Machine Learning
- PyTorch
- Tensorflow/Keras
- Heat
Step 2: Create project in FELS
Follow this link and select your organisation. Log in with your login details. Then you will see a list of all projects for which you are the Principal Investigator (PI). You now have the option of creating a new project.

Step 3: Configure project details
This page opens after you press the "Create new project" button. Here you can name your project and give it a short name as well as a group name and add a short and a detailed description. Then click on "Save".

Step 4: Select resources
You will then be taken to this page where you have the option of either deleting the project, or adding further members, or selecting a resource for your project.

Step 5: Connect with service
In this case, please click on "Connect with service" and select "HoreKa". Then press "Submit request" to submit your request for your project to us.

Step 6: Invite team members
You then have the option of inviting other people to your project. Please enter the e-mail address and name of the person you would like to add to your project. Under "Sender e-mail" and "Sender name" please enter your own e-mail address and your own name. Then click on "Send invitation".

Step 7: Upload signed form
Please upload the form which was signed by the PI to NHR-JARDS, which you received via mail after you applied for the project.
Note
This step is not needed if you are applying for "AI Fast Track".
Step 8: Notify project review
Send a message to project-review@nhr.kit.edu stating that you have created your project and informing about the ID In FELS of the project and its name.
If you apply for "AI Fast Track", please fill in this form and attach it to your mail.
Step 9: Confirm first access
After you access your project on the machine for the first time, kindly send an email to project-review@nhr.kit.edu stating that you have accessed the machine, in order for us to setup the correct resources granted for you.
Note
This step is not needed in case of test applications.
Templates for detailed project descriptions¶
During the application process you are required to upload a detailed project description. A template is provided for each project category. Using the structure given in the template is mandatory. Applications that do not follow it might be rejected.
| Category | Template | Submission |
|---|---|---|
| Normal / Large | NHR templates | NHR-JARDS |
| Test | PDF, DOCX, LaTeX | NHR-JARDS |
| Starter | TBD | TBD |
| AI Fast Track | PDF, DOCX, LaTeX | Signed, by e-mail (not via NHR-JARDS) |
| HAICORE 3.0 | TBD | TBD |
If a Normal or Large project is funded (whitelisted), choose the whitelisted template.
Other documents¶
- Resources Release form - please contact project-review@nhr.kit.edu for this document
- Access Form - please contact project-review@nhr.kit.edu for this document
Questions?¶
In case you have any further questions or inquiries, please contact us via mail (project-review@nhr.kit.edu).