Access to HPC resources

The declared goal of HPC.NRW is to provide scientists in NRW with easy and structured access to HPC computing resources.

HPC resources in Germany are arranged hierarchically in the so-called HPC performance pyramid, in which the smaller Tier-3 and Tier-2 systems cover entry-level requirements and large systems (Tier-2 to Tier-0) also enable large research projects:

Tier-0 PRACE and EuroHPC
Tier-1 Gauss Center for Supercomputing (GCS, JSC, HLRS, LRZ)
Tier-2 Network for National High Performance Computing (NHR) and other Tier-2 centers
Tier-3 local HPC centers of the universities

Which tier is suitable for me and my research project?

Scientists at German teaching and research institutions should, where available and suitable, use the resources of their local Tier 3 center. However, only members of the respective university are generally entitled to use these resources.

Good reasons to orientate yourself towards Tier 2 centers are when

more resources are required than the local Tier 3 center can provide,
a local Tier 3 center does not exist at the location,
special requirements are placed on main memory, multi-cores / CPUs or GPUs, for example.

Where can I get access?

Members of a university with its own local Tier 3 center are always eligible for access.

NHR centers and other Tier 2 systems also provide resources for scientists from all public institutions. Among other things, this is also done in the role of Tier 3 substitute provision if no local Tier 3 center is available.

The application/usage authorisation can be found in the table below:

What are small, medium and large computing time projects?

The resource requirements are usually estimated in CPU core-hours (core-h). There are often ways to get easy access for testing and development, e.g. to estimate the computing time and prepare larger requests.

For example, if you want to use a computing node with 48 cores over the entire year (24/7), you would have to apply for 48 cores * 24 h * 365 days = 0.42 million core-hours.
Small projects generally comprise no more than 1 million core-h and normal large projects up to 10 million core-h. Large projects can go well beyond this.

Computing project	Order of magnitude (per year)	Local HPC service (Tier-3 service)	NHR/Tier-2 service	GCS-Center (Tier-1), EuroHPC (Tier-0)
small	up to approx. 1 M CPU-Core-h	local center, small projects	NHR-Starter project, NHR-Test project	-
medium	up to approx. 15 M CPU-Core-h	dependent on local system size	NHR-Normal project	-
large	up to approx. 70 M CPU-Core-h	-	NHR-Large project	GCS Regular project
very large	beyond	-	-	GCS Large-Scale project

How do I reach the local Tier 3 centers?

Please refer to the Quick Reference Cards below for details of the individual centers.

We also provide an overview of the local support teams.

How do I reach the Tier 2 centers?

While the resources of the local Tier-3 centers are usually only accessible to researchers from the local university, external researchers can also apply for computing time at Tier-2 centers.

Further information can be found here:

Aachen (online): https://www.itc.rwth-aachen.de/hpc-project
Cologne (PDF): https://rrzk.uni-koeln.de/en/hpc-projects/hpc/access-and-use-instructions
Paderborn (online): https://pc2.uni-paderborn.de/go/access

Use the NHR website to apply for comp uting time to gain access to the above-mentioned and other NHR centers.
The NHR centers offer test and starter projects to make it easy to get started.

NHR centers also support users in various key areas. For example, NHR4CES concentrates on the field of "Computational Engineering Science". In the list of application support of the NHR centers you will find the contact paths, e.g. to chemistry, earth system sciences, humanities and social sciences, life sciences and physics.