Applying for External Computing Time

For some projects, the local infrastructure is not enough. We can help you choose and apply at an external computing centre with more resources.

There are various national and international calls to apply for computing time at various levels differentiated by the amount of available computing time. In general, the more computing time a project can apply for in a given call, the more involved is the application procedure.

The computing infrastructure in Germany is organised into tiers (or levels) starting at the institute level and going through Tier-3, Tier-2 and Tier-1 (national) and Tier-0 (international) with the latter referring to the most advanced systems with the largest amount of computing time available to individual projects. For more details and a helpful flow-chart to guide you, please refer to the HPC Resources page at HPC.NRW.

Tier-0 PRACE und EuroHPC     Tier-1 Gauss Center for Supercomputing (GCS, JSC, HLRS, LRZ)     Tier-2 Verbund für Nationales Hochleitungsrechnen (NHR) und weitere Tier-2 Zentren     Tier-3 lokale HPC-Zentren der Hochschulen
© HPC.NRW

Differences between Tiers

  • At the lowest level, projects are granted quickly (a few days to a few weeks at most)  based on a short description of the project and the intended outcome. In general, no stringent restrictions are placed on the application codes at this level.

  • At the most competitive levels, applications are peer-reviewed both scientifically and technically. The time betwen application and project start can be up to half a year and the technical review requires scalability and efficiency to be demonstrated for the employed algorithms and codes. When projects end, computing centers ask principal investigators (PIs) for lists of publications and the submission of projects reports. Of the latter, it may be that both a technical report as well as a project summary suitable for the general public (to be published on the centre website) may be requested.

  • At the intermediate level, the selection criteria are naturally less stringent and the time between application and project start is generally between a few weeks to a few months.

  • While not always true, calls at the most selective tiers are regular (twice to four times a year) while calls at the lower tiers are rolling or continuous and may be reviewed quite quickly. Some rolling calls have regular cut-offs, however (applications can be submitted at any time but are only reviewed n times a year).

Overarching Organisations

At Tier-3, HPC.NRW acts as a bridge and coordinator between Tier-3 and Tier-2 centres in NRW, providing support, documentation, training and fostering a cooperative atmosphere within the state.

At Tier 2, computing time is avaiable in Aachen, Cologne and Paderborn for researchers at public research institutions in the state of NRW. Nationally, the  NHR Initiative is currently in the process of reorganising access to these resources, such that researchers from any state can apply for access to resources in any other state. NHR should lead to a more unified application procedure at this level while at the same time improving the available support in particular research fields.

At Tier-1, the Gauss Center for Supercomputing (GCS) coordinates between the top computing centers in Germany, providing a unified application procedure as well as advanced training and support activities. It also acts as a bridge between Tier-2 and Tier-1.

Beyond Germany, PRACE and EuroHPC both offer rolling and regular calls at different levels. On the one hand, PRACE calls may allow the usage of resources which are not available in Germany (such as certain types of CPUs or GPUs) while on the other hand, project may apply for access to extreme levels of computing time.


Project categories, computing centres and application frequencies

We collected information for regional, national and international computing centres for you.

Download the table here

Glossary of Terms

Term Definition
EFLOP Exaflop, number of floating-point operations available to a project (per year) → unified measure for amount of computing time available on both CPU and GPU machines at JSC
EuroHPC European High Performance Computer Joint Undertaking
FAU (RRZE) Friedrich-Alexander-Universität Erlangen (Regional Computing Center)
GCS Gauss Centre for Supercomputing
HLRN North German Supercomputing Alliance
HLRS High-Performance Computing Centre Stuttgart
IAS Institute for Advanced Simulation, Forschungszentrum Juelich
JSC Juelich Supercomputing Centre, Forschungszentrum Juelich
KIT Karlsruhe Institute of Technology
LRZ Leibniz Supercomputing Centre
NHR ""Nationales Hochleistungsrechnen"" / National High Performance Computing
NHR4CES National High Performance Computing Centre for Computational Engineering Sciences
PC² Paderborn Center for Parallel Computing
PRACE Partnership for Advanced Computing in Europe
SCC Steinbuch Centre for Computing, KIT
TUD (ZIH) Technische Universität Dresden (Center for Information Services and High Performance Computing)
ZIB Zuse-Institut Berlin

Contact

HPC/A Lab

We can assist you in choosing a computing centre for your project! Contact us below.

Wird geladen