Small Compute applications (NWO)

 You can submit the following Small Compute applications for the use of compute systems, storage and support. This includes:

• up to and including 1,000,000 CPU or GPU SBU (standard billing units) of cluster computing on Snellius Supercomputer (including max. 200GB directory storage, 4 hours of support, if required max. 50TB offline tape storage, if required max. 10TB project space storage, Research Drive - via SURF max. 5TB or via institution).
   
• Up to and including 500,000 CPU core-hours and/or 14,000 GPU hours of cluster computing on the LUMI pre-exascale supercomputer (including max. 100,000 TB-hours project space storage, 4 hours of support).
   
• Up to and including 1,000,000 CPU core hours or 10,000 GPU hours of Data Processing on Grid and/or Spider (including up to 200TB disk storage, 300TB tape storage, Research Drive - via SURF (up to 5TB) or via institution, and expertise to be determined in consultation).
   
• Up to and including 50,000 CPU core hours or 5,000 GPU hours of cloud computing on HPC Cloud via SURF Research Cloud (including max. 2TB storage, 8 hours of support, if required max. 50TB offline tape storage, Research Drive - via SURF max. 5TB or via institution).

To be eligible to submit a proposal, researchers must have either a tenured appointment (and therefore hold a paid position for an indefinite period) or a tenure track appointment at one of the research organisations listed below. For proposals for small amounts of computing time only, a main applicant with a temporary appointment may also submit a proposal under the following conditions: 

• The applicant’s appointment will continue for at least as long as the project for which computing time is being requested; and
• The research organisation guarantees the computing time project’s duration and completion. 

• Universities and Universities of applied sciences, as referred to in section 1.8 of the Higher Education and Scientific Research Act (WHW) and the universities listed in the Policy Rule Universities located in the Kingdom of the Netherlands;
• University medical centres, by which is meant academic hospitals as referred to in section 1.13 paragraph 1 of the Higher Education and Scientific Research Act (WHW);
• KNAW and NWO institutes;
• Netherlands Cancer Institute;
• the Max Planck Institute for Psycholinguistics in Nijmegen;
• NCB Naturalis;
• Advanced Research Centre for NanoLithography (ARCNL).
• Princess Máxima Center.
• National Institute for Public Health and the Environment (RIVM);
• Royal Netherlands Meteorological Institute (KNMI);
• Wageningen University (WUR)/Dienst Landbouwkundig Onderzoek (DLO);
• Netherlands Aerospace Centre (NLR);
• Deltares;
• Maritime Research Institute Netherlands (Marin);
• Netherlands Organisation for Applied Scientific Research (TNO);
• IHE Delft Institute for Water Education.

People with zero-hours contracts are excluded from submitting proposals. 

It may be that an applicant’s tenure track employment contract ends before the intended completion date of the project for which computing time is being requested, or that the applicant’s tenured contract ends before this date due to their reaching retirement age. In that case, the applicant must attach an employer’s statement in which the research organisation concerned guarantees that the project and all project members for whom computing time is being requested will be adequately supervised for the full duration of the project. 

Applicants with a part-time contract must be able to guarantee adequate supervision of the project and all project members for whom computing time is being requested.

For both Small Compute Applications (NWO) and Large Compute Applications (NWO), the total capacity available is as follows:

• Snellius supercomputer: 1,902 million SBU (1,283 million SBU CPU thin nodes, 426 million SBU GPU nodes, 184 million SBU CPU fat nodes, 9 million SBU CPU high-memory nodes), with 9,896 TB of Snellius project space.
• LUMI pre-exascale supercomputer: 24.44 million CPU core hours and 1.09 million GPU hours, with 10.1 million TB hours of storage.
• HPC Cloud (via SURF Research Cloud): 26.28 million CPU core hours and 1.33 million GPU hours, with 846 TB of online storage.
• 15,000 TB of Data Archive offline tape storage.
• Data Processing (Grid and/or Spider): 150 million CPU core hours and 0.28 million GPU hours, with 28,000 TB of online storage and 14,000 TB of Grid offline tape storage.
• 200 TB of Research Drive storage.

Access to LUMI is best suited for projects that require computations at an extreme scale, specifically GPU accelerated workloads on tens to hundreds of GPUs. You can apply for LUMI access if your project requires computations at such scale, or if it targets other LUMI-specific hardware. 

A LUMI-regular application should be preceded by a LUMI pilot or another LUMI project to make sure that your code runs successfully on AMD hardware and scales well to large numbers of GPUs. 

Several universities also offer high performance computing facilities with powerful servers on which researchers, PhDs and students connected to their university can interactively work on data and run computations. These are particularly of interest if you need relatively small amounts of capacity, or if you need HPC capacity for educational purposes.  See the high performance compute facilities at universities page for a complete overview.  

Also some institutions purchase capacity directly from SURF, which their researchers and students can use. See the direct access to compute services page for an overview.

Each applicant may submit one application per type of service, per calendar year.

Example 1: in 2024, a researcher can apply for 50,000 core hours on HPC Cloud and 500,000 core hours on Snellius.

Example 2: if a researcher works for 2 research projects, both of which require HPC Cloud, this researcher can only submit a pilot application for HPC Cloud for 1 of these projects. For the other project, someone else must submit the application.

As long as the application is within the limits set by the compute service, it may be raised.



Example: applicant has a pilot allocation of 350,000 core hours on Snellius. If it turns out that the applicant needs an additional 150,000 core hours during the period of use, then an increase is possible within the same allocation.

Applicants with a permanent contract can sign their compute time application themselves. Applicants with a temporary contract for at least the duration of the project and applicants whose tenure track position or permanent contract ends before the intended project completion date must have their application digitally signed by a responsible scientific supervisor with a permanent appointment. With this, the supervisor declares to guarantee the justification of allocated computing time, storage and support. This scientific supervisor may act as guarantor for several applicants.

Within 1 to 2 weeks after submitting your application, you can have access to our systems.

Yes, on the NWO website. NWO publishes all granted Large compute applications (NWO) via news items. At the bottom of these announcements there is an appendix with all Small compute applications (NWO) via SURF.

We are happy to help you. Please create a ticket on our Servicedesk portal and we will get in touch.

Yes, a one-time extension of max 6 months is possible. 

For LUMI a one-time extension of 3 months is possible.

Please create a ticket on our Servicedesk portal and we will get in touch.

There is more detailed information available on the website of NWO: Computing Time on National Computer Facilities and in the call text.