SDU opens call for HPC resources

Researchers at SDU can now apply for so called “large projects” for HPC resources available from 1st of July to 31st of December 2024.

The deadline for applications is 14th of April 2024 at 23:59.

Large projects are projects that require more than 50.000 CPU-hours, 1000 GPU-hours and/or 20TB of data. Remember that a requirement for submitting an application for a large project at SDU is that the same application has been submitted to the national DeiC call. In order to ensure local resources for researchers, who did not receive a national grant, applications to SDU will be withdrawn in case the equivalent application receives a DeiC national grant.

You can find out how to apply for the national DeiC grant via the DeiC website (NB: application deadline for the current call is 12th of March 2024 at 12:00)

NB For the local SDU call, you are welcome to apply for less than 50.000 CPU-hours, 1000 GPU-hours and/or 20TB of data – in fact you can submit this type of application (a so called regular project) at any time. Find out more about how to apply for a regular project here.

NB There is currently local SDU resources available for the spring period of 2024. Find out more here.

Available resources

For the current, local SDU call, SDU researchers can apply for resources on the following national HPC systems:

Type1 – DeiC Interactive HPC

The type 1 system is mainly focused on interactive computing and easy access for users. The system is made of the YouGene cluster hosted at SDU (see cluster specs here) and the OpenStack cluster at AAU. SDU researchers can access the cluster resources via their UCloud account.

Resources available to SDU for the second half of 2024: 1,000,000 CPU core hours and 100,000 GPU core hours.

Type 2 – DeiC Throughput HPC

This type of HPC system typically has a large number of cores which can be a mix between cost-effective and calculation-efficient units. Type 2 also has the ability to handle large amounts of data and its main focus is on high-throughput performance.

There are three type 2 HPC systems available at national level for SDU researchers:

The hardware accessible to SDU users for a type 2 HPC project is shown in the table below.

Compurerome 2.0	GenomeDK	Sophia
Xeon Gold 6230 Cascade Lake, 40 cores, 192 GB	AMD EPYC Rome 7452, 64 cores, 512 GB	AMD EPYC 7351, 32 cores, 128 GB

Resources available to SDU for the second half of 2024: 1,200,000 CPU core hours.

Type 3 – DeiC Large Memory HPC

This type of HPC system focuses on problem solving, with a structure that cannot be easily or efficiently distributed between many computer nodes. This is a type of system that is characterized by typically relatively few cores with access to a large globally addressable memory area.

Type 3 is hosted and maintained at SDU. For the cluster specs check here. The user guide can be found at this link.

Resources available to SDU for the second half of 2024: 1,700,000 core hours.

LUMI Capability HPC

LUMI is an abbreviation for “Large Unified Modern Infrastructure”. LUMI is one of the three European pre-exascale supercomputers part of the EuroHPC project and located in CSC’s data center in Kajaani, Finland.

Denmark participates in the consortium behind the LUMI supercomputer. Part of the LUMI machine therefore belongs exclusively to Denmark and is, in this sense, considered a national HPC resource.

For more information on LUMI, check the official documentation here.

CPU resources available to SDU for the second half of 2024: 1,050,000 CPU core hours.

GPU resources available to SDU for the second half of 2024: 69,000 GPU hours.

How to apply

For large projects, SDU researchers must send both:

A filled out application form for a large project, where they specify the amount of resources asked for. Note that the amount of resources asked for will be different from the national call. Only one form is needed even if the applicant is asking for compute time on more than one system.
The applications (one or more) they used for the DeiC national call. A copy of an application to DeiC can be sent from e-grant to your email (see the e-grant guide available on the DeiC website). Note that for every machine type selected in the application to SDU, there should be a corresponding DeiC application for the same machine type.

Application form for large project

DeiC application via e-grant

Applications for large projects must be submitted via the eScience Center’s service desk.

If you need help with/advice on how to write your application, please contact the representative from your faculty in the SDU eScience Center Operational Board: