HPC ACCESS

Your Portal to Scientific Discovery 

GCS systems are available to national scientists and researchers from academia and industry through several different allocation programs. Researchers at German Universities and publicly funded research institutions are eligable to apply. Additionally, European researchers can access GCS resources through the Partnership for Advance Computing in Europe. Below, current and prospective users can find information on how to access GCS resources, including guidelines, important dates, and user obligations after recieving time on GCS resources.

Large-scale projects are characterised by projects that require a large amount of computing time over longer periods of time. In detail, projects are classified as “Large-Scale” if they require 2% of the systems’ annual production in terms of estimated availability. This requirement can be translated into different values. For the supercomputers Hawk and SuperMUC-NG we give the requirement in terms of core-hours. Thus, the projects’ minimal requirements should be at least 100 Mcore-h for Hawk or 45 Mcore-h on SuperMUC-NG. Please note that the architecture and sustainable performance of a core of each system may differ widely and that the “core-hours” of the systems are not comparable or interchangeable. 

JUWELS consists of two different modules with different hardware architectures. Thus, the minimal computing time requirements for GCS-LS projects cannot be coherently specified by use of core-hours but EFLOP is used instead. The projects’ minimal requirements summed over all requested modules should be at least 45000 EFLOP per year for JUWELS. EFLOP refer to the theoretical peak performance. Further information can be found under 6. in the fact sheet for JUWELS.

Large-scale projects go through a competitive review and resource allocation process established by the GCS. A "Call for Large-Scale Projects" is published by the Gauss Centre twice a year. Deadlines for calls are usually at the end of winter and at the end of summer of each year. An overview of the approved GCS large-scale projects is available here.

Projects which do not fall into the category “large scale” are called GCS regular projects. The peer-review process is implemented at the national level, carried out by the steering committees or allocation committees of the three GCS centres HLRS, JSC, and LRZ, respectively.

Applications for GCS regular projects on Hawk and SuperMUC-NG may be submitted at any time (so-called rolling calls), applications for GCS regular projects on JUWELS may be submitted twice a year at the same time as GCS large-scale projects.

GCS is one of the hosting members of the Partnership for Advanced Computing in Europe (PRACE), with all three GCS supercomputers listed as a Tier-0 resource at PRACE. GCS systems are thus also available to scientists and researchers residing in Europe. Computing time allocations are dispersed based on scientific criteria through independent reviewers in a peer-review process on the European level through PRACE. To apply for computing time, European scientists are invited to answer the PRACE calls for projects. For further information and for details on how to apply, please click here.

Applications for compute resources are evaluated only according to their scientific excellence.

• The project must be scientifically challenging as well as in the societal interest. 
• Applicants need to specifcy clear scientific goals and verifiable milestones in their proposals.
• Project implementation must be technically feasible on the available systems, and be in reasonable proportion to the performance of these systems. 
• The principal investigator must have a proven scientific track record, and he or she must be able to successfully accomplish the proposed tasks. Applicants must possess the specialized understanding for using high-end computing systems. Specifically, researchers need to prove that their application has thorough documentation of application performance on smaller computing systems or scaling studies.   
• Users should take advantage of the features unique to high-performance computing as they implement their respective programs. This will be checked regularly during the course of the project.

Dates for GCS Large-Scale Calls

The next call for GCS large-scale computing time proposals on Hawk, JUWELS and SuperMUC-NG will cover the period November 1, 2020 to October 31, 2021.

The call will open on  20 July 2020 and close on 17 August 2020, 17:00 CEST.

Dates for GCS Regular Calls

(A) Hawk and SuperMUC-NG:
Applications for GCS regular projects on the HLRS and LRZ HPC systems can be submitted at any time (so-calledrolling calls).

(B) JUWELS:
Applications for GCS regular projects on the JSC HPC system can be submitted twice a year at the same time as GCS large-scale projects (so-called cut-off calls – see dates above).

The application and reporting procedures slightly differ for the three supercomputers and location sites. Therefore, please carefully read the following additional information on “How to Apply” for the individual GCS HPC systems: 

Important Notice on How to Apply for Computing Time on Hawk
Important Notice on How to Apply for Computing Time on JUWELS
Important Notice on How to Apply for Computing Time on SuperMUC-NG

General Requirements for project applications:

Applications should be submitted in English.

Please structure the project application in the following way:

• Outline of the scientific challenge and your approach towards its solution (with references)
• Description of previous work in this field, exploratory studies including the experience and results obtained (with references)
• Statement of the scientific goals of the research project
• Description of the physical and mathematical methods employed in the project, including numerical algorithms
• Detailed schedule of the project, in the case of an extension of the project, please give details about any changes of your plans.
• Preliminary studies that show good scaling behaviour of the programs under production conditions (i.e. typical parameter sets and problem sizes of the planned project, including I/O).
• A detailed description of the I/O behaviour of the application (amount and size of files generated during typical runs, I/O strategy used (MPI I/O, netCDF, HDF5, SIONlib, etc.) and an estimate of the storage requirements during the project (scratch diskspace needed, amount of data which needs to be transferred during the project to and from the system).
• If the project is an extension of a previous project, please primarily describe how this project differs from the previous year's application. Please upload the previous application document as supplementary material.
• Please follow the form “Project Proposal” (link below) for a project description with respect to form, content and size.
   
•  Template: Project Proposal for Tier 0/Tier1 HPC Access at GCS (PDF, 300 kB)
•  Template: Project Proposal for Tier 0/Tier1 HPC Access at GCS (docx, 239 kB)
•  Template: Project Proposal for Tier 0/Tier1 HPC Access at GCS as a TeX file (FileTypex-tar, 408 kB)

Please make sure that all data is complete and double check it. 

The project description has to be uploaded via the application form as PDF file. Please do not include any supporting material in the project description. → If you wish to add supplemental material to the project description, please submit this information in the application form with a separate PDF file.

THE APPLICATION FORM CAN BE FOUND HERE

As the HPC technologies provided by GCS are openly accessible to all scientists and researchers, all research conducted by using the publicly funded HPC systems is aimed at servicing the “public interest”. As a consequence, results and findings achieved in the realm of these simulation projects need to be made publicly available, and certain reporting obligations do apply:

Status report

This report must be submitted after twelve months counting from the start date of the last allocation period of the project in case the project applies for a continuation or extension. The status report should not exceed a maximum of 10 pages and must be uploaded together with the application for a project extension/continuation as a separate file and should cover the last twelve months of the project.

Final report

The final report is due three months after the end of the GCS lage-scale project and one month after the end of the GCS regular project, respectively. The report must not exceed a maximum of 18 pages. It should focus on the scientific and technical outcome. Moreover, it should explain, how the granted computing time was spend within the project.

Both reports should include the following:

• The precise results which were obtained during the last computing period.
• Statements about the use of the computer in your project with respect to performance, scaling behaviour, main storage requirements and disk storage requirements.
• Publications that were published in peer reviewed journals during this period (no conference contributions or posters).
• Ph.D. theses which were completed in the realm of the project (please give name of the author and the title of the thesis).
• Attractive colour pictures which were created in your project to present your results, which can be interesting for a general public. We would ask your permission to use these pictures in GCS or JSC publications.
• Please follow the template for the status/final report with respect to form, content and size.

 Template for the Status/Final Report (PDF, 55 kB)
 Template for the Status/Final Report (docx, 25 kB)
 Template for the Status/Final Report as a Tex file (FileTypex-tar, 26 kB)

The final report has to be uploaded here.

Report for publication on the GCS website 

For all GCS computing time projects, a report for publication on the GCS website has to be submitted within three months of conclusion of the project. For on-going projects running more than two years an interim project report needs to be submitted within three months after the second project year. It should describe briefly the scientific and technical goals/challenges and results of the project achieved so far. This report for the GCS website aims at introducing the science and research activities to the general audience without a scientific background. For details about form and content see our “instructions for authors” page.

GCS Coordination Office
c/o Jülich Supercomputing Centre
Forschungszentrum Jülich
52425 Jülich
Germany

E-mail: coordination-office@gauss-centre.eu