HPC Projects at the RRZK
Current high-performance computing (HPC) projects at the RRZK cover many different disciplines.
German Human Genome-Phenome Archive (GHGA)
The aim of the GHGA is to optimize the procedures and processes for the storage and access to human data records. An infrastructure for an open handling of Omics-data is to be created, which at the same time ensures safe storage of personal data. The aim is to achieve the greatest possible technical and organisational simplification, while ensuring full compliance with the latest and highest data protection and security standards.
Researchers should be given access via a simple data access portal. This requires integrating and optimising a diverse and distributed data infrastructure with considerable data generators and scalable cloud technologies. In addition, communication with data access committees and the technical details for regulating distributed data access must be optimized.
Project participants:
- The German Cancer Research Center Heidelberg (Koordinierende Institution)
- Direktorium: Prof Dr. Oliver Stegle (Sprecher), Prof. Dr. Oliver Kohlbacher, Dr. Jan Korbel, Prof. Dr. Eva Winkler
- The German Cancer Research Center Heidelberg: Oliver Stegle, Benedikt Brors, Ivo Buchhalter, Daniel Hübschmann, Martin Lablans, Peter Lichter
- Eberhard-Karls-Universität Tübingen: Oliver Kohlbacher, Jens Krüger, Sven Nahnsen, Thomas Walter
- University Hospital Tübingen: Holm Graessner, Nisar Malek, Stephan Ossowski, Olaf Rieß
- Charité - Universitätsmedizin Berlin: Michael Hummel, Thorsten Schlomm
- Technische Universität München: Julien Gagneur, Thomas Meitinger, Juliane Winkelmann
- Europäisches Laboratorium für Molekularbiologie, Heidelberg: Oliver Stegle, Peer Bork, Wolfgang Huber, Jan Korbel
- Max Delbrück Center for Molecular Medicine, Berlin: Dieter Beule, Uwe Ohler
- Technische Universität Dresden: Andreas Dahl, Wolfgang E. Nagel
- University Hospital Heidelberg: Dirk Jäger, Julio Saez-Rodriguez, Christoph Schickhardt, Eva Winkler
- Heidelberger Akademie der Wissenschaften, Heidelberg: Fruzsina Molnár-Gábor
- Universität zu Köln: Viktor Achter, Ulrich Lang, Susanne Motameny
- Universitätsklinikum Schleswig-Holstein, Kiel: Philip Rosenstiel
- Helmholtz Zentrum München: Thomas Meitinger, Annette Peters, Juliane Winkelmann
- Deutsches Zentrum für Neurodegenerative Erkrankungen e.V., Bonn: Joachim Schultze
- Universität des Saarlandes, Saarbrücken: Jörn Walter
- German National Cohort, Heidelberg
Project duration: since October 2020
HPC.NRW
With the North Rhine-Westphalian Competence Network for High-Performance Computing (HPC.NRW), the organisational and also personnel requirements are being created for a competent and thematically wide-ranging contact and advisory centre for scientific HPC users in NRW. The know-how of the larger tier 2 HPC centres is combined with the advisory services of the lower tier (tier 3) centres. A powerful HPC advisory network is being established. In this way, the basic services for locations without level 3 centres and smaller ‘universities of applied sciences’ is to be ensured. The aim is to establish a fair, uniform and structured access for HPC users from all universities throughout NRW.
Equally important points of the realisation of the ‚HPC-Landeskonzeptes‘ are structure, efficiency and transparency. A structured provision of HPC resources supports the user in the selection of the appropriate level, the cluster and can, if necessary, provide assistance in submitting computing time applications. The state wide networking and the exchange between the locations promote an efficient and effective use of the computing power and help to improve the quality of operations. A transparent presentation of the available resources and joint coordination are necessary for the establishment of standardised and science-led application and approval procedures for all higher education institutions in NRW.
Further information under HPC-Landeskonzept der Hochschulen in NRW (german)
Contact: Viktor Achter
The HPC.NRW consortium partners:
- RWTH Aachen University, IT Center (Coordination)
- Universität Bielefeld, Bielefelder IT Servicezentrum
- Ruhr-Universität Bochum, IT Services
- Rheinische Friedrich-Wilhelms-Universität Bonn, Hochschulrechenzentrum
- Technische Universität Dortmund, IT und Medienzentrum
- Heinrich-Heine Universität Düsseldorf, Zentrum für Informations- und Medientechnologie
- Universität Duisburg-Essen, Zentrum für Informations- und Mediendienste
- Universität zu Köln, Regionales Rechenzentrum
- Westfälische Wilhelms-Universität Münster, Zentrum für Informationsverarbeitung
- Universität Paderborn, Hochleistungs-IT-Systeme
- Universität Siegen, Simulationstechnik und wissenschaftliches Rechnen
- Bergische Universität Wuppertal, Zentrum für Informations- und Medienverarbeitung
ORKA-HPC: OpenMP for reconfigurable heterogeneous architectures
Support-Code: 01lH17003A
The ORKA-HPC project deals with heterogeneous HPC architectures using FPGA accelerators. The reconfigurable nature of FPGAs enables highly efficient implementations of selected algorithmic kernels of heterogeneous application classes. As programming FPGAs is a highly specialised area and very time-consuming, they are rarely used in current HPC applications. Recent developments address better coupling between the FPGA subsystem and host. ORKA-HPC intends to make FPGAs accessible to software developers in the HPC domain and to significantly reduce the effort of porting software to FPGAs.
This includes utilizing FGPAs productively via OpenMP, which has proven to be the most strongly adapted shared memory programming model to date. Previous OpenMP implementation attempts have focused on specific aspects and do not address the performance aspects relevant in the HPC field. For an integrated approach, research into new ways of representing program code and heuristic methods for optimizing inherently parallel architectures is required.
Contact: Viktor Achter
Involved in the project:
- University of Cologne: RRZK - Prof. Dr. Ulrich Lang (Projektkoordination)
- Fraunhofer-Institut für Algorithmen und wissenschaftliches Rechnen (SCAI): Horst Schwichtenberg
- Friedrich Alexander Universität Erlangen: Lehrstuhl für Programmiersysteme - Prof. Dr. Michael Philippsen
- Konrad-Zuse-Zentrum für Informationstechnik Berlin (ZIB): Dr. Thomas Steinke
- Assoziierte Partner
- Intel GmbH
Project duration: 01.11.2017-31.07.2021
Contact
If you have any questions or problems, please contact the RRZK-Helpdesk