The state government, the federal government, the Johannes Gutenberg University Mainz and the Helmholtz Institute Mainz (HIM) have invested a total of 10.6 million euro in the new high-performance computer since 2016 in order to support the Rhineland-Palatinate scientists as part of the "Alliance for High Performance Computing Rhineland-Palatinate" (AHRP) till 2020 with computing power of the international top class.
This investment enabled the JGU to build a supercomputer of 1,876 individual nodes, of which 822 nodes each are equipped with two 10-core Broadwell processors and 1,046 nodes each are equipped with two 16-core Skylake processors from Intel. The more than 49,000 cores are paired with a 50Gbps Intel Omnipath network between the nodes and connected to a storage system with 7.5 petabytes of usable capacity. MOGON II is operated jointly at the JGU by the Center for Data Processing (ZDV) and the Helmholtz Institute Mainz (HIM).
"Advances in computer technology are enabling the new high-performance computer to consume only 40 percent more power than its predecessor system, while performing nearly seven times the peak performance with an input of 657 kW", explained Prof. Dr. med. André Brinkmann, Head of the JGU Data Processing Center. "With only slightly increasing maintenance costs, we can offer our scientists state-of-the-art high-performance computing of the highest international standards."
Natural and life science basic research today relies on the availability of sufficiently dimensioned computing resources. Users of MOGON II are working groups in physics, mathematics, computer science, biology, medicine, chemistry and geosciences. The issues studied with MOGON II include the simulation of the structure of matter and antimatter, as well as the development of new materials, the improvement of cancer therapies and our understanding of evolution, and the development of more precise weather and climate models.
The Cluster of Excellence PRISMA, the Graduate School of Excellence MAINZ and the majority of special research areas at the JGU can only achieve a refinement of their models and thus an improved understanding of our world view through more and more computing power. In the context of research at the Cluster of Excellence PRISMA, the physicists of the HIM will use the computer for the simulation of particle collisions as well as for the calculation of masses and structural properties of hadrons (= binding states of quarks and gluons). Both activities are also closely related to the construction and commissioning of the FAIR accelerator complex in Darmstadt.
"The considerable increase in computing capacity through the commissioning of the second stage of MOGON II makes Johannes Gutenberg University Mainz one of the leading national centres for the investigation of strong interaction", stated the spokesman of the Cluster of Excellence PRISMA, Prof. Dr. Hartmut Wittig.
The new high-performance computer is not only available to JGU researchers, but is also part of the Alliance for High Performance Computing Rhineland-Palatinate (AHRP). The German-wide networking takes place via the membership of the ZDV in the Gauss-Allianz, the association of the German high performance computer centres. The AHRP pools the country's computing capacity to make it available to all of the country's universities and research institutions, and provides training and advice to users of high-performance computing. The commissioning of the new high-performance computer and the integration into the AHRP are thus an essential contribution to the profile development of all universities in Rhineland-Palatinate and to the strengthening of competitiveness in the international environment.
The Gauss Alliance promotes the creation of conditions for the sustainable and efficient use of supercomputing resources of the highest performance classes in Germany, in particular by coordinating and pooling the complementary competencies of the participating data centres. Gauss Alliance's goal is to promote high-performance computing as a standalone strategic research activity with a focus on researching and developing strategies to improve the efficiency, applicability, and ease-of-use of high-performance computing.