ParTec AG, one of the leading SMEs in the HPC domain in Europe and the specialist for modular supercomputing, develops and supports a comprehensive suite of HPC systems’ management tools and a runtime environment specifically tuned for the largest distributed-memory supercomputers in existence today and beyond. ParTec’s unrivaled expertise in developing HPC software includes the self-developed ParaStation MPI and tools like TicketSuite and HealthChecker. In conjunction with professional services, consultancy, and support, ParTec was elected as the partner of choice in some of the leading HPC sites across Europe. For more than two decades, ParTec has been a strong general-purpose HPC specialist in the German and EMEA HPC market.
ParTec’s ParaStation Modulo is extensively used in production environments, e.g., on the JURECA Cluster and Booster system with 6.56 PFlop/s peak performance and ranked #52 in the June 2019 TOP500 list, both operated by the Jülich Supercomputing Centre (JSC). The existing JUWELS Cluster system with a 9.89 PFlop/s peak performance, run by the Jülich Supercomputing Centre (JSC) and ranked #30 in the June 2019 TOP500 list has just been extended by a Booster. The peak performance for the Booster is around 71 PFlop/s. It got ranked #7 on the November 2020 TOP500 list, making it #1 in Europe. Besides the mentioned large systems, ParTec also supports some experimental systems. ParTec is the chosen partner for the co-design and the support of the JUROPA family, HDF-ML and ICCP-ICEI systems, likewise operated by the Jülich Supercomputing Centre.
ParTec participates in nationally funded research projects as well as in EU-funded Exascale projects like the DEEP and SEA families of projects. Furthermore, ParTec participates in the CoE RAISE project, which is its first participation in a Centre of Excellence. The Modular Supercomputing Architecture (MSA), which plays a central role in these projects, is enabled by ParTec’s ParaStation Modulo Software Suite and further developed in the aforementioned Exascale projects. The MSA approach, first demonstrated on the DEEP-EST prototype, has been shown to work successfully also on large-scale production systems like the JURECA and JUWELS Cluster-Booster systems, and was also chosen for the Luxembourg MeluXina system expected to go into production in H1/2021. In addition, ParTec (as a member of the project management team) internally supports the coordinator FZJ in the DEEP Projects, the DEEP-SEA project and the RAISE project at management level.
In the RED-SEA project ParTec participates in two work packages. In WP1 Architecture, Co-design, and Performance ParTec supports the project in creating a basis of co-design activities by providing its own software information. WP4 Endpoint Functions and Reliability includes the optimization of MPI and ParaStation MPI libraries and thus an extension by BXI support to enable the utilization of BXI networks.
