The CLAIX-2023 expansion stage was put into operation at the end of 2023. This will offer the latest technologies in a highly efficient water-cooled infrastructure with Intel Xeon Sapphire Rapids processors, DDR5 memory, NVIDIA H100 accelerators and an NDR InfiniBand network.
Over 80% of CLAIX's capacity is called upon by applications in engineering and the natural sciences (as of May 2023). It is in line with these applications that the architecture of CLAIX has been designed and the call for proposals has been shaped. CLAIX-2023 will also support applications from the field of artificial intelligence and machine learning, respectively, which can also be coupled with classical simulations. From traditional SSH-based access to interactive use in the browser, CLAIX supports different access paths and usage scenarios and offers a variety of software packages. Accordingly, suppoLinkrt services are available for all CLAIX users.
The CLAIX-2023 expansion stage consists of shares of the Tier-2 supply in NHR, the Tier-3 supply of RWTH Aachen University and the state of NRW, and the WestAI project.
- With NHR4CES RWTH is a member of the national high-performance computing network, which is jointly funded by the federal and state governments. CLAIX is thus available as a national Tier-2 system via NHR to all scientists at German universities free of charge after a science-led review.
- The Tier-3 portion, financed by the federal state of NRW, primarily supports teaching and small-scale computing projects at RWTH Aachen University and for engineers in NRW. All employees and students at RWTH Aachen University also have access to a basic computing time quota.
- RWTH operates part of the infrastructure of the AI Service Center WestAI, which is funded by the BMBF, and this part is integrated into CLAIX. The resulting computing time contingent is available for all activities in the project.
Data Storage
Various file systems are provided for storing the data, which differ in terms of the intended usage scenarios. These differences reveal themselves in the form of performance in various metrics, the available space, and the data backup concepts. The spectrum ranges from highly available file systems with disaster-proof backup for critical data to the possibility of storing petabyte-sized data volumes and access speeds of over 100 gigabytes/sec.
The technical description explains details and architecture of the data storage.