The new high-performance computer CLAIX-2023 at the IT Center of RWTH Aachen University is an expansion stage of the CLAIX-2018 and brings with it a significant increase in performance. With powerful Intel Xeon 8468 Sapphire Rapids CPUs with a total of 96 cores per compute node and 52 special servers for artificial intelligence and machine learning, each with four NVIDIA H100 GPUs, the system achieves an impressive total performance of over 14 PFLOPS in the ML segment. The total of 632 directly water-cooled computing nodes for classic high-performance computing (HPC) are characterized by a significant increase in performance and sustainability. The different memory expansion of the nodes enables precise use while optimizing costs at the same time. The HPC segment achieves a peak performance of around 4 PFLOPs and up to 530 million core hours are allocated each year. In order to keep pace with current developments in the field of artificial intelligence and machine learning, 52 servers were procured specifically for these applications. These computing nodes are equipped with four NVIDIA H100 GPUs and 96 GB HBM2e memory per GPU, which enables the calculation of large ML-based models. The ML segment achieves a total performance of over 14 PFLOPS. Many compute nodes are used in parallel to simulate large models, which is why the system is equipped with a fast NDR Infiniband RDMA network.
A large part of CLAIX-2023 is provided by NHR4CES and is used in particular for engineering sciences, especially computational fluid dynamics.
For more information, please refer to the article in the IT Center Blog. Instructions on how to use the system can be found on IT Center Help.