Atos is to upgrade the Berzelius supercomputer in Linköping, Sweden.
The IT infrastructure provider's Eviden business unit has been awarded a four-year multi-million euro contract to boost the performance of the Berzelius supercomputer at Linkoping University for the second time.
Eviden will integrate and deploy further Nvidia GPUs into the system, adding an extra 512 petaflops of FP8 AI performance to the Berzelius.
Berzelius will see 16 Nvidia DGX H200 systems added, with 128 Nvidia H200 Tensor Core GPUs to the Nvidia DGX SuperPOD, bringing its entire architecture up to 110 nodes and 880 GPUs. The Eviden team will also integrate a VAST Data Platform into the cluster.
Dr. Cédric Bourrasset, head of HPC-AI and quantum computing, Eviden, Atos Group said: “We are honored by this new mark of trust from the Linkoping University, which relies once again on our teams to deliver a performance boost. Berzelius has been a key system in Europe to boost AI research since 2020 and this new upgrade will allow researchers to benefit from an expanded GPU architecture, helping to widen the possibilities of scientific breakthroughs.”
Housed in the National Supercomputer Centre (NSC) at the university in Linköping, the machine was first installed in 2021. Funded by the Knut and Alice Wallenberg Foundation, the system was previously upgraded in 2023. Plans for this latest upgrade were announced earlier in the summer.
Prof. Anders Ynnerman, WASP program director and professor of scientific visualization, Linkoping University added: “In the fast-moving area of AI it is crucial for researchers to have access to state-of-the-art resources for machine learning integrated in the national infrastructure for supercomputing. I am very impressed by the speed at which this upgrade of Berzelius has materialized and am looking forward to research results it will enable.”
In addition to the Berzelius system, the NSC hosts and operates the 3 petaflops Tetralith; the Sigma HPC cluster; Cirrus, Stratus, and Nebula which are all used for operational weather forecasting; the 326 teraflops Bi; and Freja, which is primarily used for climate, meteorology and oceanography.
In July 2023, it was announced that the NSC would house and operate the 30 petaflops Arrhenius supercomputer, co-funded by the European High-Performance Computing Joint Undertaking and the Swedish Research Council’s funding for research infrastructure.