Forschungszentrum Jülich will combine this new blade into its booster module, extending its current JUWELS BullSequana supercomputer, making it the primary system worldwide the use this new expertise. The JUWELS Booster will present researchers throughout Europe with considerably elevated computational sources. Some of the initiatives it would gas are the European Commission’s Human Brain Project and the Jülich Laboratories of “Climate Science” and “Molecular Systems”. Once absolutely deployed this summer season the upgraded supercomputing system, operated below ParTec’s software program ParaStation Modulo, is predicted to supply a computational peak efficiency of greater than 70 Petaflops/s making it essentially the most highly effective supercomputer in Europe and a showcase for European exascale structure.
By integrating one of the best applied sciences out there, together with this newest blade from Atos which incorporates NVIDIA’s highly effective A100 GPU into our supercomputer, we’re capable of increase our supercomputing energy. This is a vital milestone on our path to succeed in our final goal of offering exascale capabilities to science and business” mentioned Prof. Thomas Lippert, Director of the Jülich Supercomputing Centre (JSC).
“We’re proud to be one of many first supercomputer producers worldwide to incorporate NVIDIA’s A100 GPU and to have the ability to supply researchers worldwide the very best software efficiency for his or her HPC and AI workloads” said Agnès Boudot, Senior Vice President, Head of HPC & Quantum at Atos. “We’re actually excited to be working with Jülich to spice up its computing energy and to allow its JUWELS system to develop into essentially the most highly effective supercomputer in Europe.”
“Atos is likely one of the high gamers world-wide within the excessive finish of the server market. It is the main European-based vendor of HPC techniques, largely as a result of efficiency and energy of its BullSequana techniques, its robust technological experience and innovation and its means to successfully handle large-scale initiatives globally. It has doubled its presence within the TOP500 rating over the past 5 years and its server market share worldwide continues to develop. We see Atos as one of many key supercomputing producers that may seemingly profit over the subsequent few years as demand for exascale techniques turns into extra important.” mentioned Steve Conway, Senior Advisor at Hyperion Research.
The new blade options an NVIDIA HGX-A100 base-board outfitted with 4 NVIDIA A100 GPUs linked through third-generation NVIDIA NVLink expertise, alongside two AMD EPYC CPUs and as much as 4 NVIDIA Mellanox InfiniBand ports linked through a Dragonfly+ configuration. The NVIDIA Ampere structure is a groundbreaking engineering achievement that delivers an enormous efficiency enchancment over the earlier era for AI coaching and inference. The NVIDIA A100 Tensor Core GPU options the world’s largest 7-nanometer processor, with greater than 54 billion transistors. This crew of applied sciences, together with the NVIDIA market-leading ecosystem of over 700 GPU-accelerated HPC purposes and help from NGC, will enable researchers to hit the bottom operating with containerized software program optimized for the event and deployment of GPU-accelerated HPC and AI initiatives.
The BullSequana X2415 might be out there in Q2 of 2020.
