Members of the EU project “CoE RAISE” – developing artificial intelligence approaches for the next generation of supercomputers – meet at CERN

Final week, members of the European Union CoE RAISE venture They gathered at CERN for an ‘all fingers’ assembly. This revolutionary venture is creating synthetic intelligence (AI) approaches for the subsequent technology of “exascale” supercomputers, to be used in each science and business. Use instances explored by means of the venture embrace optimizing wind farm layouts, designing environment friendly plane, optimizing acoustic engineering, seismic imaging utilizing distant sensing, and extra.

CoE RAISE – European Heart of Excellence in Exascale Computing “Analysis in Synthetic Intelligence and Simulation-Primarily based Engineering at Exascale” – is funded below the European Union’s Horizon 2020 Analysis and Innovation programme. The venture was launched in 2021 and can final for 3 years.

The four-day assembly, which was held within the CERN boardroom, was attended by 54 venture members. Individuals mentioned the progress of their work to develop AI applied sciences for complicated purposes in Europe operating on future high-performance computing (HPC) methods. Exascale refers back to the subsequent technology of high-performance computer systems that may carry out greater than 10 machines18 Floating level operations per second (FLOPS). Immediately solely Frontier supercomputer At Oak Ridge Nationwide Laboratory in the USA it reached this degree. Nevertheless, with extra exascale HPC methods on the horizon, it is very important make sure that AI approaches utilized in science and business are prepared to completely exploit the massive potential. In June, the European Joint Enterprise for Excessive-Efficiency Computing (EuroHPC JU) introduced that Germany’s Forschungszentrum Jülich GmbH It was chosen to host and function Europe’s first exascale supercomputerwhich is about to return on-line subsequent yr and will likely be often called JUPITER (Joint Enterprise Chief for Modern and Transformational Exascale Analysis).

CoE RAISE develops revolutionary approaches to AI on heterogeneous HPC architectures that embrace a number of processor varieties. Such architectures can present increased efficiency and vitality effectivity, however the code should be tailored to effectively use several types of processors. The unreal intelligence strategies being developed focus round it 9 main use instances Nicely extensible and designed to run on exascale HPC methods.

CoE RAISE helps know-how switch to business, significantly small and medium-sized companies, in addition to manages schooling and coaching initiatives. Moreover, CoE RAISE additionally offers consultancy and communicates with different European initiatives to maximise synergies, exploit alternatives for co-design and data sharing. All features of the venture’s work had been mentioned over the 4 days at CERN.

CERN can also be a associate and is offering a use case to the venture. This work focuses on bettering strategies for reconstructing particle collision occasions on the Excessive Luminosity Giant Hadron Collider (HL-LHC), which is scheduled to start operation in 2029. The HL-LHC will see extra particle collisions than ever earlier than. Whereas, exabytes of knowledge are produced yearly, leading to unprecedented computing challenges. To reconstruct particle collision occasions right this moment (with datasets on the order of terabytes or petabytes), lots of of various algorithms are working concurrently: some are conventional algorithms optimized for particular {hardware} configurations, whereas others already incorporate AI-driven strategies, resembling deep neural networks ( DNNs). Mission crew members at CERN are working to extend the modularity of the methods and make sure that the code is optimized to take full benefit of heterogeneous architectures, in addition to improve using machine studying and different AI strategies to reconstruct collisions and classify particles.

“Supercomputers are reaching exascale and enabling the supply of an unprecedented scale of processing assets to HPC and AI workflows,” says Maria Geron, Chief Expertise Officer of CERN openlab, who’s main CERN’s contribution to the venture. “The analysis carried out at CoE RAISE will result in the joint design of HPC computing assets for future AI and HPC purposes for each science and business. This assembly enabled us to alternate concepts, develop them, and introduce new frontiers. It additionally offered researchers from different fields with distinctive perception into the setting and challenges going through CERN, which promotes cross-fertilization and understanding.”

Leave a Comment