Machine learning: Oxford University and Atos to install the UK’s largest supercomputer

0
Machine learning: Oxford University and Atos to install the UK’s largest supercomputer

Atos and the University of Oxford have signed a four-year, £5 million contract for a new state-of-the-art supercomputer. Dedicated to deep learning and based on the NVIDIA DGX SuperPODTM architecture, it will enable UK academics and industry to advance research in machine learning and artificial intelligence, as part of the JADE2 project (a continuation of the UK’s national JADE project).

The supercomputer, the largest artificial intelligence supercomputer in the UK with over 500 NVIDIA GPUs, is funded by the Engineering and Physical Sciences Research Council (EPSRC). It aims to build on the success of the Joint Academic Data Science Endeavour (JADE) project, a national computing resource that provides GPU-accelerated facilities for the world’s leading AI and machine learning experts, implemented by a consortium of eight UK universities and the Alan Turing Institute.

JADE2, based on NVIDIA DGX systems, uses the A3I DDN storage solution. It is expected to more than triple the capacity of the first JADE machine and will provide increased computing power to an expanded consortium of more than 20 universities and the Alan Turing Institute. It will help meet the demand for AI-based facilities that follows the success of JADE. JADE, led by Oxford University, with support from the Alan Turing Institute and 22 universities, is the national GPU system that supports multidisciplinary science with a focus on machine learning and molecular dynamics.

Professor Wes Armour of Oxford University said:

“The success of JADE has generated increased demand from local researchers and companies for powerful computing facilities capable of handling sophisticated and intensive AI workloads.

Building on the successful cooperation with Atos for JADE and significantly increasing the computing capacity of the consortium, Atos’ new supercomputer dedicated to deep learning will enable us to meet this demand and lead to important scientific discoveries. It will consolidate JADE’s status as a key national computing centre for university research in AI.

Alison Kennedy, Director of STFC Hartree, said:

“We look forward to hosting JADE2. The provision of increased computing capabilities will help develop and expand research computing skills in industry and academia. This is directly in line with the mission of the Hartree Centre and we are delighted to continue the collaboration with Atos and Oxford University which builds on the previous success of JADE. »

Agnès Boudot, SVP, Head of HPC & Quantum activities at Atos, added :

“We are proud to be working with Oxford University to deliver JADE2. The power of this new facility will provide researchers and industry with the necessary resources to remain at the forefront of research and innovation in machine learning and AI. We are confident that this high-performance system, combined with our expertise, will support the UK’s ambition to become a world leader in AI and machine learning, while helping it to meet the various challenges specific to these two fields.

The SuperPOD DGX system will comprise a 63-node DGX cluster with a total of 504 NVIDIA Tensor Core A100 GPUs, interconnected with the NVIDIA Mellanox InfiniBand Network and powered by the NMS AI400 storage – making it the largest system of its kind in the UK.

The project will be housed at STFC Hartree in Daresbury, near Warrington, UK.

Translated from Machine learning : L’Université d’Oxford et Atos installeront le plus grand super calculateur du Royaume-Uni