A consortium of eight UK universities, led by the University of Oxford, has been awarded £3 million by the Engineering and Physical Sciences Research Council (EPSRC) to establish a national high-performance computing facility to support machine learning.
The new facility, known as the Joint Academic Data Science Endeavour (JADE), forms part of a combined investment of £20m by EPSRC in the UK’s regional Tier 2 high-performance computing facilities, which aim to bridge the gap between institutional and national resources.
JADE, which will be the largest Graphics Processing Unit (GPU) facility in the UK, will provide a computational hub to support the research of the world-leading groups in machine learning at the universities of Oxford, Edinburgh, Sheffield, King’s College London, Queen Mary University of London and University College London (UCL). It will also provide a powerful resource for data science and molecular dynamics researchers at the universities of Bristol and Southampton.
Machine learning has experienced huge growth over the last five years, with applications including computer vision for driverless cars, language translation services and medical imaging. JADE is the first national computing facility to support this rapid growth.
Professor Mike Giles of the Scientific Computing department at the Oxford Mathematical Institute, who is leading the project, said: ‘For the first time, JADE will provide very significant national computing facilities addressing the particular needs of machine learning, one of the fastest growing areas of academic research and industrial application.’
JADE will be delivered through a partnership between Atos, who will provide and integrate the system hardware, and STFC’s Hartree Centre, who will host and support the system for the three-year initial duration of the facility. Exploiting the capabilities of the NVIDIA DGX-1 Deep Learning System, JADE will comprise 22 of these servers, each containing 8 of the newest NVIDIA Tesla P100 GPUs linked by NVIDIA’s NVlink interconnect technology.
‘For the first time, JADE will provide very significant national computing facilities addressing the particular needs of machine learning, one of the fastest growing areas of academic research and industrial application.’
Professor Mike Giles, Scientific Computing department at the Mathematical Institute, Oxford University
To support researchers using the system, five software engineering posts are being created by Oxford, KCL, QMUL, Southampton and UCL. This is a key investment to ensure the necessary expertise is in place to derive maximum benefit from the new facility.
Speaking on JADE’s potential research impact, Professor Philip Nelson, EPSRC’s Chief Executive, said: ‘These centres will enable new discoveries, drive innovation and allow new insights into today’s scientific challenges. They are important because they address an existing gulf in capability between local university systems and the UK National Supercomputing Service ARCHER. Many universities are involved in the six new centres, and these will give more researchers easy access to High Performance Computing.’