Compute: information note
Updated 6 March 2023
© Crown copyright 2023
This publication is licensed under the terms of the Open Government Licence v3.0 except where otherwise stated. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3 or write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email: psi@nationalarchives.gov.uk.
Where we have identified any third party copyright information you will need to obtain permission from the copyright holders concerned.
This publication is available at https://www.gov.uk/government/publications/future-of-compute-review/compute-information-note
The independent Future of Compute Review, led by Professor Zoubin Ghahramani, was published on 6 March 2023. The review outlines 10 recommendations to harness the power of compute to boost economic growth and to address society’s greatest challenges.
The government is committed to supporting a thriving compute ecosystem that keeps the UK at the forefront of science, technology and innovation. It is taking action in response to the recommendations from the review and has announced investments in a UK public exascale compute capability and an artificial intelligence (AI) research compute resource.
The government will consider the remaining recommendations in the review carefully and respond in due course.
Compute means the systems assembled at scale to tackle computational tasks beyond the capabilities of everyday computers. This includes both physical supercomputers and the use of cloud provision to tackle high computational loads.
Exascale capability
Compute is integral to the running of the modern economy and science and technology (S&T) ecosystem. Our research and development (R&D) capability, through the application of simulation, modelling, data analysis and AI, hinges upon the effective use of compute. This government is fully aware of the need to invest in next-generation compute to ensure that we can remain a world-renowned destination for R&D investment and S&T talent. The announcement of a new national exascale system responds directly to the Future of Compute Review’s recommendation to ‘make immediate investments in the pathway to public exascale capability’.
An exascale system can perform one exaflop per second, or a billion billion floating point operations. A national exascale facility will form the cornerstone of the UK’s next-generation compute infrastructure, unlocking major advances across S&T and deep tech. This includes the critical technologies that have been prioritised by this government through the Science & Technology framework, which sets out an ambitious ten-point plan for achieving science and technology superpower status by 2030.
Current generation top-end systems have taken us far, but can only do so much. An exascale system will deliver advancements in patient-specific drug treatments and our wider drug development capability. Exascale capacity will be crucial to more accurately simulating the workings of fusion reactors, allowing us to design more effective reactors which can help us to engineer a future with a potentially limitless source of low-carbon energy.
The announcement underlines this government’s commitment to the UK as a world leader in S&T, and emphasises that we are working to ensure that the UK is well-positioned to remain so going forward. International counterparts such as the EU and Japan have already made significant progress towards exascale capability, and we have seen from the US’ operational Frontier system that the exascale generation is already commencing. The government is acting quickly to ensure that our world-leading R&D community has the tools it needs to remain competitive with the world’s best and brightest.
Artificial intelligence research resource
AI model sizes have exploded in recent years, best exemplified by GPT-3 which has up to 175 million parameters. Despite this, we have seen consistently impressive results from assigning ever-increasing quantities of accelerated compute to large-scale model training, delivering improvements to self-driving car capability and better understanding the causes of rare genetic diseases. This government recognises the growing and specialised compute needs of the AI research community. Existing national and regional systems have served us well, but we understand that more specialised hardware-accelerated systems will form a crucial component of the UK’s future compute landscape.
We are taking the appropriate steps to ensure that our world-leading AI sector has a dedicated public compute resource for both academic and promising commercial research, ensuring that the UK remains an attractive destination for talent, investment and innovation. The announcement of an AI research resource underlines this government’s commitment to supporting deep tech investment, R&D and innovation. This responds directly to the Future of Compute Review’s recommendation to ‘immediately and significantly increase compute capacity for AI research’.
The resource will provide hardware-accelerated compute capacity at a scale that is usable for large-scale model training, as well as being available for exploratory research at the smaller scale.