Event round-up: Future of compute campaign launch - how can we make the UK a leader in compute?
Large-scale computing is a key enabler of science, research and innovation and sits at the heart of value chains that extend deep into the economy. LSC is essential to R&D in a range of sectors, including healthcare and drug discovery, engineering and manufacturing, financial services and risk analysis, particle physics, materials science and more.
On 28 September 2022, techUK launched a new supercomputing campaign - Is the UK ready for the Future of Compute? - deep-diving into the UK’s current compute ecosystem to look at where the UK can position itself for success, where we are falling behind, and what government and industry can do to make the UK a leader in science and technology.
The panel included:
- Dave Snelling, Director Artificial Intelligence, Fujitsu
- Alison Kennedy, Strategic Advisor, STFC
- Vasilis Kapsalis, EMEA Business Leader, SambaNova Systems
- Hannah Cooke, Co-Chair, Future of Compute Review Secretariat, DCMS
- (Chair) Sue Daley, Director of Tech & Innovation, techUK
You can watch the full session here, or read our summary of the discussion below:
Please note that the below is a summary of the event, and readers are encouraged to watch the webinar to understand the full details of the discussion.
Following opening remarks, the panel discussed the importance of compute at scale to the UK’s economy, and Dave Snelling made the case for a long-term approach - compute may be vital for scientific research but it isn’t a quick win that you can pour on the economy for instant growth.
Vasilis Kapsalis argued that compute also has a role to play beyond early-stage science, for example in training natural language processing (NLP) models, that could have immediate benefits in sectors like education and healthcare.
The panel discussed use-cases for compute and the need for a heterogeneous ecosystem that can deliver different types of architecture for different problems, including hybrid systems harnessing AI, classical HPC, and eventually even quantum.
Planning for the future
The discussion turned to the importance of taking a long-term view and putting in place the building blocks for a thriving and self-sustaining ecosystem, particularly when it comes to skills and software.
Alison highlighted the ongoing need to update and evolve our HPC software to get maximum value from the latest infrastructure, including by embracing international collaboration. Skills and software will be key to UK success over the next 10 years.
Convergence with other technologies
The panel noted the convergence of large-scale computing with emerging technologies like quantum in hybrid classical-quantum systems.
Vasilis made the case for compute at scale being key to delivering the most modern AI capabilities, often using systems designed specifically for training AI models, and the opportunity to build new AI-enabled industries to drive economic growth. A wide-ranging discussion followed about some of the ethical, legal and governance challenges of compute and AI scaling beyond our ability to verify the results of complex models.
Hannah highlighted the role of compute as a fundamental enabler of other technologies and explained how the DCMS future of compute review relates to other work in government like the AI Strategy, Quantum Strategy, Integrated Review and National Data Strategy.
International collaboration
The panel explored how the UK could position for success in a global compute market and the opportunities for international collaboration.
Hannah reminded the panel that one ambition of the DCMS review is ensuring that the UK’s domestic compute ecosystem is embedded within the larger global ecosystem and equipped to maintain and grow the UK’s strengths in innovation, science and technology. To deliver infrastructure, supply chains, and academic and industry collaboration, we need to work internationally - partnerships are a crucial component that can deliver huge benefits.
Dave suggested that if the UK wants to build its own high-end system, then partnering with a well-aligned country like Japan would make that much easier, but the harder part of the puzzle is the surrounding ecosystem - including software, application design, integration - and that all sits above building the hardware.
Vasilis agreed that we need to pick partners we are aligned with and think about how compute partnerships relate to broader collaboration on science and technology within trade deals.
Future of compute review
Hannah talked about the DMCS future of compute review and how the discussion needs to move towards policy decisions Government can take to support the compute ecosystem.
Sue asked the panel to close with final comments on key interventions they would like to see from the Government in response to the review:
Vasilis - “Investment in mapping compute architectures to different classes of problems - and then linking that to economic benefits”
Alison - “More investment in training - not just PhDs but also a much wider range of people that we need to contribute to these efforts"
Dave - “Something to create stability in the market for intellectual property and start-ups so the UK does not lose skills and IP generated here.”
Future of Compute Hub
techUK is exploring with members how we can develop our compute infrastructure, safeguard the legacy of Turing and Lovelace, and position the UK for success.