Supercomputers that model the climate and weather are themselves consuming enough electricity to become part of the carbon problem they are meant to solve. The CONTINENTS project, a collaboration between the UK’s EPCC supercomputing centre and national climate research centres in both the UK and the US, aims to slash the energy footprint of high-performance computing (HPC) without sacrificing the scientific power needed for climate modelling. Data centres already account for the largest share of HPC’s operational carbon emissions. As national economies push toward net-zero targets by 2050, these facilities cannot remain exempt from the same efficiency demands placed on other sectors. The project will tackle the problem from multiple angles: optimising how data centres are run, making better use of specialised hardware, applying machine learning to reduce the computational load of data analysis and numerical modelling, and redesigning software to be inherently more energy-efficient. If successful, the research could transform how the UK and US operate their largest scientific computing resources. More efficient climate models would run faster and cheaper, improving forecasts and our understanding of climate change. The methods developed could also ripple into any field that relies on HPC—from drug discovery to materials science—quietly reducing the energy burden of the infrastructure that underpins modern research.
View original technical description
The UK's Net Zero strategy is for all sectors of the economy to meet the Net Zero target by 2050. Similar long-term aims to move towards Net Zero emissions on the same timescales exist in other countries, including the USA. High-performance computing (HPC) is not exempt from needing to adapt to these strategic challenges, and pushing HPC towards sustainability and Net Zero is crucial if the scientific community is to keep justifying the use and cost of large-scale HPC resources in the face of climate change. Electricity consumption of data centres and systems is by far the largest contributor to the carbon footprint of operational HPC and minimising the energy that is consumed, and reducing/reusing waste, is therefore key in achieving Net Zero. The International Collaboration Towards Net Zero Computational Modelling and Simulation (CONTINENTS) project will build on the expertise of EPCC, the supercomputing centre at the University of Edinburgh, the National Centre for Atmospheric Science in the UK and the National Centre for Atmospheric Research in the USA to transform the state-of-the-art in sustainability and power/energy efficiency of computational modelling and simulation through an ambitious programme of research that will drive innovations in: data centre and HPC system operation; optimal use of hardware; machine learning applied to data analysis and numerical modelling; and software design and development strategies. As a specific use case, CONTINENTS will focus the application of its research on weather and climate modelling. These are both scientifically and computationally challenging, and compute resource intensive, domains that are vital to furthering our understanding of climate change and the factors that influence it. CONTINENTS is a unique interdisciplinary collaboration between leading centres of HPC research and service provision, atmospheric science experts, and numerical and machine learning application developers, providing a comprehensive approach to addressing the Net Zero challenge from the data centre all the way through to individual applications using both computational and climate science expertise. The projects objectives include: establishing a long-term collaboration between leading supercomputing researchers and centres in the UK and USA; pushing the boundaries of sustainable operation of data centres; exploring the use of novel and specialised hardware; developing new methods for performance, power and energy efficient software development and deployment; minimising the resources needed for moving, processing, analysing and storing data; and creating a collaborative research environment that encourages sharing of expertise and knowledge.
Adrian Jackson (Co-Investigator)Ioana Colfescu (Co-Investigator)Mark Parsons (Principal Investigator)Michele Weiland (Co-Investigator)Nick Brown (Co-Investigator)Stephen Mobbs (Co-Investigator)
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know