Partnering up to reach net zero computing
Celebrating Net Zero Week 2022 by announcing our project partners!
Digital research infrastructure (DRI) ranges from supercomputers to data archives, from software to the people using and maintaining it. These infrastructures are essential for many scientific studies that are looking into the most pressing issues our society faces. Operating and using DRI has a carbon impact largely due to energy consumption and the embodied carbon from the manufacturing footprint of computer hardware. The carbon impact needs to be assessed across all of the UK Research and Innovation (UKRI) community if we are going to reach our net zero goals by 2040 (or sooner). This is a huge task that requires collaboration and partnerships to be formed across all UKRI sectors. The UKRI Net Zero DRI Scoping Project is building a community of experts to help tackle the task - 18 institutions have recently been made project partners.
About the scoping project
The UKRI Net Zero DRI Scoping Project is collecting robust evidence and formulating recommendations that ensure we can provide carbon neutral computing and data services for the research and innovation communities. Evidence collected during the 19-month project will cover all areas that influence net zero targets – ranging from policy decisions, procurement processes for hardware, changing behaviours of users and managers of DRI, and much more. Our scope covers all UKRI owned or majority funded DRI - this is spread across a range of organisations, including universities and research councils.
Building a community of experts
Due to a vast scope and short timescales, and to increase the breadth of expertise and experience informing the project, the core project team has enlisted the help of partners across various organisations to help gather the necessary evidence. Some of the work being undertaken by project partners is summarised below. Funding for many of the partners was allocated via sandpit events which attracted new ideas from the community, while the consortium is put together to address problems identified in the original project funding call.
Dr Martin Juckes, the project lead, said “The target of net zero digital research infrastructure across UKRI is a massive challenge which will require both exploitation of emerging technologies and a transformation in the way we think about our use of resources. We are building a community of experts who will help contribute to our evidence based roadmap.”
The work undertaken by project partners is being coordinated by the core project team, based at the Centre for Environmental Data Analysis (CEDA). All evidence collected will be translated into recommendations that will ensure UKRI’s DRI can reach net zero. The final report is due for publication in Summer 2023. An interim report (including initial recommendations) will be published in Summer 2022.
To keep up to date with the project’s progress and findings, sign up to the mailing list or follow us on Twitter. If you have any questions please contact the CEDA helpdesk, making clear that you are interested in the UKRI DRI net zero scoping project.
| Project name | Summary of planned work | Institutions involved |
|---|---|---|
| Sandpit funded work | ||
| Giving Voice to, and Empowering Stakeholders of UKRI DRI: A Net Zero Workshop Series (Go Zero) | Different DRI stakeholders need help to understand and plan for meeting net zero challenges. Many of the skills and knowledge needed to solve net zero challenges are already in the community – but we lack the mechanisms for sharing them and enabling behaviour change. This project will lead a series of participatory workshops which will be undertaken with the aim to connect and enthuse the DRI community in order to meet the net zero challenge. The workshops are designed to empower all stakeholders to take ownership of the path to net zero. Reference document DOI:10.5281/zenodo.6787567. |
Brunel University London Heriot-Watt University Science and Technology Facilities Council University of Reading/National Centre for Earth Observation National Oceanographic Centre |
| ‘Value’ and Net Zero Decision Making | This project will develop a methodology to aid net zero decision making for UKRI in relation to Digital Research Infrastructure. This ‘net zero decision making’ is made up of: Comparison: the carbon emissions of a task vs. the ‘value’ it generates Action: assigning which tasks get to emit carbon via the use of UKRI electricity and infrastructure ‘Task’ here can be any use of DRI, whether it’s for computation or storage etc. Any use of infrastructure or electricity will have a carbon cost associated with it and will therefore affect the pathway to net zero. The project will review how UKRI currently make decisions around DRI use (if, when and how a task gets to use DRI), undertake a literature review of how emissions-vs-value decisions have been managed in similar infrastructure environments, and provide specific recommendations to UKRI of a NetZero decision making methodology. Adopting the methodology outlined above, not only helps with keeping UKRI accountable to the net zero pathway, but also starts a mindset shift which brings the real carbon impact of DRI activities to the front of all UKRI decision making. Reference document DOI:10.5281/zenodo.6787673. |
Heriot-Watt University |
| Learning from the Big Picture: Applying Responsible Innovation to the Net Zero Research Infrastructure Transformation (ARINZRIT) | Creating a net zero digital research infrastructure needs an understanding not just of the technical factors, but also wider social and organisational factors which drive demand for it. By understanding the broader landscape of policy, practice and engagement with it, this project seeks to ‘map the landscape’ of how policies and practices interplay and affect growth in digital research infrastructure, providing key evidence at a ‘policy & cultural’ level, and identifying ‘risks’ and opportunities to start answering critical questions toward net zero. The project will identify the ways in which socio-technical forces shape, and are shaped by, the research infrastructures and their environmental impacts. It will suggest alternate ways in which these factors may shape the evolution of the infrastructures in the future, in line or exceeding expectations for net zero. Opportunities for impact reduction through intervention points in this system, how they might reduce infrastructure demand, and risks of potential unintended consequences from net zero initiatives will also be identified. Reference document DOI:10.5281/zenodo.6787422. |
Lancaster University University of Bristol Science and Technology Facilities Council Edinburgh University |
| IRIS Carbon Audit SnapshoT (IRISCAST) | Making good decisions around infrastructure procurement and provision requires understanding, as much as possible, the full carbon costs for operating, maintaining, and using the infrastructure; going beyond accounting for electricity and cooling, and including the full chain of costs embodied in the infrastructure. Enabling good decision making therefore requires a source of accurate real-world data on infrastructure climate impacts. This project will perform an audit of carbon costs across a multi-site heterogeneous infrastructure (called IRIS) by collecting and analysing snapshots of actual usage. It will identify where the carbon cost sits across the infrastructure. It will also find out what the drivers are that influence carbon cost. The evidence collected will be used to allow for consistent reporting, evaluation and decision making. Reference document DOI:10.5281/zenodo.6787626. |
Queen Mary University of London University of Cambridge University of Edinburgh |
| ENergy-aware hEteRoGenEous compuTIng at sCale (ENERGETIC) | Current leading-edge high performance computing (HPC) systems are often heterogeneous, comprised of combinations of multiple compute units and accelerators, including (but not limited to) CPUs, GPUs and FPGAs. HPC is a significant contributor to energy usage. However, the energy-to-solution varies between these architectures. In terms of minimising energy consumption, this choice of possible architectures presents various challenges to HPC managers and users. Currently, there is little data on the energy efficiency of codes or algorithms across different architectures, a lack of an established framework for measuring this on new systems, and little use of existing tools like Performance Application Programming Interface (PAPI) for measuring and comparing power draw from hardware. Due to this lack of data, it is difficult for DRI managers/users to effectively design codes and systems to meet quantifiable energy budgets that contribute to achieving net-zero goals. This project aims to test whether the use of heterogeneous architecture could significantly reduce the energy-to-solution and thus the energy consumed by UKRI digital research infrastructures. A workshop and other community engagements are planned involving key stakeholders to aid understanding the current state and enable dissemination of project findings of reducing energy consumption. Reference document DOI:10.5281/zenodo.6787467. |
Manchester Metropolitan University Newcastle University EPCC/University of Edinburgh University College London |
| HPC job efficiency and energy usage: monitoring and reporting (HPC-JEEP) | Traditionally, resources on high performance computing services have been granted in units of compute time (core-hours, node-hours, etc.) with users not informed, nor paying much attention to their energy use; and services not generally reporting breakdowns of energy use other than for total electricity charges. Without an understanding of how research areas, projects and users use energy and how efficiently they use resources it is difficult to plan future procurements or service strategies towards a net zero goal. This project will look at what level of energy and efficiency information can be extracted from current and historical per-job data from ARCHER2 and DiRAC services; and how this data can be analysed and synthesised to provide the information required for funders (who procure and set service strategy and objectives) and researchers (who make decisions about how best to use the resources they are granted) to make informed decisions about how to manage HPC resources to extract the maximum research benefit in the most energy efficient way. Reference document DOI:10.5281/zenodo.6787599. |
University of Edinburgh Durham University |
| CQuanDRI: Quantifying the Carbon Emissions of Digital Research Infrastructure | How does variability in the utilisation of infrastructure translate to energy consumption? Which are the indicators that minimise uncertainty of associated carbon footprints? How does the varying grid carbon intensity affect carbon footprints, and can this be exploited to reduce emissions? This project will provide a measurement trace that captures the sustainability aspects in the operation of a DRI service, including dependencies between different indicators (e.g. data flow, power consumption, and carbon intensity). Reference document DOI:10.5281/zenodo.6948779. |
University of Bristol University of Oxford STFC Newcastle University |
| Consortium funded work | ||
| In-depth case studies: JASMIN, JADE, Scafell Pike | An in-depth analysis of JASMIN, JADE (for AI focus) and potentially Scafell Pike (high ranking in Green 500 list) will be undertaken to determine their carbon footprint. | Scientific Computing Department within the Science and Technology Facilities Council (SCD STFC) |
| In depth case study: ARCHER2 | Detailed analysis of ARCHER2 carbon footprint (energy use, and, as available, embedded energy, e.g. from PAIA or empirical analysis based on area of printed circuits) and how it relates to research outputs; methods used to optimise processes which impact on the energy footprint of users. The team will also contribute to the design of a questionnaire which will be used to gather information from other facility and service hosts (Tier 1, 2 and 3). The team may also contribute to data gathering, by talking to service managers to fill in the questionnaire - this will provide input for the mapping activity (see below). |
Edinburgh Parallel Computing Centre at Edinburgh University |
| Mapping the DRI | This work will develop plans for a database that will enable mapping of the entire UKRI DRI estate. The team will look into using the code base of the UKRI Infraportal, potentially extending the level of detail to support aggregation of some key capabilities (e.g. FLOP rating, transistor count, power consumption). They will also look for compatibility with the TOP 500 and Green 500 lists. |
Scientific Computing Department within the Science and Technology Facilities Council (SCD STFC) |
| Providing advice and expertise on roadmap development | The team at UCL will contribute to the literature survey, particularly in the area of developing roadmaps for net zero and related sustainability targets. They will also assist in developing the roadmap itself. Other areas include stakeholder engagement with the ESRC and AHRC communities and gathering evidence relevant to the ethical dimension of the transition to net zero DRI. |
Institute for Innovation and Public Purpose (IIPP), University College London |
| Providing expertise on sustainable computing | Glasgow University will provide expertise on advances in computing science, software engineering and hardware technology supporting the goal of low-carbon and sustainable computing. They will also provide links to the EPSRC user community. They will assist and contribute to the literature survey and developing the roadmap itself - both by contributing to workshops and engaging with the project to guide the evidence gathering process (e.g. via interviews with EPSRC funded facilities managers.) |
Glasgow University |
| Community survey | Edge Hill University will contribute on outreach to the broad community, reluctant stakeholders and probing implicit attitudes through a user and stakeholder survey. | Edge Hill University |