This funding opportunity is designed to deliver on the ‘Investing in People’ pillar of the ExCALIBUR programme.
RSEs are integral members of research teams who work alongside researchers co-designing scalable, robust, verifiable and maintainable software to deliver research impact. RSEs work across almost all the UKRI research base from cosmology to digital archiving.
At present a subset of RSEs work in the domain of high performance computing (HPC). With the imminent arrival of the next generation of supercomputers the demand for RSEs with the expertise to meet the challenges of new programming paradigms along with novel and heterogeneous architectures has never been greater. Highly skilled RSEs will be required across both academia and industry to exploit these advances.
Provision for the development of HPC skills within the RSE and research communities
It is acknowledged that many sections of the RSE community do not currently work on software in the HPC domain. However, with the anticipated growth in demand for skills in HPC and beyond it is necessary to enable the existing RSE community to grow their expertise in this domain. The RSE Knowledge Integration Landscape Review identifies a set of skills required for RSEs working in the HPC domain which include:
- core skills
- programming skills
- debugging, performance profiling and optimisation
- software career skills
- artificial intelligence and machine learning
- application and data workflows.
Provision for the development of exascale skills within the RSE and research communities
The RSE Knowledge Integration Landscape Review highlighted the following areas where skills development was required within the RSE and wider research community:
- parallel paradigm
- data and application workflows and their optimisation
- HPC and artificial intelligence benchmarking of systems, codes and models
- heterogeneous architecture programming
Long term career path development
To ensure the UK research community can effectively exploit the potential of exascale computing, recognition is required of the valuable contribution software engineering focused staff make to research outcomes. A strategy is therefore required to develop a long term career path for these individuals within both national laboratories and institutions.
The ExCALIBUR programme
ExCALIBUR (Harnessing Exascale Computing Algorithms and Infrastructures Benefitting UK Research) is a £45.7 million Strategic Priorities Fund (SPF) programme led by the Met Office and UK Research and Innovation to meet this challenge by delivering research and innovative algorithmic development to harness the power of exascale HPC.
Radical changes to supercomputer architectures are on the horizon. The current simulation codes, that much of UK science relies on, are designed for current supercomputer architectures. These codes will, at best, not be able to fully exploit the power that the supercomputers of the mid-2020s will deliver; at worst, they will run slower on those machines than they do now.
Future computers will be more energy efficient and so the longer we rely on the current approach, the more expensive the solution will be. Therefore, it is essential that we invest now in redesigning those simulation codes so that they perform well on the future generations of supercomputers.
ExCALIBUR will be delivered over five years and will meet this challenge by delivering research and innovative algorithmic development to redesign the high-priority simulation codes to fully harness the power of future supercomputers across scientific and engineering applications.
It will achieve this by bringing together an unprecedented range of UK domain experts, mathematicians and computational scientists who will identify common issues and opportunities in the high-priority simulation codes and focus their combined scientific expertise and resources to accelerate toward interdisciplinary solutions.
The programme objectives have been designed to specifically address the benefits sought, which are:
- efficiency: the UK’s most important scientific simulation codes will be able to harness the power of the supercomputers of the mid-2020s, resulting in an increase in scientific productivity for a given investment
- capability: capitalising on this efficiency will enable the UK to continue to push the boundaries of science across a wide range of fields delivering transformational change in capability
- expertise: a new, forward-facing, interdisciplinary approach to RSE career development will position the next generation of UK software engineers at the cutting-edge of scientific supercomputing.
ExCALIBUR is built around four pillars:
- separation of concerns
- data science
- investing in people.
These pillars describe the fundamental principles that guide the development of research within the ExCALIBUR programme and are designed to ensure that the outcomes are future proofed against the constantly evolving landscape of hardware design. It will be delivered through six main activities:
- the redesign of a core set of simulation codes (use cases) chosen to span a wide range of science domains
- knowledge integration across the programme through widely applicable cross-cutting themes
- application of learning from these activities to a second wave of use cases
- exploratory research to identify and develop emerging high-performance algorithms in areas with significant potential impact
- an interdisciplinary research software engineer knowledge integration activity
- an annual capital investment to support the development of novel test beds to enable co-development with industry.
The ExCALIBUR programme defines cross-cutting research as: a coordinated approach addressing a known technology or infrastructure issue, which, if resolved, will lead to significant progress across a range of exascale software development challenges.
The themes of the cross-cutting research opportunity were defined following a market engagement event jointly held by the Met Office and EPSRC in November 2020. These themes were included within the two calls subsequently run and themes were categorised as common approaches and solutions or potential disruptors.
Common approaches or solutions:
- input, output and storage
- data workflow
- verification, validation and uncertainty quantification
- domain specific languages.
- exposing parallelism: parallel-in-time
- exposing parallelism: task parallelism
- machine learning: optimising numerical methods and augmenting physically based applications
- future computing paradigms.
The grants funded through these calls will utilise the lessons learnt from the use cases and design and development working groups to address common issues that impact scientific code under development for use at Exascale.
EPSRC is fully committed to develop and promote responsible innovation. Research has the ability to not only produce understanding, knowledge and value, but also unintended consequences, questions, ethical dilemmas and, at times, unexpected social transformations.
We recognise that we have a duty of care to promote approaches to responsible innovation that will initiate ongoing reflection about the potential ethical and societal implications of the research that we sponsor and to encourage our research community to do likewise. Therefore applicants are expected to work within the EPSRC framework for responsible innovation.