ARCHER2

High performance computing for national research

High performance computing (HPC) is crucial to the UK’s national resilience. UKRI continues to invest in world class advanced computing resource for  national research.

UKRI’s latest investment, ARCHER2, which became operational in 2021, is the UK’s most powerful supercomputer yet - putting world-leading capabilities in the hands of UK-based scientists.

Just like its predecessor, ARCHER1, but now with significantly advanced speed and capabilities, ARCHER2 provides crucial support across the nation’s science and innovation landscape – from ultra-precise weather forecasting and flood prediction to jet engine design, new drug development, the creation of sustainable technologies and materials, and helping to tackle the global pandemic – to name but a fraction of the uses it can be put to.

ARCHER2 is a 23-cabinet HPE Cray EX supercomputing system with an estimated peak performance of 28 PFLOP/s. The machine has 5,860 compute nodes, each with dual AMD EPYC Zen2 (Rome) 64 core CPUs at 2.2GHz, giving 748,544 cores in total.

The new resource is housed at EPCC, the University of Edinburgh’s Advanced Computing Facility. The project is being delivered and supported by UKRI’s Engineering and Physical Sciences Research Council (EPSRC) and Natural Environment Research Council (NERC), with a combined investment of £79 million.

Available to thousands of academic, business and government users, ARCHER2 has a peak performance equalling around 250,000 modern laptop computers all working together, or 10 billion times faster than the first Cray supercomputer in 1964. And ARCHER2 is capable, on average, of over 11 times the science throughput of ARCHER1.

But it’s not just their speed and power that makes supercomputers like ARCHER2 so special, it’s how they can be used by scientists. Traditionally scientific research was performed in a lab, perhaps with a computer interpreting results, but research in many disciplines is now heavily dependent on specialist supercomputers to perform experiments entirely virtually, in ways that would be impossible if addressed by experiment alone. Being able to model and simulate real-world problems enables the exploration of new areas that cannot be addressed by experiment alone.

ARCHER2 gives UK researchers the ability to perform more complex, larger and higher resolution simulations than were possible previously.

ARCHER2 is already being used to model and predict the spread of toxic life-threatening gases from a volcanic eruption in Fagradalsfjall, Iceland which began in March 2021. Not only is it interpreting huge amounts of data, it is also enabling real-time research to take place in a hazardous environment that would have been impossible to work in without endangering life.

Using High Performance Computing to tackle the pandemic

UKRI led a consortium of High Performance computing (HPC) facilities across the UK to focus their power on the world’s most pressing, complex problem. The consortium supported global efforts to tackle the pandemic, including the US-led COVID-19 HPC Consortium.

It includes Distributed Research using Advanced Computing, which is the Science and Technology Facilities Council’s supercomputing facility, and ARCHER1 and ARCHER2.

Researchers used ARCHER1 during the pandemic to put together the known biology, chemistry and physics of the virus to create computational models of molecules.

Professor Syma Khalid from the University of Southampton explains: "If we imagine two pieces of LEGO, there is one that represents the virus and another that represents humans. Unfortunately for us, those two LEGO pieces come together and fit rather well. We’re using computer simulations to design new LEGO pieces that will come in and prevent that tight fit, and thereby prevent the virus from infecting people."

Improving cloud modelling for better weather and climate prediction

The turbulent behaviour of clouds is the reason for many of the uncertainties in weather and climate prediction. Researchers at the University of St Andrews and University of Leeds have worked with the EPCC to develop software which allows vastly improved computational modelling of clouds.

Current weather and climate models struggle to resolve the complex details of the interactions between clouds and their environment.

The team of researchers had previously developed a new numerical model, MPIC (Moist Parcel-In-Cell), which deals with the dynamics of clouds. The team has now incorporated this code into a community code used by atmospheric scientists around the UK, the MONC (Met Office NERC Cloud) framework.

The resulting new PMPIC (Parallel Moist Parcel-In-Cell) code can be run on many thousands of processors on supercomputers such as ARCHER. Compared to the original MPIC code, this permits much larger simulations to be carried out, allowing the dynamics of the cloud to be resolved in much greater detail. This is particularly important at the edge of the cloud, where evaporation of cloud droplets and the difference in velocity between the cloud and its environment create turbulence.

Improvements made to the MPIC codebase have also led to the code running significantly faster. The new code is widely adaptable and could in future be used to model other aspects of the atmosphere, including atmospheric chemistry and the behaviour of rain and snow. It could also be applied to other fields of research, including ocean modelling, astrophysics, and modelling of gravity currents.

For more information about ARCHER2, visit: www.archer2.ac.uk