Ian Ritchie / The Herald Business HQ / September 6th, 2024
FORTY YEARS AGO, in the early 1980s, the University of Edinburgh won a competitive bid to set up a new centre – the Edinburgh Parallel Computing Centre (EPCC) – which was to host the most powerful computer in the UK; a “supercomputer”, to be made available to any in the UK’s academic and industrial computing community who have a need to undertake massive complex calculations.
Over the past four decades the EPCC has regularly upgraded the machines it operates and its current one, called ARCHER, installed in 2021, has a performance of 28 PetaFLOPS per second.
A FLOP is a ‘Floating Point Operation’ a single basic calculation, and a PetaFLOP is one thousand trillion, or 1017, of these.
28 PetaFLOPS every second may seem incredibly powerful but these days it is a relative minnow compared to the leading supercomputers installed across the world.
It currently ranks 39th in the league table of global supercomputers, about 50 times less powerful than the current leaders. We are outgunned by supercomputers in the USA, Germany, Finland, Japan, Switzerland, China, Italy and Spain. Indeed, individual companies such as Microsoft, Nvidia and IBM own computers which are considerably more powerful than any in the UK.
Also, the Brexit deal negotiated by the UK Government has restricted access by British institutions to EU supercomputer facilities in Finland, Italy, Germany and Spain.
These supercomputers are used to allow industrial users across the UK to undertake complex calculations to model processes such as climate behaviour, drug discovery, material science, astrophysics, and nuclear fusion, and provide an essential tool for the development of many areas of advanced scientific development.
This type of costly shared scientific capability is usually enabled by strategic government research funding, provided by UK Research and Innovation (UKRI).
Twenty years ago, when I was a trustee of one of its predecessors, the Particle Physics and Astronomy Research Council (PPARC), we were responsible for the UK’s share of the funding of the European Space Agency (ESA) and the CERN physics research centre, both of which were fundamental to much of the UK’s academic research in aerospace, physics and astronomy. The provision of a powerful shared supercomputer was the equivalent strategic scientific resource to support a wide range of the UK’s research capability.
It has been widely accepted for some time that the UK had fallen behind in its provision of access to high-speed computing capability. The trade association TechUK stated in 2022: “The UK lags other nations when it comes to the infrastructure, skills and resources needed for UK businesses to engage and lead the future of computing”.
So, it was no real surprise that earlier this year, the Conservative government announced that the supercomputer facility at EPCC would be upgraded to acquire an exascale machine, at 1018 FLOPS, about 50 times more powerful than the current ARCHER machine, catapulting us back up to the top of the league table of global supercomputers. There are currently two exascale computers in the US, while Japan, Saudi Arabia and Europe are currently investing billions in similar provision.
But in what has turned out to be a shock to the UK computing community, the new Labour government has pulled the project and cancelled the £1.3 billion of funding for advanced computing research, including the £800 million assigned to the new exascale computer and a further £500 million for AI research, a field in which the UK excels.
This shock was particularly felt at the University of Edinburgh, which has already invested £31 million on the new building needed to host the machine.
After Parliament broke up for the current recess Chancellor Rachel Reeves identified this project as one of the "unfunded" projects in the spending plans that she has inherited from the previous government, so this controversial decision has not been able to be challenged in Parliament. In the past, Ms Reeves has declared that a "stop-go" approach to capital spending was the “British disease”. Our newly elected government has declared its overwhelming aim to be growth – growing our economy so that we can afford the society we want, in health, education, justice and so on.
In that context it is very surprising that the exascale supercomputer is being cancelled. It is bound to damage our ability to attract investment to the UK in fields such as pharmaceutical development, which might now find that Germany, Switzerland or Spain are more suitable to model their huge complex computing calculations.
The government has asked a leading entrepreneur, Max Clifford, to review the UK’s computing needs before October’s budget, and the fact that Edinburgh’s Vice-Chancellor Peter Mathieson has been careful so far not to openly criticise the government’s decision would indicate that it might still be reversed.
That would leave Rachel Reeves responsible for another example of what she dubbed the British disease – "stop-go’" spending.
Comments