The rocket behind retailing: A British firm is stealing the technological limelight from the US and Japan, but that's not all, says Tom Wilkie

Last week the Geneva-based European Laboratory for Particle Physics, Cern, took delivery of the first of a new generation of supercomputers - a machine designed and built in Bristol. Cern has one of the world's largest computer installations and this is the first time in 30 years that the most powerful computer in the Cern system has been designed and built in Europe, breaking American and Japanese dominance.

The principal author of this renaissance in European computing is a small company called Meiko. Last year, it pulled off a spectacular coup in the United States by landing a multi-million-dollar contract to supply the world's fastest computer to the Lawrence Livermore nuclear weapons laboratory in California. Although Meiko had transferred much of its operation to the US, the Livermore decision provoked outrage from US computer companies at what they saw as the award of a contract to an outsider.

In spite of these successes, Meiko does not want to be seen only as a purveyor of processors to 'rocket scientists'. It wants to sell technology honed at the frontiers of science to mass-market retailers, high street banks and supermarket chains. In an electronic equivalent of beating swords into ploughshares, machines tested in calculations on the hydrodynamics of thermonuclear explosions will be turned to handling payroll and personnel files, stock control and inventories. Meiko has already shipped more than two dozen supercomputers to commercial customers. In the UK alone, these include Bass Taverns, British Shoe Corporation, Lloyd's of London, East Midlands Electricity and South Eastern Electricity.

Although it has a Japanese name, meaning 'clever technology' (or 'trainee geisha girl'), Meiko is a British start-up company. It is a pioneer of a new direction in supercomputing: massively parallel processing. Its machines link together many individual microprocessors - more than 1,000 in some cases - so that computational tasks can be split up and performed simultaneously. The trick is to ensure that every processor knows what the others are doing so that if, for example, one processor needs the result of another's calculations before it can move on to the next step, the processor supplying the result knows when it needs to finish its calculation and where to send it.

This control and communication is such a difficult task that conventional supercomputers have tended to avoid it by using a single ultrafast central processor to carry out calculations one after another as quickly as possible. That approach, however, is expensive: the central processor is costly to design and manufacture and often uses exotic materials. Meiko's massively parallel machines, on the other hand, use (comparatively) cheap and cheerful processors. At the company's small offices in Bristol, one of Meiko's founders, Miles Chesney, gestured casually to a row of four steel boxes. Behind their flashing lights lay electronic circuitry worth more than pounds 250,000, the heart of one of the fastest and most powerful computers in the world. 'This is technology that you're supposed to be able to buy by the yard,' he said.

The supermarket philosophy of pile 'em high and flog 'em cheap might appear a little strange coming from one of the high priests of high technology. Mr Chesney is the man who, in the Eighties, masterminded the design of what was then the world's most technologically advanced computer chip, the Transputer. But from the white Porsche in the company car-park to the diamond ear-stud, it is clear that he has moved far from the abstract mathematical logic of computational science. His designs now are on commerce. If he has his way, Meiko, of which he is co- founder, could be the Microsoft of the late Nineties. He denies the ambition, but it is possible that Mr Chesney could be the British version of the American computer billionaire Bill Gates.

But while he presides over one of the highest of hi-tech computer companies, Mr Chesney is refreshingly dismissive: 'Computing is an immature industry. You can use the telephone exchange without having to know how it works. Computers have been falling short on that score for years.'

He cites how personal computers have become user-friendly only recently - as a consequence of powerful microprocessor chips that allow PCs, for example, to run graphical interfaces such as Windows. Parallel computers have been difficult even for professionals to use because they had to be programmed differently from conventional ones, but now Mr Chesney believes 'the fact that it's parallel can be submerged'. Ultimately, he hopes his computers will become mundane items: 'I look forward to the time when our machines are hidden in the basement, not housed in a cathedral-like room.'

It will not be the speed of calculation that brings parallel processing into commerce, but its ability to handle huge quantities of data in ways that conventional computers cannot. If you have several accounts with a single high street bank, it is difficult for a conventional computer system to appreciate that behind several different account numbers there is a single customer. And integrating a company's knowledge base will get harder in future because the trend in many companies is to move away from highly centralised data processing to distributed computing where employees have PCs on their desks linked to each other in a local-area network.

But Meiko's philosophy goes against this trend. According to Mr Chesney, 'the costs of distributed IT management are disappointing. It may appear cheaper but the costs are hidden because they are distributed, too.' Most important is the fragmentation of the company's knowledge about itself and its business. Meiko's parallel computers can, Miles Chesney believes, reintegrate that knowledge and serve a business better. The company has been working for nine years since its foundation. Now, Mr Chesney says, the time has come.

(Photograph omitted)