By continuing to use the site you agree to our Privacy & Cookies policy

Your browser seems to have cookies disabled. For the best experience of this website, please enable cookies in your browser.

Close

Your browser is no longer supported

For the best possible experience using our website we recommend you upgrade to a newer version or another browser.

Close

From code-breaking to building design: supercomputers move on

martin pawley

There can hardly be a sizeable architecture firm in the world that is not updating, reorganising, expanding, restaffing, outsourcing or trying to bring under control its computer operations. Why is this? Because in the past few years there has been a revolution of expectations in the world of CAD, with the realisation that falling prices may not mean cheaper operations forever.As a result, design firms that today exult over the increasing number of plotters, monitors, scanners and so on that they can buy or lease for less than they cost last year, already know that the offthe-peg supercomputer is peering over the technological horizon, threatening the biggest onslaught of artificial intelligence since the arrival of the PC - the replacement of computer-aided design by computer-dependent design - as in such fields as defence, distribution and financial services.

During the Cold War, the US government spent an estimated $4 trillion on intelligence gathering, code-breaking and computer-controlled weapon systems - far more money than any consortium of corporations could have mustered for a civilian task. But when the Cold War combatants stood down, the knowledge associated with their defence activities lost its protected status. In 1996, as part of this process, Cray Research, whose founder Seymour Cray designed the Control Data 6600 and 7600 supercomputers that simulated nuclear explosions and cracked enemy codes, was bought out by Silicon Graphics, a firm that announced its intention to turn the extremely expensive Cray machines into a range of supercomputers available to anyone.

Until this breakthrough, supercomputing had been a slowgrowing phenomenon, the outcome of years of research regardless of cost.That is why there are very few supercomputers in the world.

They can not only perform astronomical numbers of calculations per second but, more importantly, relate different sets of data from dozens of different disciplines so as to produce an immediate synthesis - something that cannot be done simultaneously in any other way. For this reason supercomputers have carried out exotic tasks: testing hypotheses about the universe, designing supersonic aircraft and advanced meteorological studies. Understandably, the broader market for the kind of work they can do was confined to government departments, defence establishments, universities and major industrial corporations. But not any more. Knowledge of the outcome of a process before it is implemented - whether it is a There can hardly be a sizeable architecture firm in the world that is not updating, reorganising, expanding, restaffing, outsourcing or trying to bring under control its computer operations. Why is this? Because in the past few years there has been a revolution of expectations in the world of CAD, with the realisation that falling prices may not mean cheaper operations forever.As a result, design firms that today exult over the increasing number of plotters, monitors, scanners and so on that they can buy or lease for less than they cost last year, already know that the offthe-peg supercomputer is peering over the technological horizon, threatening the biggest onslaught of artificial intelligence since the arrival of the PC - the replacement of computer-aided design by computer-dependent design - as in such fields as defence, distribution and financial services.

During the Cold War, the US government spent an estimated $4 trillion on intelligence gathering, code-breaking and computer-controlled weapon systems - far more money than any consortium of corporations could have mustered for a civilian task. But when the Cold War combatants stood down, the knowledge associated with their defence activities lost its protected status. In 1996, as part of this process, Cray Research, whose founder Seymour Cray designed the Control Data 6600 and 7600 supercomputers that simulated nuclear explosions and cracked enemy codes, was bought out by Silicon Graphics, a firm that announced its intention to turn the extremely expensive Cray machines into a range of supercomputers available to anyone.

Until this breakthrough, supercomputing had been a slowgrowing phenomenon, the outcome of years of research regardless of cost.That is why there are very few supercomputers in the world.

They can not only perform astronomical numbers of calculations per second but, more importantly, relate different sets of data from dozens of different disciplines so as to produce an immediate synthesis - something that cannot be done simultaneously in any other way. For this reason supercomputers have carried out exotic tasks: testing hypotheses about the universe, designing supersonic aircraft and advanced meteorological studies. Understandably, the broader market for the kind of work they can do was confined to government departments, defence establishments, universities and major industrial corporations. But not any more. Knowledge of the outcome of a process before it is implemented - whether it is a stretched version of an airliner, or the verification of global warming - is power indeed.The prospect now is of a free market in this type of machine at such a low cost that it will devolve down to relatively small organisations, such as those responsible for the design of buildings.

Architecture on its own used to be regarded as a small sector of the computer market. Now it is becoming a small sector confronted with a big product, a machine that can handle all the mysteries of architectural form and still have enough calculating power left over to run structural design programmes, analyse fire risks, design building services, simulate wind-tunnel tests, write specifications, calculate costs, codes, claims, contracts and project management and, best of all, interrelate all of them. So who will stump up the millions - no longer hundreds of millions - needed to become the possessors of all this synchronised knowledge?

It would be nice to say 'architects will', but will it be true? What is about to become available, at a price, is the supercomputational capacity to turn any consultant (building economist, lift consultant, engineer, surveyor, project manager) into a total building design organisation. If they are not careful, more than a few medium-sized architectural firms are going to find themselves competing with people who used to work for them.

Have your say

You must sign in to make a comment.

The searchable digital buildings archive with drawings from more than 1,500 projects

AJ newsletters