Michael Brooks meets research director Karl Kummerle and tries on a computer for size
Some Japanese girls are waving and dancing 3cm in front of my right eye. No one else can see them, or hear the music in my ear.
I can, with a bit of focusing effort, look right through their bobbing heads. When the video clip ends, the 2,5cm-wide screen displays the Windows 98 desktop. Using the joystick mouse in my left hand, I click on the video icon and repeat the experience.
I am wearing this computer. The screen is mounted on a headset like spectacles with only one lens. The computing power (equivalent to a 233MHz notebook) is held in what looks like a portable CD player weighing less than 300g, including the battery pack. It slips into a pocket or clips on to your belt.
Connecting wires run from this unit to an earphone and the mouse, which can be held and operated with one hand. The mouse also has a built-in microphone, because this computer will also respond to spoken instructions. To wear one is to want one.
This is the way computing is heading, according to IBM, the machine’s developer. The wearable PC is part of the `”Ease of Use” initiative, designed to get everyone using computing technology – especially IBM’s – and loving it. It is about to go into trials and the company hopes to market the first versions sometime next year.
Within a very short time, emergency services personnel or aircraft maintenance technicians could be accessing online manuals while working, and commuters might download e-mail or surf the Web while watching the countryside through train windows.
This new technology is Exhibit A in the attempt to prove that a once-dominant force in computing technology is back in business. Not so many years ago the popular definition of a personal computer was an “IBM machine”; since then, the firm has almost crashed and burned.
Karl Kummerle, director of the Zurich research laboratory where the Wearable PC and several other innovations are on display, understates the situation beautifully. “A few things went wrong. IBM made some strategic mistakes.”
These “strategic mistakes” include selling off IBM’s share of Intel, and ditching the mainframes business just before they were put to use as high-end servers for network computing. “We had written them off and now it’s a booming business,” Kummerle admits.
But IBM’s research divisions, based in eight laboratories around the world, plugged away at the fundamentals. They were successful, too, despite enormous cuts in the research budget. Kummerle’s Zurich laboratory gained two Nobel prizes, in 1986 and 1987, for the Scanning Tunnelling Microscope – now a standard tool in almost every university physics department – and the discovery of high-temperature superconductivity.
The research effort began to bear fruit in computing with the 1997 victory of IBM’s Deep Blue II computer over the chess grand master Garry Kasparov. They have taken this success and developed Blue Pacific: currently the world’s fastest supercomputer.
Every second it can perform nearly four trillion operations (that’s 63 000 years of using a handheld calculator), and its memory is equivalent to around 80 000 desktop computers. At the end of October, IBM’s Deep Computing division delivered one to the United States Department of Energy who see Blue Pacific as a way out of the problems caused by the cessation of nuclear tests.
The US nuclear stockpile has been sitting around for a few years now without testing. Blue Pacific is being used to run the Monte Carlo simulations developed to follow individual nuclear reactions; it does this fast enough to be able to work out the amount of degradation in the weapons’ nuclear material, and run a test, simulating an explosion.
That, says the Department of Energy, should mean no more underground tests will ever be necessary.
Plenty of other lucrative applications are awaiting Blue Pacific: pharmaceuticals production and data mining (finding patterns in consumer behaviour, for example) are on the agenda. Just more than a year ago, IBM signed a deal with biotechnology company Monsanto to hunt for hidden patterns in genetic information. Next year, Blue Pacific will be superseded by a machine that is 10 times as fast. Researchers are aiming for 100-trillion operations per second within a few years.
This kind of progress requires research one step ahead of the next anticipated problem. IBM’s rivals have openly acknowledged that, in one area of chip technology, they have been left standing. Last year IBM began producing a new copper chip that works up to 40% faster than current silicon technology – with a 30% reduction in manufacturing costs.
Copper replaces the aluminium used to link transistors. It is a better conductor of electricity and so signals travel faster. But it is more difficult to connect to the silicon transistors, and has a tendency to contaminate the chip’s other components. There is also the problem of inertia in the semiconductor industry: aluminium has been the standard for 30 years.
However, miniaturisation has reached the point where aluminium’s electrical resistance is becoming prohibitively high. IBM’s breakthrough could not have come at a better time – and with a better head start. Intel, and most other major chip-producing companies, are 18 months behind.
Rumours abound within IBM that while the company was trying to axe the copper chip project one researcher stubbornly refused to give it up. It is a process that Kummerle is familiar with: 10 years ago, while he was heading the communications research area of the Zurich laboratory, he produced a prototype machine that could integrate voice and data traffic.
“We couldn’t get the interest of the development people,” he says. “They were impressed by the data part, but not with voice integration.”
In the end Kummerle and his team gave up. “It might have been the right decision then but, in light of Internet telephony, it looks a bit different now,” he says.