Martin Banks, Personal Computer World 08/89 - checked

August 1989

When you've been around as long as me, approximately 1.3 billion micro-fortnights, you get to thinking you've seen it all. The 'been-there-done-that' syndrome is quite often to be found creeping over the horizon like a cynical old friend.

This is most noticeable, perhaps, in the area of semiconductor technology, where everything is all nicely planned out according to the laws of physics, isn't it? Well, up until the back of last year, that probably was the case.

Now, however, something has occurred which is almost interesting enough to stir these aged bones towards excitement. Only almost, of course, nothing is THAT exciting when you get to my age.

But, from a purely technological, gosh-isn't-that interesting standpoint, the latest gizmo news from the good old US of A is... well... quite interesting, I suppose.

You see, they have come up with a whole new way of making semiconductor devices to a whole new set of rules. That means that old set of rules, while they work well enough, are (or will be some time in the future if this all works properly) passé.

That standard set of rules relate to the production of semiconductor devices as we know and love them today. Open up your PC and look adoringly at all the little black slabs of plastic. Inside every one is a chip you can feel love for. Those rules suggest that, if you make a semiconducting structure too small it will no longer work as a semiconductor and will, therefore, stop being lovable by the average PC aficionado.

However, a new set of rules has been discovered. These suggest that, if you make the semiconductor structure small enough, the rules of quantum mechanics start to come into play. For the life of me I don't understand a word of it, but I am reliably informed that, if the structure is small enough, the electrons within it start acting as waves and stop acting as particles. It's all terribly impressive.

Indeed it could be. On a 1 to 10 rating of applied long words, this new technology probably rates around 11.5, which makes it easy to dismiss. And that would be to miss its point, for these quantum structures could change just about everything to do with computers. For example, how would you like a cheap laptop that has the power of a Cray supercomputer? (whoever said 'will it run DOS?' kindly leave the room).

One of the key components of any such system will, it is reckoned, be things called quantum wires, laid out in a lattice structure. 'Wire' in this context is a relative term, for in the world of sub-atomic physics which constitutes the environment of quantum structures, some 6 million of them are needed to make up the width of a human hair.

At the back end of last year, Texas Instruments announced that it had come up with a quantum effect transistor. This uses quantum structures and materials to produce a transistor that has critical geometries 100-times smaller than a conventional silicon device. They may be interesting on the 'so-what' scale of things, but the real advantage is that it should also make a transistor which switches 1000 times faster.

What does this give us, I hear you ask? Well, imagine it. The subject of geometries, for example, means that line widths within an integrated circuit can be reduced to around 0.02 microns - just a few atoms wide. If you consider that the current crop of new processors, the Motorola 68040 and the Intel 80486, pack over 1 million transistors into a chip and run at up 12 MIPS, you can work out what might be possible.

A processor with 100 million transistors is a tempting prospect, though quite what functions might be designed into it is a matter for considerable conjecture. With that power potential available it would be possible to jam 10 or more 386 PCs, complete with 1 Mbyte of memory each, on to a single quantum circuit. Is that what someone would want? (At least the old DOS applications would be safe for a bit longer).

Now add in the fact that quantum structures obey rules which make things happen much faster, approximately 1000-times faster in this case, and the prospects get even more interesting. Just imagine the PC makers advertising the fact that their new toys run at 25,000 MHz and have 10 complete systems inside them. Now, that really would make windowing technology something to play with.

Last, but by no means least, quantum structures will use hardly any power, so all this performance potential could easily be squeezed into a laptop system run from torch batteries. The prospects are quite mind-boggling.

In practice, of course, none of this might actually come to pass. The fact that Texas Instruments has made such a transistor in the laboratory, with others such as IBM hot on its heels, does not mean that the technology will finally see the light of day in working products. Even the most enthusiastic researchers admit that there is a great deal of work yet to do before quantum devices are available. None of this may come to pass.

But what if it does? I know I was joking about DOS earlier, but think about what software will be available in ten years time to run on processors like these. There is no sign of DOS going away, and if the developments in Unix and the rumoured 386-version of OS/2 are any guide, DOS applications will be with us for a good while yet.

Then what will we see in the news? How about: 'Lotus says 12-3 Release 18 Is Coming Soon', 'Ashton-Tate's dBASE VIII Nearly Ready' or 'Amstrad Clones IBM's PS/7 In Wrist-Watch System'.

end