Experience shows us that new technologies turn into usable products much more slowly than vendors predict. So it is with Intel's Itanium 2 processor. Nearly all the server vendors are jumping aboard this bandwagon, and many pundits now say that the days of Risc systems are severely limited.
Some industry observers seem to be unsettling IT managers, by suggesting that the Risc upgrade path will soon end and vendors are about to cast their customers adrift. Well, the pundits are basically right - except that the change is going to happen much more slowly than they suggest.
Yes, Risc processors are doomed in the long term. Simple economics suggest this is so. To design and develop a new processor would cost a company an arm and a leg. Setting up the new production facilities in order to build it would cost the rest of the body, plus those of some relatives.
That is why Compaq and HP gave up on Alpha and PA-Risc chips respectively, and why they have not gone back on those decisions since they merged. IBM has sufficient financial clout to continue its Power line of chips, but I doubt we'll see much beyond Power 6, unless IBM copies the tactic of Sun with UltraSparc - using third parties to manufacture the parts.
Sun has persuaded few other computer makers to use its UltraSparc chips in their own products, so the demand for its processors is relatively small. This limited demand reduces the scope for investment in new designs, and makes the technology less attractive to third-party manufacturers.
In other words, the price/performance equations for Risc processors are likely to become more unfavourable in the future, to the point where it is no longer possible to produce new, higher-performing components at prices that anyone in their right mind would be willing to pay.
But this is not going to happen tomorrow, or next year. It is at least five years away, and quite probably 10. Itanium will creep up on us all, not come storming in like an avenging angel, so IT managers don't have to panic about being inconvenienced just yet.
Itanium will gradually change attitudes towards technology. It doesn't matter what one thinks about it as a chip, its arrival has, for good or ill, cast the die for the next 20 years, probably more, in terms of hardware development. From now on the technology is no longer sexy stuff: it is mainstream and ordinary, today's equivalent of pen and paper.
Its development is now likely to slow down and the technology will mature. At the same time, old technology, as users of Digital Vaxes and PDP-11s will no doubt testify, will have a worthwhile and honourable place for years to come.
Risc systems won't die a sudden, vainglorious death; they will just slowly fade away. And IT managers of the future may well look back and wonder why such systems were ever considered good.
Have your say: contact IT Week
reader comments