Richard Bradbury and Nick Chalk assert that the world is about to end, and that Intel's Merced technology is going to be the culprit! I think this is not the case.
Richard and Nick's articles labour on a few common misunderstandings, namely:
Firstly, NT isn't anywhere near killing UNIX. NT doesn't measure up to UNIX in any way at all, it doesn't have any stability worth talking about when compared with a properly installed UNIX machine. NT doesn't scale well - a good UNIX implementation will perform twice as fast given twice as many processors - not so with NT. How can NT compete with big machines like Sun's E10000, with 64 333Mhz processors and 64Gb of RAM? The answer is simple: it can't. NT is unsupported, in real terms. Microsoft's poor support is legendary - how can you hope to run a mission critical server on NT when you know that you can't get timely help? UNIX vendors have separate companies devoted to support functions.
Now I have to admit that my bias is definately UNIX, since I used to work for Sun's support function, SunService. Consider this fact, though - SunService had somewhere in the region of 100,000 machines covered under support contracts, but had only 150 engineers employed to deal with the problems that were logged. On average, we'd get a call for a particular machine every two or three years.
Secondly, Intel doesn't want to kill its opposition. Not really. You only have to look at the antitrust cases that have gone before to understand what Intel stands to lose. The conjecture made that flaky American law will fail to punish Intel is naive. An open market will not put up with a big monopoly, whether it's the UK, where the Monoplies and Mergers Commission will intervene, or in the US, where antitrust rules will apply. Flaky or not, they're part of the (beloved!) US constitution.
A competitive market is a healthy market, and Intel will be happy to let the likes of ARM compete with them in the more esoteric products, where pickings are thinner and smaller companies can thrive.
Finally, StrongArm hasn't got anything to do with anything, not on the grand scale. It may power a few mobile phones and the Acorn RiscPC may resurface as something else in the future, but we all know that won't do anything that hasn't been done before. StrongArm is a well designed and successful device for the embedded market, that is in my opinion not at all suited to powering a workstation. I just don't understand why anybody would render an expensive workstation impotent by omitting FP!
I don't deny that Intel has a very strong hold on the market, but I really don't think that it has any strong bearing on the supposed battle between NT and UNIX for market share of servers. Let us not forget UNIX is written in C and thus has flourished because of its portability. So there's another processor? Bring it on!
64-bit horror story - Even more horrific...
Disagree? Then reply!