VMS coming back to life 1297

part:

On the other hand, I have to admit that I see no evidence that the people in "the executive suites in Santa Clara" are doing anything to preserve the Itanium from oblivion.

The Itanium is supplied on chips with large caches, which help to cover up the weakness of the architecture, in requiring programs that occupy more memory to do the same thing.

But it is supplied on chips with only rudimentary pipelining, and slow clock speeds, compared to what a mbutt-market Pentium IV offers.

Intel's great strength is that the 386 architecture, for all its clumsiness, is the industry standard. Intel has the market power to piggyback the IA-64 architecture on that standard, and make it the new standard - provided, of course, that this architecture can provide, or even be made to appear to provide, an improvement in performance.

But they've sedulously avoided doing this, thus ensuring that the Itanium - like the 68020-68881 architecture, like the Alpha - is doomed to languish in the face of the 386 architecture... *regardless* of merit.

Of course, given that the Itanium has a population count instruction - and at least *nibble* matrix multiply - perhaps there are export control issues operating here. But then, if only the U.S. could use the *new* standard microchip, while the rest of the world had to hobble along on AMD chips, that would give the U.S. a compebreastive advantage.

Now, it might be that Intel could do better than the Itanium. I would very much welcome them trying to do this. But EM64T isn't better, it's only more compatible.

Besides population count, though, I like the fact that the Itanium can be switched into big-endian mode. That means that the Macintosh could be ported to it, and yet *stay* incompatible with Windows PCs, and retain access to its old data formats. If the Itanium were "the standard chip", then, Macs would not be dependent on some other company keeping up with Intel, the way they are now.

John Savard