Win, your cynicism today would normally amaze me, but I think I am beginning to accept it, coming from you. As you should already know, Compaq and Itanium aren't the only ones supporting Itanium. It was nearly the entire industry, including IBM, Dell, Unisys, NEC, Bull, SGI, and others. Though HP and Compaq are very important to Itanium. After all, both companies are giving up their RISC architectures to embrace Itanium; not because they were both force fed by Intel, but because they felt it was the best architecture for the future.
Frankly, I can't understand your complaints about Itanium. Yes, it's a big die, but so are all the other RISC architectures. Yes, it dissipates a lot of power, but so do the other RISC architectures. Sure it scores low on SPECint, but McKinley promises a 70% improvement, which puts it clock for clock better than the Alpha EV68. Sure it doesn't have a lot of software right now, but half the industry has announced software support, and new press releases flow every week about new commitments. And the most important: what other competition is there?
Sun is Itanium's only real competitor, but the UltraSparc is slow, it's losing momentum, and it has no support outside of Sun. IBM's Power4 will be larger and lower volume than their already humongous Power3. Hammer is a joke, and no one is supporting that. It will eventually end up being AMD's high end desktop chip, and it will probably have a follow on version to the Athlon MP, which isn't an Itanium competitor to begin with.
Without Alpha or PA-RISC, there is a ton of opportunity for Itanium, and Sun can't possibly gain all the ground that Alpha and PA-RISC have given up. Itanium will fill in those holes, because Compaq and HP will have many loyal customers willing to make the transition. And even if they weren't, there isn't much else, and plenty of other Itanium vendors will be there to pick up market share.
AMD was never in the running for this, and not even a miracle SOI chip with 64-bit x86 extensions is going to convince any OEM to think twice about supporting it for any of the high end markets. You're in denial if you think Itanium will still fail.
wanna_bmw |
