As technology enthusiasts, we're usually excited by the latest breakthroughs in IT. So today's launch of Intel's Montecito chip should have had us salivating. But it didn't.
This lack of passion is partly because the dual-core Itanium 2 processor is almost a year late, a delay that only helped rumours about lack of support for the platform from outside, even inside, Intel. Even had the chip been on time, to speed and with its full feature-set, that may not have been enough to guarantee success.
Yet Montecito is a big step forward. The top-end 9050 chip runs at 1.6Ghz, with a cache of 24MB, all with reduced power consumption over the previous part. Intel says that Itanium 2 has double the performance for 80 percent of the power: even with the vendor benchmark optimisation field in operation, those are good figures.
That's a lot of bang — for a lot of bucks. That 9050 is going to cost $3,692, in quantities of a thousand. At that rate, Intel's eight vendor partners won't be selling their Montecito servers on the cheap: it's not coming out of its top-end supercomputer niche in a hurry.
Over-specialisation can be dangerous. Intel's Itanium vision all those years ago was to create a thoroughbred 64bit chip to power servers and high-end workstations, perhaps even to ease out the x86 chips over time.
Things didn't turn out that way, thanks to the performance increases achieved on the x86 platform by Intel, sometimes despite itself — as when attempts to protect Itanium by refusing to add 64 bits were bypassed by AMD.
Pat Gelsinger insists that with Montecito, Intel has the perfect chip for 'mission-critical computing'. This claim would have more substance if IBM and Sun were on-side: it's not clear that Itanium will evolve to interest them.
But even if these dual-core Itanium 2 machines do deliver what the top-end user needs, its going to be a continual struggle to justify the development dollars to match the constant advances being made by the architectures IA-64 was once intended to supplant.
Like someone running up a flight of stairs to avoid a fire, the launch of Montecito may simply delay the inevitable for Intel. At some point, there will be no higher ground to which to retreat — and that's overlooking the fact that the very top of the food chain is already x86.
There's only one way that we can see Itanium 2 being a box-office success. If no more bang, then less bucks. Build an ecosystem that makes Itanium developers — that endangered species — feel like getting up in the morning. Performance at a great price is today's mantra. If Intel and its supporters can deliver, it may still have the last laugh.