Intel’s Architecture Shift Could Be a Shaky Platform

It’s hard to argue with success. Take a look at Toyota, for example, which builds roughly a zillion different car models (or so it seems) based on the Camry platform.

There’s the Camry itself, the Avalon, the Highlander SUV, the Lexus ES300 and the Solara. Each drives just a little differently, but each also builds on a solid automotive design foundation.

Perhaps more by accident than strategic planning, Intel created a runaway success with its Centrino mobile platform.

In fact, Centrino was so successful, it’s clearly starting to color the decision making at Intel.

Centrino was a bright spot at Intel when the updated Pentium 4, code-named Prescott, came up short on performance and long on thermal output.

I’ve even heard the term “Desktrino” emanate from the mouth of one Intel employee.

The company has always been stronger at marketing than some pundits believe, but at its heart, it’s been an engineering-driven company. But that seems to be slowly shifting.

Whatever Intel’s upcoming new architecture offers in terms of technical accomplishments, it’s also looking like Intel will be wrapping a lot of marketing bubble wrap around it.

The whole “platform” aspect involves not just CPUs, but core logic, networking, security and other building blocks, in an effort to extend the Centrino concept. Will it work? Who knows? To paraphrase that great philosopher Yoda, always in motion, the future.

It’s good that Intel is tacitly acknowledging that the P6 architecture was a good one, and will be incorporating some of those ideas into Merom.

It will be interesting to see what new ideas are incorporated. Intel’s track record in new processors’ architectures hasn’t been all that stellar lately, given the problem-plagued Prescott core and the less-than-rousing success of Itanium.

Intel really needs to hit a home run on the fundamental architecture itself, especially if it’s betting on a “one-size-fits-all” platform strategy.

Read more here about Intel’s plans for a new processor architecture.

If the CPU architecture itself proves anemic, it won’t matter how good the memory controller or networking subsystem is.

If Intel’s new platform falters, it won’t have a convenient Centrino product to fall back on.

So it really is “bet the farm” time in Santa Clara. And with the AMD lawsuit and multiple national jurisdictions investigating Intel’s business practices, Intel can’t afford to be overly aggressive on the sales front.

Still, there are a lot of smart architects and engineers churning away on this, and looming in the background is Intel’s manufacturing prowess.

And I’m sure that the production guys, still smarting from the less-than-stellar 90nm Prescott conversion, want to prove their mettle as well.

Meanwhile, plucky AMD is pushing pretty hard, though not quite firing on all cylinders. For one thing, AMD lacks an ultra low voltage solution for ultralight laptops.

But its Alchemy embedded processors are garnering design wins in some market segments, though it doesn’t really go head to head with Intel’s Xscale, giving AMD additional growth in a key area.

If intense competition creates clarity of purpose, then Intel should be seeing the world in perfect focus.

Whether that means that the company can actually respond to the competition is a different issue. We’ll find out in 2006.

Check out eWEEK.com’s for the latest news in desktop and notebook computing.