Gordon Moore's 1964 prediction was based on the idea that we could shrink the size of components indefinitely. (Oh, and yes the title of this is a pun.)

If you limit your look at Moore to that one point, last week's announcement by Intel that it will change the way it looks at chips is, indeed, Gordon Moore's Last Sigh. (What, you haven't bought the book yet? What's wrong with you?) That method for increasing chip speeds is, henceforth, inoperative. (Picture of Gordon Moore from CNN.)

But Moore's Law is going to keep on keeping on. Moore's article cited a method for making chips faster, but Moore's Law itself was really a challenge to the industry, to keep those improvements going. Here's how the challenge will be met:

First, what has done in Moore's ideas about line shrinkage isn't cost, and it isn't physics. It's heat. Intel has hit a "thermal wall." In typical Intel fashion, CEO Craig Barrett was extremely forthright about it:

"Our strategy is to go 200 miles per hour until we hit a brick wall," said Craig R. Barrett, Intel's chief executive. "At the same time, we go down parallel tracks to make sure we're ready to move in new directions when we have to."

In this case, Intel is going to keep Moore going by dumping Von Neumann. John Von Neumann (1903-1957) pioneered the "stored program concept," the idea that programs would execute one instruction at a time until the program was complete. For decades, this was in fact, the only way that computers could operate. (For more on Von Neumann, and his principles, read Keith Devlin's excellent December 2003 column on him.)

Parallel processing, on the other hand, had many fathers, but if you wish to pull someone from the pile my vote goes to Danny Hillis. Hillis co-founded Thinking Machines, a company that pioneered the concept in the mid-1980s. The idea is that work is parceled-out among several machines, then re-assembled. The time lost on the front end is made up on the back end. (See what Danny's doing now here.)

Parallel processing works. Parallel processing made SETI@Home possible. It made Google possible. It let some guys build a super-fast supercomputer out of game machines.

And now, Intel is going to begin go apply those lessons directly into chips.

Intel is going to face real competition now, for the first time in ages. IBM may be ahead of it in this area. AMD may be ahead of it. It's to Intel's credit that it chose to bite the bullet now, when it's possible to make a turn, rather than waiting until its problems were more-obvious and more difficult to correct.

But it has which means Moore's Law, like love on the Titanic, will go on-and-on. Good news, indeed.