The End of the Era of Speed

Intel plans to junk the desktop PIV path in favor of its notebook series.

This is an admission by Intel that long-term, it can no longer conduct business as usual because it can’t handle the resultant heat and power problems.

In other words, you can’t just crank it up when the crank is broken.

It is the end of an era. The end of the Era of Speed.

Performance ¹ Speed

It’s not that we’ll never see a faster processor; it’s just that the monomanical drive for raw speed is going to take a back seat from now on to functionality and power consumption.

That’s a revolution.

Intel’s decision marks the start of the CPU industry turning in unto itself and pursuing different goals. Few of those new goals will have “faster, faster, faster” attached to them.

Stop obsessing about speed, and start obsessing about functionality and power, and pretty soon you’ll be likely to have a real “computer in a chip” that you can stick in any device you feel like. Like an HDTV. Or a phone.

Once you start thinking that way, you start looking upon CPU technology a whole different way.

For instance, if like Intel you develop technology to greatly reduce power leakage, rather than thinking “we can go faster,” you think “we can reduce power consumption not only for quieter, cooler, more portable computers, but now we can stick them into a lot more other places.”

Within the next few years, the logic of fab design and processes shrinkage will leave even a single fab capable of making all the world’s PC processors. Capacity is growing geometrically while demand grows just arithmatically. Cost of manufacture will drop into the single dollar digits. The CPU folks are going to need new markets, so “stick them anywhere” is going to sound like a really good idea.

All together, that means a second revolution in computing that will start becoming quite evident towards the end of the decade.

For those who really need more computing power, it will be available. Most likely, it will be a matter of clustering these tiny little processors together in specialist machines running specialist programs. On the low end, you’ll have gaming/multimedia specialist boxes. On the high end, you’ll have your desktop supercomputers.

But what’s going to happen is that the ancillary technologies required to make these specialist machines go fast will often have little to no benefit for mainstream computing. Take multicores, for instance. There’s no point to multithreading word processing or web browsing. Yes, two CPUs will give you a smoother ride than one, but do you really need four?

More importantly, once the computing elements get cheap enough, the idea of a single all-purpose PC for computing becomes silly. If you ask real people what they don’t like about computers, the leading answers are “they’re too hard to use and/or they break down too much.” Such people would much rather have six $100 easy-to-use single purpose computers that don’t break down to one $600, hard-to-use all-purpose computer that does break down.

This revolution will be a big net plus for the world. The winners will be the average person who only needs so much computing power. He or she will get it for next to nothing, and end up with a better experience.

The big losers will be the speed geeks. Those relative few who have real super needs will have to pay much higher prices for their superneeds rather than having the rest of the world pay for their R&D.

Gaming will inhabit both worlds; good will be reasonably priced, better will cost a lot more.

Time For An Attitude Adjustment

Intel isn’t going to do this until 2006-2007, but the writing is on the wall.

And that writing says, “If you’re expecting 5GHz tomorrow, and 50GHz ten years from now, you ain’t gonna get it. That record is broken. We’re not going to play that song anymore.”

If you’re saying to yourself “What about AMD? What about IBM?” this article ought to be an eye-opener. They’re all singing the same tune.

You’ve got a few years to get used to it.

This is going to be traumatic to a lot of people, and especially to those who make these things. It’s so against the prevailing culture. When your whole life and focus have been dedicated to “faster, faster, faster,” it’s going to be hard to shift gears.

Nonetheless, it’s going to happen, and you don’t adapt, well, the dinosaurs didn’t, either. There will be many dinosaurs going extinct five-ten years from now.

Will this mean the end of overclocking eventually? Wrong question.

We need to realize that if we continued to go down the same path, eventually our world would die a heat death. Not all would perish, but the number of people willing to freeze themselves into an overclock is but a small proportion of today’s overclocking population.

Instead, we’ll face a new world and new unknowns. Some will be good, some will be bad, most if not all of them will be new and not even imaginable today.

But somehow, someway, if it’s at all doable, many of us will be around to get it done. The processors, the architecture, the technologies will change, but the human desire to push the envelope will remain the same.

Ed