Sledgehammer and the Future of Computing

Add Your Comments

I wish the mainstream computer media would get a clue.

I once worked for a place when a new development occurred. A meeting of all interested parties was called, and there was panic. Not panic because of the development, panic that the representatives had to think. One put it very well, “Nobody’s told us what to think!”

AMD apparently had a few things to say about Sledgehammer. See here. Nothing that hasn’t been known for months, but apparently a revelation to the reporter.

Sledgehammer is AMD’s 64-bit processor. It is due out late in 2001.

Why should I care?

Right now, the AMD/Intel “war” is just a bunch of love taps. AMD can sell every single Athlon it wants and not even touch Intel’s total production numbers. Until the end of the year, increases in Athlon production pretty much only offset decreases in K6-2 production.

AMD will only start making significantly more chips than it has in the past during the 4Q. Even then, this shouldn’t fluster Intel too much since AMD looks to be meeting demands Intel couldn’t meet anyway. It’s more profitable to keep prices high in a time of heavy demand, and not worry about a couple percentage points in market share.

None of this is a knock on AMD at all; even David took some time to figure out how to knock out Goliath. Unlike David, though, AMD is in no position right now to knock out Intel, but it’s progressing nicely towards the day when it could.

The recent announcement that AMD was going to build another fab is a sign that they might not want to be a number two forever.

Sledgehammer is the other. The name is appropriate. Athlons can at most annoy Intel. Sledgehammer could kill Intel as we’ve known it.

Here’s why

Over the next few years, we’ll be shifting from 32-bit to 64-bit computing. Intel has been planning for this for a long time, and Itanium is the first Intel product to do this. More are coming, most notably McKinley in 2001.

Intel doesn’t expect 64-bit processing to reach the desktop for a long time, roughly 2003. Intel wants to make a break from the old x86 standards and limitations with its 64-bit products. Not that the early ones won’t run 32-bit products, there will be provision for that, it just won’t run them too quickly.

AMD has taken a much different approach to this. Their 64-bit processor is not a revolution but a further evolution of the x86 standard. While this appears to hurt performance to some degree compared to Intel in 64-bit processing, it allows for full, fast compatibility with 32-bit applications. AMD also plans to beef up hardware FPU even further for Sledgehammer; something Intel appears to have given up on in favor of software solutions.

AMD also plans to bring 64-bit processing to the desktop a long, long time before Intel, and they are aiming at the desktop early on with reasonably affordable machines. In contrast, Intel’s 64-bit chips won’t be cheap.

So AMD may very well have 64-bit products the average person can buy a year or more before any Intel equivalents, and they’ll run current software better than what Intel has planned. This is not a “me-too” product; it’s a fork in the road. Athlon vs. Coppermine or Willamette is just a beauty contest. Sledgehammer is a bid by AMD to determine the 64-bit war on its turf.

And they might just succeed.

Put up a good product soon, and it often beats a better product much later on. Especially if the good product runs your current software better right away, while the better product later on effectively makes you toss out what was perfectly good software. AMD may have a winning formula here

Not saying AMD will win, or probably will win, but it could win.

All the ZDNet guy could do was wonder about the number of 64-bit applications available. That’s not the point. If it runs your old stuff better, there’s time for the new stuff to develop. It’s only when the new CPUs don’t run the old stuff too well that you need the 64-bit apps on launch.

Intel had better watch out.

Email Ed


Leave a Reply

Your email address will not be published. Required fields are marked *