AMD's future: an opinion

[wtd]

After skimming a recent Arstechnica article on AMD's plans for the future, I cannot help but think that they are headed in the wrong direction.

It would seem their plan is to develop truly flexible systems, that allow for CPUs and chipsets to be highly customized for different customers. Technically speaking, this is a brilliant idea.

As a business plan, though, it is my opinion that it ignores history. Success has always seemed to be in pushing a given technology toward ubiquity. Customers tend to like this as it gives them not less confusion, but a base set of features to fall back on that is everywhere, and still relatively decent from a technical perspective. Let's consider some examples.

Apple computer pushed the command-line out of their computers when they launched the first Macs. This prevented developers from continuing old command-line habits. They made the GUI ubiquitous.

Intel pushed the x86 chip. Sure, there were and are better chips, both in general and for the various niches that make up computing. But they succeeded because they created a pretty darn ubiquitous instruction set. They could not have done this if they had offered the best custom chip for every niche.

Microsoft created the Win32 API. It's an atrocious thing, but Windows succeeded because of it. Write a Win32 app and you'll be able to run it on somewhere in the neighborhood of 90% of the computers in the world.

AMD saw a real kickstart in its own business when they pushed 64-bit CPUs really hard. Did consumers need $500 Radio Shack HP systems with 64-bit processors? Heck no, but they got them.

Now AMD wants to abandon this path to sell very specialized chips. Sure it's a beautiful, technically elegant idea. As a business, though, AMD has the wrong idea. The right direction would be to take these neat features, cram them onto a chip, fab as many as possible, and sell them to everyone. If someone doesn't wish to use the features it gives them, they don't have to. But, if they decide they do, they won't have to worry that they bought into the wrong platform. That's the way to ubiquity, and that's where hope lies for AMD.

[md]

I think amd went wrong when it bought out ATI. On the other hand their 4x4 systems aren't nearly as impressive as 4 core intel systems. For all the hype about how intel just put two dual core chips on one package, intels solution performs much better then AMDs.

Oh well, it'll be a shame to see AMD go back to being the minor CPU manufacturer. Just as I almost started considering getting an AMD chip next computer upgrade.

[Tony]

who's AMD targeting with those specialized chips? For PCs, the features will be OS dependent anyways (the system ether takes advantage of it or not). Having different sets would only make sense for propriatory hardware systems, say... game consoles.

Though in such a case, I don't think that PS3's 8-core system is due for an upgrade any time soon. Lets face it, hardware is way ahead of the software that's feasable to write. The only thing coming close to taking advantage of all the resources and features are those multi-million dollar budget video games, and it's said that noone will even bother pushing PS3 to its limit.

Which I suppose brings up a separate debate - what future is left for hardware inovation.. going faster just doesn't seem to make as much sense as it used to.

[wtd]

I think the thought is that they'll be targeting the High Performance Computing crowd.

[apomb]

well, md, and others, including myself are correct in assuming AMD's buy-out of ATI was ill-fated, they've already been slashing prices, to combat Intel's strong Core2 performance. link
although, this isn't exactly DUE to the merger, it certainly doesnt look good for them.

[Andy]

i honestly dont understand why you guys think it was a bad idea for AMD to purchase ATI. infact, i think it was a pretty good move. Now they can compete with the entire intel system, instead of just the cpu

[wtd]

As it happens, I am not criticizing that purchase.