No, not today, not tomorrow, not next year. But next decade?
People are beginning to ponder, and maybe stew a bit, too, over what all this Fusion/Torrenza stuff means. Here’s one example of that.
That article concludes with the unsettled feeling that, with a few exceptions, this really doesn’t make a lot of sense for the PC world.
And I think they’re right, but not in any sense they meant it.
I think the answer is: it’s not meant for the PC world.
You Can’t Beat Intel At Its Own Game
The last few years has been exhilarating for AMD fans. Viewed strategically, though, it’s been a lot less fun.
The last few years can be described as follows: AMD tried to beat Intel at its own game, gambled on a few technologies (onboard memory controller, SOI, x86-64 and Hypertransport) to build a better mousetrap, and did so just when Intel was pushing its second straight architecture to destruction. It also handled the media infinitely better than AMD ever did in the past.
The stars don’t align better than that.
And what did it do? It got them about five marketshare percentage points. It gave the Evil Empire a black eye.
And now the Empire is striking back, and AMD’s learned a few things about Intel as a result. It will swallow its pride and copy AMD, so any edge Green comes up with will only last a couple years. It took a while to get the ship going in a different direction, but once that happened, Intel is now going to use all its wherewithal to pound AMD into the dirt, and do a few things differently in the future so that AMD won’t get a chance to repeat the last few years again.
Intel created the microprocessor. It invented the microprocessor game, which is all about big resources and mass production and economy of scale and “I got billions to throw at problems and you don’t.”
And so long as that’s the game, AMD can’t win it. The most they can do is survive and occasionally give Intel a sucker punch.
So what can Green do?
Change the Game
Maybe the quickest way to describe this is, “Yes, Intel can build a 16-core-all-running-at-6GHz 22nm CPU for you in 2011, and AMD can’t.”
“But why would you need one in your phone or TV?”
If you look around even today, the desktop is a dinosaur. It dominated in its day, and it will never die out, but its days of dominance are doomed.
That doesn’t mean the New Era of Notebooks, either. They’re more like raptors in this picture. Again, they’ll always be around, but they won’t rule, either.
Instead, we’ll see the CPUs migrate to other places, maybe every place. Just how and where is still unclear; something will be in your phone, something will be in your TV, eventually, maybe something will be in your T-shirt, and a thousand places inbetween.
If you want to know where they’ll end up, you’ll just have to keep living, but it’s probably pretty safe to say they ain’t going to be no sixteen-core CPUs running at 6GHz each. It’s probably just as safe to say the one in your TV set isn’t and shouldn’t be the same as the one in your phone or T-shirt. They all may do a few of the same things, but it won’t be the same CPU trying to do everything.
You’re not going to be hooking up an 1100, 800, or even 500 watt power supply to your phone, or typing on your TV or T-shirt, either. The issue will become “How much computing power do you need” as opposed to “How much can you get?”
Since there’s likely to be CPUs all over the place, rather than one or two, you won’t expect as much from any particular one. You probably won’t use your TV to create multinational databases, and you probably won’t use your phone to do spreadsheets. If you can cut things like that out, it makes circuitry (and programming) much simpler. More importantly, if you can make the things you’d most likely do with TV, phone or T-shirt really easy, who needs that big or small ugly box or multigigabyte OS?
Futurists have been talking about this for years, and that day isn’t so far away anymore, but what is Intel talking about? Feeding the dinosaurs and raptors.
The point is not “Who can build a better PC CPU?” but rather “Who is going to be using a PC ten years from now, and what will everyone else be using?”
Over the last couple years, AMD has been talking about, well, “AMDEverywhere,” sticking their stuff inside anything that will take it.
Fusion/Torrenza/whatever is how they’re going to do it. AMD is hoping, praying that flexibility will matter more than anything else next decade. Give each manufacturer his own custom optimized computing thingee while Intel is sixteen-coring, and that might give AMD a rather big edge in the new markets that will matter (and, not so coincidentally, wipe out all those economy-of-scale advantages Intel has).
They’re probably also hoping that Intel won’t be so willing to give up the old game so fast, either. Abandoning the PC as we know it isn’t likely to be as easy for Intel to do as adopting x86-64.
Will they succeed? Who knows, but it’s not like they’re going to do any better playing the same old Intel game.
So when you look at Fusion/Torrenza, don’t think PC. Think outside the beige or black box. Think phone or TV or even tee.