Taking Exception

There’s an article over at the Inquirer about 64-bit.

Since the article raises some important points, and even more importantly, there’s isn’t any important happening at the moment :), we’ll address
these point by point.

If nothing else, let this be an example of civilized disagreement, which can hardly hurt given what we usually see in cyberspace.

(Text in bold italics come from the Inquirer article)

THERE’S SOME fine old mudslinging controversy going on over at Realworldtech, Overclockers and some other places over whether anyone in their right mind would want a 64-bit PC.

No, no, no. At least here, the controversy is over whether or not everyone in their right, wrong, or no mind is going to scurry over to 64-bit any time soon, especially without Intel pushing the world along.

At least here, there’s no argument that some people will want it or even “need” it. We simply disagree that most or all will do so, and we have doubts that the uncaring or unknowing can get swept up this time around due to the needs of a relative few.

One of the arguments being advanced is that no-one needs this level of technology and there’s some merit to this one.

All the additional power that Intel and AMD have so generously given us out of the kindness of their hearts over the last few years hasn’t increased our typing speed at all, we must admit.

Again, no one is saying no one needs it. What we’re saying that most don’t need this any time soon, and that “business as usual” isn’t, or at least isn’t going to be shortly.

The prime reason why we think “business as usual” won’t be is due to the growing belief that current PCs are powerful enough for the tasks the average person uses them for. This belief has even taken firm root among the overclockers.

My God, if even the lunatics in the asylum are saying, “Enough,” this ought to tell you something.

These people aren’t technological Luddites. They just want something else out of advancing technology rather than sacrificing (sometimes literally) burnt offerings to the Goddess of Power.

Instead of faster, faster, faster, faster for no apparent personal benefit or purpose, they want small, quiet, cheap, inobtrusive (or at least one or more of the above).

A note from Paul DeMone to Paul Otellini at Intel and to Hector Ruiz at AMD would no doubt have top priority at the meetings of the next boards of directors.

Mr Otellini would realise straight away that the Intel business model was just a crazy idea, it would stop investing in $2 billion fabs, halt the progress of the Prescott 90 nanometer chip, lay off 40,000 people and just carry on selling 3.20GHz CPUs. Or Celerons, maybe.

No one needs a 64-bit chip, so Steely Paul would patiently explain to his shareholders and to the financial analysts that the billions of dollars spent developing the Itanium wasn’t necessary any more. . . .

In fact, there is a solid argument that we don’t need all this processing power for our PCs, but that’s not something that’s music to Otellini’s or to Hector’s ears.

Just to get it out of the way, again we have this curious inability to recognize that there’s this notion called “some” which exists between “no one” and “everyone.” Itanium is meant for some people, not everyone. Just like Opterons are meant for some people, not everyone.

Not everyone likes strawberry ice cream. Does that mean you stop making it? No, if there’s enough of a market willing to pay the price to make strawberry ice cream production worthwhile, you make it, and you are not the least bit concerned that everyone doesn’t buy it.

The major point, though, is that Mr. Magee is quite correct in saying that neither Intel nor AMD want to hear this.

Indeed, this will destroy Intel and AMD and Microsoft and many others as we know them, or just destroy them, period, since none of the above can charge one or two hundred dollars for its products in a one or two hundred dollar computer.

And yes, these dinosaurs faced with extinction will fight this any and every way they can.

And they will eventually lose. Or morph into the “enemy.” Period. The only question is whether it will be sooner or later.

Keep in mind that even “sooner” won’t be around for quite some time. The battle hasn’t even started yet, and probably won’t in earnest for a couple years to come. Given that, “sooner” may be five years from now; “later” may be ten years or a bit more.

It could be music to Via’s ears however. If one of its screaming Nehemiahs can deliver all most people would need for a few cents in a tiny lunchbox (that’s small form factor in marchitectures), why don’t Intel and AMD pack all their 64-bit troubles in a big carpet bag and march off and do biotech or something?

Precisely. That’s the big story for the next decade or so. Which version of the mainstream PC will win out?

The PC has never been solely about technology. It has always been about affordable technology, and the definition of “affordable” changes only one way, downward.

Like any other technology, it started off expensive, and the cost reserved it to a relative elite.

Like any other technology, a lower price meant a bigger, more inclusive market, so those who produced cheap and good enough beat those who made expensive and better. This is why Wintel won and Apple lost.

In the twenty-plus years since IBM introduced its PC, (inflation-adjusted) prices have dropped by well over 80%. This has made them affordable to the average people in the developed world.

And what has occurred? We’ve gotten saturation in a few parts of the developed world; most of the rest will follow in a handful of years.

Where are the next growth markets? Middle-income countries. Places where for the average person, price matters, and right now, the price is too high.

Technology Plateaus

There seems to be a belief that technology expands indefinitely. It doesn’t.

Once technology reaches “good enough” levels for the mainstream and you hit the point of diminishing returns, cost becomes the decisive factor in that technology.

Thirty, thirty-five years ago, many if not most people thought that supersonic flight was the next inevitable advance in air travel. After all, the whole history of air travel had been “faster, faster, faster.” Why should that change?

However, when you get on a jet plane today, you travel little faster than your father did (or would have) at your age. How can this be? Why aren’t we all flying around at 5,000 miles an hour today?

The reason why we don’t is that cost became more important than performance, and for most travel, 500-600 miles per hour was fast enough, or at least people weren’t willing to pay the cost of flying twice as fast.

What became more important than flying fast was flying cheaply. At least in the U.S., it’s a good deal more likely that you’ve flown in an airplane for non-business travel than your father did at your age, and far more likely you’ve done it pretty often compared to your father. That’s simply because it’s much cheaper to fly now than it was in 1965 or 1970.

Now there are a relative few for whom a six hour transatlantic flight is just too much, and are willing to pay a whole lot more to save a few hours flight time. These people take the Concorde (which was the only supersonic passenget plane that got off the ground in that earlier age).

Those who did fly the Concorde no doubt pay a good deal more than they would have had supersonic travel become mainstream, simply because the cost of R&D on such planes would have been spread out among a lot more people.

I’m sure they’re not happy about that, but so what?

It didn’t work out that way. The people who wanted low prices more than high speed prevailed.

How many Concordes cross the Atlantic as opposed to slow Boeings and Airbusses?

The computer world is starting to look a lot like air travel did a generation ago. There’s certainly a constituency that wants faster, faster, faster.

However, there’s a much bigger constituency that is saying, “They’re fast enough, now we want cheaper.”

The airlines at the time didn’t want to hear “cheaper” any more than Intel does today. Nonetheless, there were those who heard the call and delivered, and the earlier dinosaurs either adapted or died.

And those who wanted faster . . . have the Concorde. A little, expensive niche.

Unless we see some tremendous software advance that most average people will want which will chew up a ton of CPU power, it’s hard to see how the personal computing world won’t split into expensive workstations for those who need and will pay for them, and cheap small boxes for everyone else.

That doesn’t mean the end of progress. What it will mean is that progress will be defined through different parameters.

For instance, you could still have a multibillion dollar fab pumping out 65nm chips a few years from now, but those CPUs will likely have fewer transistors, be cheaper to make, and be cooler-running than they would otherwise. Speed would go from the only criteria to just being a factor.

Making factors other than speed a big say in CPU design doesn’t preclude advances like 64-bit (indeed, it just might encourage them if x86-64 provides a bigger bang for the buck than other improvements, look at gaming machines), but if a Via chip running x86-64 two years from now, it will be quite a different beast than an Opteron workstation.

Going small and cheap will not end progress, it will just change the pace and direction of that progress.

64-Bit, Just A Pre-Fight Skirmish

When I write articles about this, the emails often say, “Why are you against 64-bit processors?” like I personally hated them.

I’m not personally against 64-bit processors at all. I just don’t think 64-bit computing is going to set the mainstream computing world on fire any time soon. I think it’s going to become the hardware equivalent of Linux; pretty popular among the server people, but a niche group on the desktop (OK, a bigger niche than Linux). AMD will not conquer Intel, much less the world with them.

Mind you, this is not the same as saying Hammer will be a failure. I think most people who buy Hammers, especially the cheaper ones, will end up running in 32-bit.

I don’t think Joes Sixpack and Suit are going to drop everything and go running like lemmings to AMD to get them right away. I don’t even think they’re going to go running even if Intel goes x86-64.

The issue is not “does Joes Sixpack and Suit want 64-bit;” it’s becoming “Does Joes Sixpack and Suit want a new computer at all, and if so, will cheap and small beat fast?”

It is always a bad idea to confuse what you want to happen with what you see happening. Especially when other people are doing the “happening.” Just look where that’s gotten Saddam Hussein? 🙂

Nor do I think this malaise is going to be an AMD-only problem at all. It’s much bigger than that. Intel will face exactly the same problem with its processors, 64-bit or not. Celeron sales keep creeping up as a bigger and bigger share of Intel sales.

Right now, Intel and AMD are Twiddledum and Twiddledee when it comes to where CPUs are going. There’s no real alternative (a hyper-cheap Mac could be, but we know that’s not going to happen) yet, outside of buying the cheapest processor one can (which is what people are increasingly doing now).

Via has a different vision, and I think in the long run the correct one. They aren’t quite ready yet, but they should be in a couple years.

Of course, this is Via, so victory is hardly inevitable. They could blow their chance, or end up showing the way for some bigger dog to win the war. Or maybe the old CPU dogs will learn new tricks.

But it’s hard to see how the typical PC in 2015 will look or cost anything like it does today.

Email Ed

Be the first to comment

Leave a Reply