The Real State of AMD

Add Your Comments

I read this piece, and I have to reply because it’s just so . . . wrong . . . about AMD and where it stands, and where it may go.

“Let’s Make A 3GHz Processor!”

The primary point of the article is that AMD would be much better off if they introduced 3GHz Hammers right away rather than wait and play cat-and-mouse with Intel.

Well, of course they’d be better off if they did that, just like I undoubtably would be better off if I got paid a million dollars a month to write these articles.

That’s not the issue. The issue in both cases is not “Should it be done?” but rather “Can it be done?”

And, in both cases, it’s pretty obvious that the reason these wonderful things aren’t being done is that they can’t be done. Joe Citarella can’t pay me a million a month and AMD can’t make a 3GHz Hammer at the moment.

There is this myth that CPU companies can sandbag their opposition. It is one of the silliest myths out there in geekland.

If there has been any incontrovertible truth about the CPU industry, it has been that any technological advance has a short period of time in the sun. CPUs are not like wine and cheese; they don’t increase in value with age, their value just rots away.

If AMD could put out a 3GHz Hammer today; they would put out a 3GHz Hammer today. There is no downside to doing such a thing. By doing so, AMD would get higher prices for its CPUs and in all likelihood force Intel’s prices down quite a bit.

But, you might say, that would show AMD’s hand to Intel’s advantage. I hate to burst your bubble, but don’t you think Intel can’t buy themselves a few Hammers, test them out, rip them apart, figure out what they can and can’t do? They certainly have done so in the past, even said so, and you would have to be naive to the ninth degree to think they don’t do so now.

Even comparatively nitwitty folks like us could figure out such a sandbag, simply by overclocking the things. If Hammers could do 3GHz under default conditions, we’d find that out pretty quickly, now wouldn’t we?

Now some might say AMD has secret process technologies they’re just holding back for the future, but frankly, if you believe that, you probably also believe the U.S. government has a fleet of alien spacecraft parked in a Roswell lot.

AMD certainly wouldn’t be doing things like making 130nm 2.6GHz chips if they had such wonderful technology in reserve. That would make no fiscal nor technical sense.

No, the simplest explanation that fits the facts is best, and when a company does things that make no sense if something does exist, but plenty of sense if that something doesn’t exist, simple logic concludes that that something isn’t there.

None of this is any knock on AMD; they’ve hit the same wall Intel and especially IBM have hit. No doubt people are working very hard to push that wall back a bit, and in the course of time, they’ll probably push it back a bit.

What is of rather more concern is that a lot of people out there simply will not believe such a wall exists, and that it’s business as usual, with major speed increases every year as an entitlement. It is no longer business as usual, it won’t be for years to come, and, insofar as the inexpensive desktop is concerned, perhaps never be again.

Other Sillinesses

The article derides the look of the AMD website, while praising the beauty of Apple’s site. Well, given that AMD sells at least six times the number of processors Apple uses, beauty isn’t all it’s cracked up to be. 🙂

The author is rather upset that a Sempron 3100+ runs more slowly than an Athlon 64 2800+. Well, the notion that a Celeron runs rather more slowly than a PIV with the same MHz rating, too, and that is why AMD recalibrated Sempron PR to its Celeron competition in Joe Sixpack land never seems to register. Then again, the author still seems to find PR a rather dubious notion.

The lack of an AMD chipset for motherboards is also found to be quite disturbing. While an AMD chipset was a welcome alternative in the days when Via was the only other (and often less than desirable) game in town, times have changed, and both Via and nVidia put out perfectly creditable mobos.

Shades of Rip Van Winkle

Look at the piece as a whole, and it’s so . . . 2000. It’s like the last three or four years didn’t happen, or at least didn’t register.

Sorry, but reality is a good deal different now then it was then. You can’t crank up MHz speed 30% or more year in and year out, and even more importantly, most people don’t care anymore. Only the mentally challenged have problems believing that a 2GHz Hammer can actually do better than a 3Ghz PIV. AMD motherboards have gotten better in the last few years.

But even that doesn’t get to the root of the problem with this article, and even my comments suffer the same problems. When you are judging AMD success or failure, the problem isn’t that outdated geeky factors are considered important and decisive, the problem is that geeky factors are considered important and decisive at all.

It’s a different world now.

Today’s World…

Today’s World

There’s something of a reality disconnect when one talks about AMD “winning” when it has about 17% desktop market, less than that in the mobile market, and about 7-10% in the server markets. Even if one concedes AMD technical superiority in these fields, it obviously doesn’t do them much good. Clearly the other 83% or 93% are marching to the tune of a different drummer.

AMD has two core problems when it comes to seriously challenging Intel, and neither of them have anything to do with technical merit.

The first problem is that the average human being has been brainwashed over the years to think Intel, so the average person buying a computer who neither thinks much nor wants to think much about computers has a built-in Intel bias.

This is the 800-pound gorilla in the whole discussion. Many buy Intel for the same reason they buy Sony televisions; the brand name has been branded into their brains, and buying that requires no additional effort on the part of the buyer.

Those willing to put more effort into the buying process can scream all they like about it, just like Toshiba fans can give all kinds of reasons for buying their product, but these folks aren’t around to listen. They’re not watching your channel.

This is not going to change anytime soon, and probably never will. This is not like Coke/Pepsi; AMD is more like Dr. Pepper/Seven Up. Even if AMD could match Intel dollar-for-dollar in advertising (which they can’t; they don’t have the money), it would take years and years to wear away that ingrown Intel Inside People’s Heads.

Realistically, the only relatively quick way for that to be erased would be huge, long-term negative news about Intel. Not little geeky faults, more like Intel processors bursting into flames and killing users.

This is not to say people can’t be convinced, but for all practical purposes, the only AMD advertising that the average person in this group gets is the price tag they see in the computer store. In the past, that has often been quite persuasive, if financially damaging to AMD.

It’s not much of an exaggeration to call AMD’s marketshare an alliance of the geek and the cheap.

Get rid of the cheap part, and you remove the prime advertising for the less-than-interested.

Making a better product doesn’t matter to this audience; they won’t know and probably wouldn’t care anyway. Convincing all the geeks you have a better product matters little more. When AMD had a bigger performance advantage over Intel than it does today, market share went up . . . five points, from about 18% to about 23%.

This is a true measurement of how much technical superiority really matters in the real world. That’s the level of influence this factor has.

That’s problem number one.

Problem number two is seen mostly on the corporate side, where AMD sales have been and continue to remain an anchor on the company’s marketshare. That is primarily due to the generally held perception that AMD just isn’t as reliable or dependable as Intel.

While AMD is trying to change that perception, there’s a difference between talking the talk and walking the walk. While there’s been some improvements here and there, overall, it still looks like the same old AMD.

Hammer rollout has been a disaster for any party depending on it for their own products. Just to illustrate, imagine if Dell had signed on to the Hammer bandwagon a year ago and decided to make desktop Hammers co-equal with PIVs. AMD couldn’t have delivered enough product to Dell, period. True, Intel had problems delivering Prescotts initially, but there was a just-as-good supply of Northwoods available, and Intel did get Prescotts ramped up pretty quickly.

Right now, the product line looks like a junkyard, with products competing more against each other than against Intel’s, and it will only get more cluttered in the year ahead.

It’s not good enough to come out with a better mousetrap. Corporate buyers consider other factors, too, and this is where AMD has been and continues to be weak.

The geeks all love Opterons, but so far, Opterons are the greatest story never sold. Yes, they’ve gone from nothing to 7% in a year-and-a-half, and probably will double that in another year-and-a-half, but that still gives you only 15%.

That’s still Dr. Pepper numbers.

Made By Geeks For Geeks

If you had to combine the two big reasons why AMD isn’t more successful than it is, you could say that AMD is still a company whose unofficial motto is “Made By Geeks For Geeks.”

The problem is, there aren’t that many geeks out there. It’s not 1982 any more. Joe Sixpack and Suit rule now, and he’ll determine what gets built. He views a computer as a necessary evil, a far-too-troublesome TV or stereo or Xerox machine, not a wondrous challenge.

The company that does more to make those people feel happy and comfortable, one way or the other, on their terms, the one that lets them do the least bit of thinking and make the least bit of effort in buying and owning a computer, is going to get their business.

And so far, that’s been Intel, whether it’s playing that ding-dong in a million ads, coming in a Dell, or promising easy Centrino wirelessness. Forget about reality or merit, perception is what counts here.

That’s why Intel rules (for that matter, that’s why MS rules, too). If AMD (or Linux) ever hopes to seriously challenge for overall market leadership rather than do well in certain niches, they’ll have to address these two main areas.

And I don’t think they can.

It’s somewhat like being a Republican in New York City thinking he can carry the city for George W. Bush in the upcoming election. Any such person is delusional. It doesn’t matter how good your arguments are, or how brilliantly you campaign; or what the merits of the case actually are. It’s not going to happen. New York City hasn’t voted for a Republican President in eighty years. You’d have to change the ingrained (and in many cases, ingrained to the bone and beyond) biases and beliefs of a few million people. The only way you could do it is if John Kerry got a bad case of post-traumatic stress disorder and started gunning down people at campaigns stops left and right, and even then, I’m not so sure about the end results. 🙂

That’s the real problem AMD faces, not a speed bump or a chipset.

Ed

Leave a Reply

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