AMD - "Out With The Old BS, In With The New"
A very astute assessment of AMD's screwed up Hammer program.
-SZ
http://www.overclockers.com/articles703/
"Out With The Old BS, In With The New"
Ed Stroligo - 2/8/03
page 1 of 4
--------------------------------------------------------------------------------
There's an interview with some AMD honchos over at c't, and it's essentially what the title says.
AMD has apparently decided that they can't BS anymore about their manufacturing problems like they did less than a month ago (talk about being used). So now they've been admitting both in the c't article and here.
I guess it finally dawned on them that people can still blame them for no product even if they don't know why there's no product.
It's much better to blame someone else for that, and AMD looks like it found an ideal bogey man: Microsoft.
Now the line is, "Oh, we're all ready, it's just those slugs in Redmond that are depriving the world of our revolutionary new product."
I must admit, this replacement appears to be far higher-quality BS. If they can pull this one off, this will approach Intel devaluing the worth of a MHz in the PIV design, then having people blaming AMD for PR ratings.
So long as MS doesn't come up with the product, AMD has an excuse for not coming out with the product that they can't be blamed for. If September rolls around and there's no Windows x86-64, just delay again thanks to those dead slugs at Microsoft.
What's MS going to do? They aren't going to stop development and leave that market wide open to the Linux people. If they stall a bit to show their disapproval, it really does make them look like slugs.
They probably won't speed up development because that will make them look like AMD's servants (and converting all those Windows files and drivers from 32 to 64-bit is a ton of donkey work).
So they'll probably ignore AMD's comments and get the product out sometime in 2004 as they initially planned. Which will make AMD perfectly happy because that's just what they want to happen.
page 2
--------------------------------------------------------------------------------
A Twisted Web They Weave
From every indication out there, AMD can't make PIV-beating Athlon64s right now. Period.
If they could make Athlon64s that could run as fast as the latest Athlons, we wouldn't be hearing diddlysquat about Microsoft; AMD would be cranking Athlon64s out.
Remember, Athlon64s were supposed to beat PIVs in 32-bit mode. That was the big selling point a while back: You could keep all your old software and OSs and stuff and still get world-class performance, and when you decide to go 64-bit, you get a performance bonus.
Remember that?
If there were really something to this 64-bit OS stuff, you'd think AMD would delay the server Opterons, too. After all, it is servers that can use 64-bit right away, if only to increase the amount of memory that can be directly accessed (and keep in mind that only big servers use that kind of memory).
True, there are 64-bit Linuxes around, and it's a much more level playing field between Linux and Windows than there is on the desktop, but I'm sure a very hefty chunk of Opteron servers will begin life running 32-bit Windows server OSs.
Well, things didn't work out that way, and AMD's partnership with IBM indicates that this is no small problem with an easy fix. IBM isn't blowing the doors off the GHz barriers with its SOI chips, either.
Even if the two companies' scientists come up with a quick solution, it's hard to see how such a find could be taken out of the lab and into finished products before the fall, which is just what AMD is saying at the moment.
You need SOI to build a .13 Hammer, otherwise the damn things would melt. Maybe SOI won't be so necessary at .09 micron, and I'd lay money AMD has looked or is looking into an SOI-less .09 Hammer as a contingency plan if SOI falls on its face.
Even if you need SOI for .09; holding off until .09 gives the scientists more time to come up with something.
So what appears to be really happening is that AMD is trying to get its SOI act together, blaming Microsoft for the delays, and using the nonexistent Windows x86-64 as camouflage to hide the nonexistent Athlon64s.
page 3
--------------------------------------------------------------------------------
The Mighty Mouse Scenario
An added bonus to all this is that AMD can drop this excuse the moment they don't need it anymore, and make themselves look even better than they would otherwise.
Let's assume the IBM/AMD scientists figure these SOI problems out quickly and AMD can come up with product before MS does.
AMD can then come up with the product they were supposed to make in the first place and say, "Oh my God! Aren't we amazing? Our brilliant geniuses were able to come up with a killer product despite those slugs at Redmond!!"
And if you have the memory retention of dynamic RAM with the power turned off, like most people, you'll probably believe it.
What If MS Hurries Up?
The worst thing that could happen to AMD is MS deciding to screw them by getting the product AMD wants from them out faster.
Reread that sentence. If that's the worst that can happen, how bad can that be?
If Windows x86-64 shows up early, in all likelihood, both server and desktop versions will become available. That increases the potential Opteron server market.
If Windows x86-64 shows up early, it will give enough of a performance boost to even current Hammers for AMD to proclaim if they'd like that they have a PIV-beater in 64-bit mode. If AMD manages to fix the real problem in the meantime, that's just gravy. Then they can say Athlon64 beat PIVs in one ad, and say Athlons64 destroy PIVs in 64-bit mode in another.
At absolute worst, people will realize this is all a scam. But it's unlikely MS will hurry up that much, and even if they do, AMD still ends up with a 64-bit OS to show for it.
Sounds Great To Me; What's The Downside?
page 4
--------------------------------------------------------------------------------
Sounds Great To Me; What's The Problem?
It's hard to admire scam when you're the one being scammed.
Specifically, the Prince of Darkness at Redmond doesn't have the memory retention of dynamic RAM with the power turned off. He and his cohorts will remember this one.
How would you feel if somebody manuevered you into getting blamed for something that wasn't your fault and made you look like crap?
Now imagine you are Bill Gates and the Microsoft heavies.
Would you like to be the person who did that?
More generally, AMD seems clueless as to why they can't sell to people and businesses with money, those who don't see only dollars.
Someone who is responsible for buying a thousand computers or servers has to look at more than the price tag. He's not buying computers, he's putting his career on the line because he's putting his company on the line. This isn't like buying a million paper clips.
Such people look at a supplier and say, "Can I rely on these people to come through for me?" It's not the product; it's the company.
This is something Intel knows down to the marrow of its bones. So much of the way Intel runs its business is simply meant to say and resay to its customers, "We're solid. We're reliable. We won't blind-side you, and if technology makes us change things, you'll get plenty of warning about it. You can plan with us." At the business level, Intel isn't selling product, it's selling a relationship.
And AMD does not comprehend that. In comparison to Intel, they run the company by the seat of their pants, and their operative motto seems to be "whatever gets us through the night."
I mean, really, when did Intel ever delay a processor coming out and blamed someone else for it?
Granted, none of these things matter much compared to the price tag when you're playing Quake, or the choice is an AMD server or no server at all.
But when you get to serious business with serious money at stake, it's death. That's why AMD has negligible big business market share, and antics like these just reinforce the image of a not-too-serious company selling to not-too-serious users.
Email Ed
News 
Market Data 
Discover 
)) 
AI
