With some “help” from Intel, AMD’s Hammer could have a rough time gaining acceptance. — Ed
We’re hearing more and more about Hammer as the days progress, and the stream will turn into a flood in the next six months.
One item that hasn’t been addressed, though, is the eventual additional costs Hammer will bring.
True, the great advantage of AMD’s Hammer over Intel’s Itaniums and McKinleys is that you can run all your current software with it, and you certainly can do that when and if you buy one of these systems.
However, to get full advantage of Hammer, you do need 64-bit OSs and applications. Estimates I’ve seen indicate a 64-bit application will do 10-20% better with Hammer than its 32-bit equivalent. That’s nothing to sneeze at.
Neither is the price to replace those 32-bit apps (provided you pay for these things). People making real money from their computers where time is money won’t think twice about it (unless they’re still Intelverts); but then there’s everybody else.
We’ve been at 32-bit so long that people have forgotten (or weren’t around to know) how awkward the transition can be when you widen the software data path.
If They Build It, Will They Come?
It’s going to be hard for AMD to persuade software-companies to compile x86-64 applications if that remains an AMD-only deal. Remember 3DNow.
It will be a lot easier if Intel’s me-too Yamhill ever sees the light of day, but if that happens, AMD loses the main advantage of going to x86-64.
The problem AMD may face is that they’ll have a really good CPU, but getting that good CPU means buying a new mobo and probably memory for it, and to get the absolute most out of it, you’ll also need to somehow procure 64-bit applications (that in any case won’t be too available for some time to come).
Add to that what looks to be a significantly lower GHz number than what Intel will have (and rest assured, Intel will point this out), and Hammer could be a really tough sell to Joe SixPack or Suit. For Hammer to succeed, AMD must persuade Joe to do so (or at least scare Intel enough to get them on board and give Joe no choice).
I can see Intel engaging in much marketing mischief to try to strangle this baby. If you think Intel advertising is unfair and deceitful now, just wait until Hammer comes out. They’ll try to make Hammer look like a Mac.
Intel is in a no-lose situation. If they succeed in trashing Hammer among the general population, they will at least seriously damage AMD. If they can’t, they just go with Yamhill and at least maintain the status quo.
AMD may find itself with its hands full just persuading the more persuadable to abandon Socket A, especially if they continue to develop processors for that platform. If they pretty much abandon Socket A after Hammer comes out (which they seem to want to do, sooner rather than later), they will aggravate many, and then they’ll sink or swim with Hammer.
In the past, when Intel has shifted, it took years to get people to transition over. Intel had 80-90% of the market back then. It could afford to wait and eventually get its way.
Can AMD be so patient with 20% of the market? Can they afford to be patient?
AMD is essentially betting that with a whiz-bang product, they can get the tail to wag the dog. My point is that the dog may not wag no matter how whiz-bang it is. The dogs may not wag because they’ll “know” 2GHz can’t possibly be better than 3GHz; they will “know” they’ll “need” 64-bit software for it; they will “know” that there isn’t much 64-bit software for it, and it will “know” that Hammer will cost them more.
Of course, everything that they’ll “know” will be wrong. So?
Hammer could become the next Alpha, a wonder to a few, ignored by the rest.