Temptation

Add Your Comments

Take a look at Intel’s expected pricing come late October:

CPU SpeedMotherboard

Official Price

3.4GHz

$640

3.2GHz

$420

3.0GHz

$275

2.8GHz

$220

There’s nothing particularly unusual about that for Intel. What will be unusual will be the effect these prices are likely to have on the competition.

“Real” Athlon64s (as opposed to “Opteron64s” should show up in November.

It’s likely that AMD will initially try to price Athlon64s on par with PIVs; they tried doing that with Athlons earlier this year, and if anything, they’d have more reason to try again with A64s.

There’s a rather big price gap between the 3.0 and the 3.2/3.4.

This raises a dilemma for AMD.

PR is at least potentially a much more flexible concept than MHz. AMD could use all the money it could get. It is very tempting to “stretch” PR a bit to get to 3.2 and/or 3.4GHz and go for the gold. Some would argue that AMD did just that with the Athlon 3200+.

On the other hand, it is the debut of Hammer on the desktop. First impressions are important. If AMD’s PR claims prove to be a bit exaggerated, there are those out there who’ll make that the news story and call the initial Hammers DOA. That could lose AMD a lot of sales.

The matters gets further complicated when you consider 64-bit processing. From the 32-bit numbers we’ve seen so far, a single-channel memory Athlon64s would seem to fall somewhat short from the claimed PR performance, but even a conservative estimate of the boost x86-64 should provide would easily justify expected PR ratings. Of course, there won’t be too many x86-64 apps around at that time.

How will the reviewers (and public) look upon all this? Which will be considered the “real” measurement?

This will be confusing enough, but there will be one additional element to the noise.

There’s more than a few people out there who instinctively hate PR, hated it from day one, didn’t and don’t believe it no matter what evidence AMD or anybody else ever presented to prove it’s a reasonable estimation.

I’m not talking about people who have questioned some of AMD’s later PR ratings; I’m talking about people with brain damage on the subject. I’m serious when I say that (well, maybe not tissue damage, but cognitive failing). They simply can’t accept that due to the architectural changes made to the PIV, an estimate like PR is the only way AMD can equate the performance of its processors to Intel’s.

(This is not the same as saying that a particular AMD estimate is bad; such people find any estimate inherently bad. If they can’t directly verify it themselves, it doesn’t exist.)

You can just imagine how these folks will react when AMD tries to claim that not only that their 2 is more than Intel’s 2; but that it equals 3.4. Should it turn out that AMD’s 2 doesn’t quite equal 3.4, but, let’s say 3.2; they won’t say, “2 equals 3.2.” They’ll say, “See, I was right. It’s all BS.”

Do You Have A Better Idea?

Why, yes. 🙂 If you haven’t seen it already, today’s other article shows how AMD could have its cake and eat it, too.

Ed