This article from XBit Labs states that PR ratings for Barton will be different than for what we’ve seen for Palominos and TBreds.
Essentially, they add another 600 to the + number.
A current AMD CPU running at 1.67GHz is called a 2000+. A Barton running at the same speed will be called a 2600+.
As of right now, the only difference we know about between Barton and TBred is that Barton will have 512K L2 cache vs. 256Kb cache for earlier socket A Athlons.
Does that make a difference? Sure, it made roughly overall about a 7% difference when the PIV got it. Would that change alone justify some adjustment in PR? Yes. This much of an adjustment? No. Hell, no.
Are there likely to be some other changes to Barton that will improve performances. Some AMD exec mumblings could be interpreted that way, but it’s hard to see internal tweaks accounting for what would be another 15% or more jump in CPU performance. At least not any improvements that would leave Barton compatible with current socket A mobos.
No, I’m afraid AMD is not going to be as conservative in its PR ratings as they’ve been in the past. Some of it is already built into some dubious results from certain benchmarks.
For instance, SysMark 2002 scores will probably jump up a whole lot for Barton because they jumped a whole lot when Northwood with its increased cache was introduced. I suppose that kind of exaggeration is justifiable.
The rest will probably be AMD becoming PR liberals, and even that will likely to be mostly justifiable.
What we’ll probably see is AMD going from a conservatively-rated processor to a somewhat better processor whose PR is accurate to slightly-exaggerated.
That’s the likely reality, but the perception is just as likely to be the opposite. The knee-jerk reaction is going to be, “male bovine excreta,” and it’s more likely to stick than the initial reaction to PR.
I’ll tell you right now what the Russian Roulette comparison is going to be three-six months from now.
Somebody is going to take a TBred running at 2400+ or 2500+ (1.93-2.00GHz) and benchmark it against a Barton 2600+ (running at 1.67GHz), and God help AMD if the Barton loses.
Should that happen, when that future benchmarker pulls the trigger, AMD’s credibility will get blown out in the minds of many, maybe most. You’ll find more virtual expressions of “Male bovine excreta!” in the forums than real ones in a stockyard.
The underlying problem will not be the Barton being overrated, but the TBred being a bit underrated, but if AMD “revalues” the TBred, the cries of “male bovine excreta” will just come earlier.
This is all presuming I’m right. I think that likely, but hardly certain.
Let’s presume AMD is as honest as the day is long at the North Pole, and Barton’s ratings are just as or even more conservative than the current PR ratings.
If that were the case, why buy a TBred? Were all this true, Barton looks to be a much better processor. Unless you just like to tinker, why buy a TBred now for a measley improvement when you can buy something better not too far down the road?
Looks like damned if you do, damned if you don’t for AMD. Provided the damners count.
AMD probably thinks revaluation will increase, not decrease sales. Should that happen while the hardware sites are howling about bovine byproducts, this hardly would be a monument to the power of such places and people.