Replies to post #1094 on AMD strictly moderated (AMD)

[pfosse]

Does anybody every remember a Server part clocking 2 bins lower than a desktop part? And these Xeons aren't even expensive at $850. I agree something is fishy. I think Intel is trying to FUD OEM's into not selling Opterons, but to wait for Woodcrests. It might work too.

[chipdesigner]

That is strange-- maybe they know they really need to stretch to hold server marketshare, but for desktop, they need to be able to manufacture at reasonable yields? But why are desktop parts on 1066FSB, server on 1333FSB? Is it only 'Conroe EE' that will have 1333FSB and a faster bin?

Merom tops out at 2.33GHz, in the 3rd article.

Not much for AMD to worry about here.

[wbmw]

Re: Woodcrest 3.0, 2.66, 2.33, 2.0, 1.86, 1.6.

Do these bins strike anyone as very odd? When do you ever see a product intro where the lowest bin is about 1/2 the frequency of the highest bin? I have two guesses here. Either 3.0 is a stretch or 1.6 is an ultra low voltage/low power part.

Nothing is odd here. It is like I've been saying all along. Woodcrest comes with a 1333MT/s FSB, which has a base frequency of 333MHz. With standard integer multipliers, they can only launch products that align with those frequencies:

3.0GHz = 333 * 9
2.66GHz = 333 * 8
2.33GHz = 333 * 7
2.0GHz = 333 * 6

I'm guessing for the other two, Intel dropped down to a 1067MT/s FSB, so:

1.86GHz = 266 * 7
1.6GHz = 266 * 6

Re: The Conroe article next to this one showed a top bin of 2.66.

This also makes sense, since Intel has said that Conroe will be 65W, while Woodcrest will be 80W. So the extra 15W power envelope enables a higher bin, much like the Opteron SE. You should also expect a Conroe Extreme Edition at 80W and 3.0GHz for the desktop, if this is true.

[CombJelly]

"I have two guesses here. Either 3.0 is a stretch or 1.6 is an ultra low voltage/low power part."

That is about the only choices. And if the 1.6 GHz is a full voltage part, then the 2.66 and 3.0 GHz parts are likely to be in short supply.