Replies to post #67317 on Advanced Micro Devices Inc (AMD)

[alan81]

Absolutely nothing!
Has Intel said anything about expected speeds out of the new cores?
Intel has said that double the power is about a 30% speed gain though. If we use this rule, and have Yonah at 31W, 2.32Ghz and assume Conroe is around double in power, we get almost exactly 3Ghz. There was a rumor that Conroe has a longer pipeline than Yonah but this is an unknown at this point. It may be longer, but used for adding 64b and other features rather than higher clock speeds... I don't think anybody outside of INTC knows right now.
--Alan

[wbmw]

Re: Well I'll be impressed if Merom/Conroe/Woodcrest can get over 3Ghz given Yonah is not anywhere close to that now.

Don't go by Yonah, since it is thermally contrained by a 31W power envelope. Conroe is slightly constrained at 65W, but Woodcrest opens it up to a larger 80W.

It will also depend on the FSB versions that are available at launch. Keeping in mind that frequencies are based off multiples of the FSB base clock; therefore, we have the following possibilities:

1067MT/s FSB:
2.13GHz(x8)->2.4GHz(x9)->2.67GHz(x10)->2.93GHz(x11)->3.2GHz(x12)

1333MT/s FSB:
2.0GHz(x6)->2.33GHz(x7)->2.67GHz(x8)->3.0GHz(x9)->3.33GHz(x11)

So if the top speed bin for these chips is ~2.8GHz at launch, the best they can do is 2.67GHz. If it's ~3.0GHz, they can do 2.93GHz for 1067MT/s or 3.0GHz for 1333MT/s. If it's ~3.2GHz, the can do 3.2GHz for 1067MT/s, but still only 3.0GHz at 1333MT/s.

Personally, given that Woodcrest has a larger thermal envelope, I would predict that Intel intends to keep one bin ahead of Conroe, and probably implement the faster front side bus, too. That's why I think we'll have 2.67GHz or 2.93GHz on the desktop and 3.0GHz in servers. After a quarter or two as Intel ramps in frequency, we'll probably see Conroe up to 3.2GHz and Woodcrest up to 3.33GHz. This excludes Extreme Edition parts, which will probably also use 1333MT/s like the highest Xeon bin.