- Joined
- Feb 8, 2004
- Location
- Chicago - USA
Well, excessive thermal dissipation & the inability to scale MAY be releated (as in early/current Pressys), but 'technically' are two seperate issues. But agreed, we still do not know what the actual stepping improvement will be.
Willamette had many early development probs, in terms of performance and OC'ing, but eventually evolved into the P4C northwood, which is often a great overclocker (though it took a core revision). Also, both process engineering, & product revision engineering, are very much 'arts' of refinement.
I hope some E0's hurry up into retail, or at least to the reviewers, 'cuz I've really been holding off on building another cruncher, due to indecision. I don't want to move to H20 for Distributed Computing, unless required, 'cuz I don't find it cost effective (I build caseless & headless systems running thru a KVM).
IF they do hit 4.3+ on air (big IF), and seeing the current relatively low demand for socket T cpus due to the reluctance to move to LGA775/9.xx mobos, we might be able to buy an E0 16X 'J', drop it in a hybrid mobo (DFI 875P-T/Asus P5GP800), skip the multi change feature, and run 'em straight up to ~270 FSB 1:1 w/some good TCCD!
That's how it was w/the early P4C's before they started binning them more accurately. Most anything was getting into the 3.2 - 3.3 range on air (often enough w/stock HSFs!) pretty easily, whether it was a 2.4 or a 3.0, at least at the very beginning of the P4C releases.
But I must admit: Intel publically dropping thier 4.0 release does seem troubling. Weird, considering they seem to be moving forward quickly w/the 6xx series of Pressys in Q1 05, which will have 2MB of cache, so I'd expect they might have a little better handle on the power/heat issue, since doubling the cache will add a fair amount of additional transistors requiring additional power.
Perhaps this has something to do w/marketing? If not, then if done for technical reasons, this is starting to sound more like it MIGHT be a scaling issue, rather than a power dissipation issue, at least for the current Prescott architecture, and E0, or any possible further stepping revs.
JMO's.
Yshhh....sorry for another rant...
Strat
Willamette had many early development probs, in terms of performance and OC'ing, but eventually evolved into the P4C northwood, which is often a great overclocker (though it took a core revision). Also, both process engineering, & product revision engineering, are very much 'arts' of refinement.
I hope some E0's hurry up into retail, or at least to the reviewers, 'cuz I've really been holding off on building another cruncher, due to indecision. I don't want to move to H20 for Distributed Computing, unless required, 'cuz I don't find it cost effective (I build caseless & headless systems running thru a KVM).
IF they do hit 4.3+ on air (big IF), and seeing the current relatively low demand for socket T cpus due to the reluctance to move to LGA775/9.xx mobos, we might be able to buy an E0 16X 'J', drop it in a hybrid mobo (DFI 875P-T/Asus P5GP800), skip the multi change feature, and run 'em straight up to ~270 FSB 1:1 w/some good TCCD!
That's how it was w/the early P4C's before they started binning them more accurately. Most anything was getting into the 3.2 - 3.3 range on air (often enough w/stock HSFs!) pretty easily, whether it was a 2.4 or a 3.0, at least at the very beginning of the P4C releases.
But I must admit: Intel publically dropping thier 4.0 release does seem troubling. Weird, considering they seem to be moving forward quickly w/the 6xx series of Pressys in Q1 05, which will have 2MB of cache, so I'd expect they might have a little better handle on the power/heat issue, since doubling the cache will add a fair amount of additional transistors requiring additional power.
Perhaps this has something to do w/marketing? If not, then if done for technical reasons, this is starting to sound more like it MIGHT be a scaling issue, rather than a power dissipation issue, at least for the current Prescott architecture, and E0, or any possible further stepping revs.
JMO's.
Yshhh....sorry for another rant...
Strat