I think the main thing Intel did here was make the chip more efficient. It is a slight performance boost over Clovertown but not by much - perhaps about 10% as was expected, but with that efficiency kicker.
2.0Ghz Barcelona actaully holds in own in some benches against this higher clocked 3.0Ghz Xeon. Of course, it seems the Barcelona's are still more energy efficient - but the margin has been reduced. If AMD moves to 45nm in 9-12 months as they see AMD maintain that crown - UNTIL Intel release their HT answer and does away with FSB - then in my estimation the crown would go to Intel since their CPU's are more efficient than AMD's - but their platform isn't.
Well - is Harpertown the nail in the coffin for AMD - NO! It certainly isn't that. AMD was given a brief moment to breath - very brief - and in that momen they need to release the 2.5 and 3.0Ghz Barcelona.
I suspect they also wanted to wait for following reasons:
1. See what Harpertown can do.
2. Release the CPU parts with best yeilds, before going into the higher problem 2.3Ghz + range (they need money you know!!)
3. Wait for large inventory of 2.5Ghz + before official release.
4. Problems with performance - I suspect the 2.5Ghz+ parts consume way too much power for what AMD wants.
I wish they had a better comparison, but I guess it'll do. They should have compared Cloverton at 3.0Ghz to Harpertown at 3.0Ghz. Instead they used Clovertown at 2.66Ghz and then tell us it is a marked improvement!??!?! Of course it's going to be faster - why not apples to apples.
Of course the 2.0Ghz Barcelona was not expected to beat the new 3.0Ghz Harpertown - but the fact that it wins on 25-35% of the benchmarks against a higher clocked CPU should make you wonder what 3.0Ghz Barcelona (if they can make it) will do.
So it would seem that on the whole Harpertown is at most 15% better performer than Clovertown at same clock speeds. If AMD wants that crown they had better role out a 3.0Ghz Barcy . . .