Woodcrest performance lead

[ahock]

Hi Guys;

Just wanna ask honest opinion here and no flame war please. I just wanna ask your opinion whether woodcrest can last its leadership longer. For conroe we know that it can trounce FX series big time running only at 1066 FSB. If AMD decides to push some improvements or speed bump through 65nm process, Conroe can still be pushed through 1333 FSB which I think can still hold on leadership in desktop plus speed bump.

The way I see it in woodcrest, it is now on 1333FSB running at 3GHz. Yes it can push through above 3Ghz but my guess is that the reason why Intel want to have Clovertown as early as possible is to have time to market leadership which I can translate that woodcrest/bensley is already stretched. Do you know whether Blackford can go beyond 1333FSB? Well if yes I think it can still hold on much longer but Intel is somewhat seeing quad core as its saviour. Makes sense?

 

[hardwareking]

in my opinion intel should try something like hyper-transport.Cuase even great processors will be held back by a saturated bus.
But i'm pretty sure blackford will go above 1333.And woodcrest will stay on top for quite some time.(K8L will be the competition)Right now it's decimating opterons.And don't forget 45nm core 2 duo's,xeon etc.. are yet to come.And those will bump the speed up even more.

 

[classy]

Well servers are a different animal. Socket F ddr2 optys will not only compete against Woodcrest well, but will win its share of benchmarks because of bandwidth, which is something the ddr2 optys will have plenty of.

 

[Maximilian]

Core 2 will be fine with a 1333mhz bus until this comes out. Intels answer to AMD's hypertransport.

 

[Furen]

Woodcrest will be ahead until AMD introduces its 65nm process, after that things get trickier. For example, AMD will likely hit 3.0GHz which will bring performance of 2-socket parts very close to Woodcrest's and will lower power draw dramatically (or it should, anyway). On the lower speed grades this powerdraw will be even lower which will probably give AMD a massive "performance per watt" advantage at around 2.4-2.6GHz. A 2-CPU socket F Opteron system looks like it will match a 2-CPU Woodcrest system in power draw (thanks to Intel's reliance on FBDIMMs) so slashing power draw by anything at all will be an advantage here.

Then there's K8L, which should increase floating point performance quite a bit (it should also bring a slight Int performance increase due to front-end optimizations but the degree of improvement is still debatable).

EDIT: Edited for Clarity.

 

[Griswold]

What Furen said. Or at least that is AMDs plan according to this info. Quad cores with the same platform power draw as dual cores will give woodcrest and clovertown something to think about in the performance/watt department.

 

[ahock]

Yes I agree, K8L might (as we dont know yet its capability) be ahead of clovertown and woodcrest in performance and performance per watt BUT do note that clovertown have the time to market advantage of say trending to 6 months which by the time AMD will release its 65nm part clovertown will be out. Who knows what happens next after clovertown or how fast intel can release its CSI thing.