Intel's "Core" Architecture; Successor to the P4 Announced

As AMD have clearly been investing in developing AM2, which doesn't seem to have any great improvemnt over 939 other than implementing DDR2, I fail to see how they'll match Conroe tbh. That's a serious performance increase.. not only over netburst architecture, but over the latest dual core AMD chips.

Could this herald the day I move back to Intel after being with AMD for over 8 years now?

 

Wow, this is looking good, I think sharing the L2 cahce is quite the idea, would this mean if one cpu could use all the cache at once or is there like a limit for each core?

 

Haha

I was there when they first released 478 P4s and Im still there now

Combining the cache means you dont have to copy the data twice, but it gives problems with multiple accessing of data by two cores at once. Although, I read that's not a problem.

They *are* good and I want one!

I like they way they've already got a quad core Kentsfield FAB'd. NEEEED.

 

Dunno, if I was AMD I'd probably be keeping the secret weapons under lock and key untill theyre needed so as to not give intel any more ideas ?

Various rumours I've seen for AM2 were improved hyper-transit, quad-core, the confirmed DDR2 of course .. we'll see what else they have in mind when the time comes.

It might just be a case that we end up with intel thinking they have the lead back and going back to there old standby price point of twice whatever AMD is charging though

Edit add : all the thinking about what might happen made me completely forget what will happen : that being the FX-60 stats being demoted to just a run of the mill X2 processor (can't remember if they were calling it 5000 5200 or 5400 now) when it comes to AM2.

 

If one core is idle, the other core can use the entire 2 megs of cache, but when they're both active, they share it.
However, I'm not sure if there are quotas for each core when they're active or if they take what they need. For example, if one core were to need 70 percent of the cache and the other were to only need the remaining 30, would the cores be intelligent enough to be able to split the cache like that? Or would they both be limited to 1 meg each when they're both active?

 

Unless that memory issue is fixed, AMD is going to get its ass kicked as things stand, plain and simple. Problem is, Im not sure how things stand.

Theres no way the performance improvements should be that dramatic with no more difference than the operating platform and the processors. A Pentium 4 3.0 GHz and an FX-57 perform, visually, almost identically in the VAST majority of games when the same GPU is equipped, as most games are GPU, Not CPU, limited. Where did these massive performance differences, in GPU limited games, come from?

 

Thing is, using Crossfire x1900XTs it's gonna be hard to really GPU limit them unless you go for silly stuff. Remember these are still "Intel benchmarks" and even when Anand runs his FEAR benchmark it's only 1280 on Max without a mention of AA/AF.

To the other reply (above Roto's) Regardless of whether 1 CPU or more is using the cache it's there for everyone with only 1 set of data. Currently even with the 9xx range having 2x2meg of cache it's still effectively a 2meg cache processor as they can potentially (and are likely to) both carry the same data making them massively inefficient. If they share data, whoever wants it, whenever can get it saving die space and money and making it more efficient if two processors pile data into and out of the same cache.

 

Ok, its official: Conroe is indeed MONSTEROUSLY fast.



With that dramatic a lead, AMD is going to be hard pressed to compete with Intel; AM2 is shaping up to be a big dissapointment, much like Prescott at the moment. AMD engineers look to have their work cut out for them. In the mean time, God Damn those numbers are impressive

 

looking at that graph above, i cant believe that no one has noticed that the "overclocked" FX60 scores lower than the reference one!hmmm...

 

Yea, it's intel overclocked, whereas the reference one is in house here at bit-tech. We tweak all the systems for uber performance but I doubt intel really does the same with an AMD system.

 

Rahul also brought out the point that the BIOS in use doesnt properly support the FX-60, which could make a fair amount of difference.

 

i wouldnt trust indian guys with anything dealing with numbers....as they allways over estimate.
either way i want to see independant benchies to show how good the new cpu's are, and only then i shall make my judgment on weither my new rig will be AMD or Intel.

 

same im gonna definitely wait before i upgrade

 

We cant rule amd out just yet. am2 is just a new platform--940-pin sockets, ddr2 and im not sure if theres a new hypertransport. Theyre probly using this to get enough am2 mobos out there so that they can easily release a new marchitecture without having to worry about the platform. amd isn't like intel--new intel cpu new mobo, more cash

 

Nothing's out yet, and almost all Conroe - well, all Conroe to AMD comparisons right now are under a HUGE amount of bias.

 

Holy thread revival batman

We'll see how it performs when it comes out for real; its coming out in approximately 23 days - we dont have long to wait