Well, maybe K8L would be comparable to Conroe IF IT EXISTED.
You all deemed it fair to put a 3 year old architecture against a 5 year old, but you know, AMD is just better and we're AMD fanboys so we can do whatever we want as long as AMD wins. Morons.
Compare what's available to what's available. Core 2 Duo will launch in 9 days, so I consider it currently on the market. Intel wins. K8L sounds good, but it hasn't been taped or even shown in public, which makes me suspect that it's nothing special, not to mention the loads of L3 cache and serious lack of L2. L3? You might as well go to memory. L2 is what will make a processor fast, and thanks to the IMC on AMD's chips, you won't be touching anything above 512K.
K8L is like what Prescott is to Northwood. It's just a revision of an old architecture. Even if it really
is better than Core 2 Duo, they still can't touch Intel's roadmap. Quadcores before the end of 2006, desktop in Q1 2007, and another brand new architecture in 2008. AMD better get off their asses and come up with something new. You can compare K8L with Core 2 Duo, but there will already be another CPU on the market by then, so you're comparing it with something that's obsolete by the time K8L actually comes around.
K8L = Waste of time, money, IMO.
You have to compare Core and K8 as they are the current offerings by both Intel and AMD. When K8L is released we will do the same exercise all over again to see how they compare.
I suppose the only angle not really being covered is that we are comparing a very immature architecture against a very mature architecture. I may be tempted to build a Core system after the CPU is out about 12-18 months, hopefully all the bugs/glitches have been ironed out with them (will also sit on the fence on K8L until it has matured a bit).
Core looks like it will take the performance crown in the 1-4 core market segaments for the forseeable future. Opteron still owns 8+ core market segaments.
What I do like is that Intel and AMD have competly different approaches to how they improve their architectures at the moment. Intel is strongly focused on pure CPU performance, while AMD is looking outside the CPU with HT improvements and co-processor type cards that fit in the CPU slots.
Has AMD made the right choice again "thinking outside the box" for improvements, or should they have focused more on getting K8L out faster? It's a bit risky, but this kind of thinking has paid off for them in the past with things like 64bit, HT, and Integrated Memory Controller.
While I'm not particurlarly interested in the faster CPU, I am interested in the better platform. I still think AMD has the lead in having a more complete platform, but Intel certainly has the performance edge.
Agree on the 'compare what we have at the moment' but the approach on the CPU core now is almost identical for both: Conroe is an improved "copy" of K8 architecture, even slower but more efficient (14 stages vs 12 of K8 while netburst topped @ 31 giving fast but highly inefficient processors), finally, they even adopted (their own way) the hated numbering system of AMD.
Problem being Core 2 Duo is nowhere close to what K8 is. It's closer to a Pentium 3, and if you think about it, that means that K8 is a 'copy' of the Pentium 3 architecture. OMG AMD COMPYING INTEL AGAIN. Well you can sit here and think that somebody is copying somebody else. Nobody is copying anybody.