Conroe/Penryn:
Pro's: Massive IPC increase and still highly clockable compared to competition
Cons: You have to choose whether you want dual core or quad core performance (both from price, power consumption, and absolute performance in games)
Cons: Need massive amount of cache ($$) as a band-aid to off die memory controller
Nehalem/Lynnfield:
Pro's: Another big IPC jump, similary clocks to Core 2 but more overclocking headroom, moving memory controller on board was a good idea
Pro's: Finally an affordable quad core, and turbo mode that gives it single threaded performance of a dual
.....
Moving BCLK generator to CPU caused drama, but was likely because of the massive overclocking headroom Intel knew about. Unlocked CPUs make it so easy to overclock everyone on tech forums think they are 1337
........
I wonder things like: how did Intel get clocks to scale so high? Was that their intent or a side effect?
Why did Intel decide to make fatter cores rather than more cores? Was that solely because of hyper threading? Or better margins?
Just a point regarding the Core2's and gaming. Absolute performance for quads in gaming was not really what the market was for at the time. Multi threading games at the release of the q6600, for example, was pretty poor in terms of performance gain it gave. Going dual over quad was the recommendation if gameing was all a person cared about.
As to power on those quads, it as a bit of a joke as IIRC the q6600 was more of two e6600 in one package than being designed as a quad from the start. I coule be wrong, but at that point, the development for quads were not as far advanced as the step from Pentium 4's was to the Pentium 4 D series.
As to large cache on the Cores, that was the "in" thing at the time. The later revision of the Core 2's had a larger than expected cache only because at the time AMD had moved it's memory controller onto the CPU and was getting a lot better performance.
At the time, if you had memory hungry applications, AMD was a better choice for some for this reason alone. Part of the reason I think the server range from AMD at the time was a better choice.
moving on.
Nehalem/Lynnfield memory controller onboard was more of a catch up than a "good idea". It was expected, not anything else as I recall.
Affordable quads depends on how you count it. The Q6600 was considered the first affordable quad at under $300. From intel anyway.
As to turbo, I think that was more of an idea in the same way hyperthreading was an idea to address a problem. While hyperthreading addressed the long pipelines intel was using at the time, turbo addressed the thermal headroom issue when running light tasks.
Now while I like turbo (gives access to something otherwise lost), hyperthreading can take a long walk off a short pier. It is/was a bandaid to a mistake in it's design approach of the time (long pipelines to allow for massivly high clock speeds). But it is still around for one reason or another (ie: marketing purposes mostly in my book). Sure some people find it useful, but without the need of it as a bandaid, those people would not be a enough group for intel to develop the tech normally.
Core 2 did away with most of the needs of it, (just like large cache sides did when the memory controller was moved on die), but that IIRC was resigned from the ground up based on the P3 and powering down unused cache (a feature from the mobile development arm of Intel). The P4 was just a dead end development and was cut loose.
As to the on chip BLK, that was ment to be a design issue to allow for tigher timings inside the cpu, which helped with getting better performance. The unlocked cpus were not because of the headroom on offer, but Intel knowing that some of their customers want to push the hardware. This then gave intel a excuse to charge more for re-enabling a feature that otherwise they were turning off.
The issue at the time was that FSB overclocking was the only option as Intel had long ago removed clock adjustments on the grounds of people re-badging slow cpus as faster ones. As mentioned at the time, taking from one hand to give with the other is not generally something worth being thankful for.
As to the overclocking head room, that is most likly not a side effect of the design. It is needed if you are going to have a high turbo mode (worth mentioning) and still have a stable cpu.
As to faster cores vs more of them, that is generally well known that Intel take the approach that most software is still single threaded. As such, it benifits more from faster cores than more of them. AMD has taken the other path as going forward it is cores over speed (well, we are expecting that anyway) and that more cores is what businesses want (more money in it on a per chip basis).
in short, it is in intel's favour for software code to not develope beyond single threaded applications and for software bloat to occur.
Intel is no shining white knight, they just have a good collection of spin doctors and similar to make the less liked features/habbits be forgotten.