Again not quite true. I don't know where you are getting all this from honestly.
Diploma in computer science, long time industry enthusiast, a job in IT, among other things. What about you?
We don't have high end CPUs with 20 cores. We just have high end CPUs with casual number of cores that have good performance per Core ratios. Even server CPUs have 4-6 cores, where performance of single core is equal to dozens if not hundreds of cores which you are are setting the limit for here. There is no limit of any kind. CPUs are getting more and more performance per single core and that's where those benchmark figures come from, certainly not from the number of cores. And the TDP of server CPU is still very much in Desktop PC range, that is 100-130W. So again, let's say you have CPU with given amount of computing power. You can divide this amount as you want. 2 cores; 4 cores or 22 cores if you wish. But total power at 100% load will never change. You will affect performance managing when multitasking, but not CPU potential. And there is nothing more to it in this regard. Can't make this more clear lol.
This is total nonsense. There is not such thing like 'computing power' in your explanation. You have execution cores which run at certain speed. The faster they run, the more computations they can perform within the same time interval. The TDP is a derived parameter which depends on the speed. The physical limit is how fast that speed can get. The record for max speed of a Haswell CPU AFAIK is 7.0Ghz, but you need some very special equipment to achieve that which makes it impossible in a normal system. Fastest Haswell CPUs shipped by Intel are configured for max clock of around 4.4 Ghz, if I am not mistaken (haven't had look at the Intel listings this year) and this only when the CPU operates in a single-core mode. This is your limit. Intel does not have a dual core has well with a TDP of 84W simply because they can't clock the cores high enough to reach that TDP in a stable fashion.
And now to your claim: 'CPUs are getting more and more performance per single core and that's where those benchmark figures come from, certainly not from the number of cores'. First, if you look at the benchmarks it is kind of obvious that the multi-threaded performance scales almost linearly with the number of cores. Second, the single threaded performance of the 13" and 15" is virtually identical. Why? Because the benchmark is run at the maximal dynamic CPU clock, which is 2.9Ghz for the base 13" and 3.2Ghz for the base 15". Still, the 15" is almost twice as fast in the multi-threaded benchmark, precisely
because it has more cores.
Now as for the turbo boost, what turbo boost essentially does, is substantially boosts frequency of one or more cores if given activity/process can't take advantage of all cores. Let's say we have 100W TDP CPU with 4 cores. It can temporarily give 2 cores 50W each and according to stepping, temporarily increase overall TDP to 110W or so and overclock the whole CPU. Now similiar thing can happen with dual core CPU. It can increase the frequency and give more power to single core if that's needed therefore possibly increasing overall TDP just as well
Sure. And that is why a quad core is faster. If your software does not take advantage of multiple cores, both the dual core and the quad core can boost their clock to a maximal safe limit (which is the same for both CPUs). In these scenarios the performance of a dual-core and a quad-core will be the same.
If your software does take advantage of multiple cores, the quad-core will fly, because it can more efficiently distribute the workload. For cores @2Ghz will beat two cores @3Ghz if you can 100% load them with work. The current multi-core designs are flexible and take the best of both worlds. This is why they are so successful. Again, times have changed since Q6600. Current CPUs work differently.
so like I said the difference between number of cores/threads comes down to multitasking, but it applies more when we compare 2cores or threads with 4, and not that much when comparing 4 to 8. The efficiency should be all what matters since these machines are TDP limited and you certainly don't NEED more than 4 threads, especially when there is so little performance to begin with when compared them to certain desktop CPUs with 4 threads which eat everything you throw at them, which just shows one more time that 4 threads is not a limitation for overall performance of any kind.
And that is what people in this thread are talking about. Multitasking. If you have a task which is easily split in parallel workloads, you want as many cores as possible. My statistical simulations run 2x faster on my quad core than on my older dual core. They run 2.5x faster yet on my 12 core Mac Pro Server (even though my laptop is faster in single-core performance).
I guess the message you want to convey is 'dual core is enough for most users'. I would certainly agree. Most of the activities an average user performs (browsing, emails, video etc.) are not parallel. However, a quad core becomes a huge bonus if you start doing parallel workflows (e.g. encoding/editing video), where a quad core is almost twice as fast as a corresponding dual core. For me, quad core is a necessity it makes a very big difference whether I need to wait 15 seconds or 30 seconds for results of a computation.
P.S. Just a note for you. 'Core' and 'thread' are very different concepts and have nothing to do with each other. Single-core CPUs were running hundreds of threads at the same time long before the industry switched to multi-core designs.