With fixed chip size, core crowding heats things up faster-> throttling.
So here´s the processor specs:
4-core 3.7GHz
6-core 3.5GHz
8-core 3.0GHz
12-core 2.7GHz
Where´s the double processors? Why only single processor? Where´s the 2 x 12-core (24-core)? Why 12-core has so slow clock speed? Would getting a 6-core 3.5GHz be the fastest machine?
Even with the previous Mac Pro people (people with lot of knowledge in these things) were saying that the clock speed is what matters, not the cores. And that the 6-core was the fastest machine.
I´m confused. I don´t know what to buy. I really want the 12-core machine, but the 2.7GHz clock speed is really pathetic speed.
Clock speed matters and the number of cores matter, but what matters most is dictated by what software you're using. If you were one new to computer systems and hadn't decided on what software you'd be using, then, I'd suggest that you look at the speed question from the perspective of potential and this is how I'd recommend that you look at it:
1) the 4 core system: 4 * 3.7 GHz = 14.8 GHz
2) the 6 core system: 6 * 3.5 GHz = 21 GHz
3) the 8 core system: 8 * 3 GHz = 24 GHz
4) the 12 core system: 12 * 2.7 GHz = 32.4 GHz
You can do the same analysis taking into account the various Turbo Boost stages and amounts.
The reason why core speeds are made to decrease as core count increases is thermally related. By way of example, if you were in a hotel room on a cold winter night with no heating appliance, you might want that room packed with as many heat emitting, high metabolic human units as possible. But over heating a CPU causes it to throttle (down clock), at best, and to fail if it can't throttle down while the same pressures persist.
On a related note, my theory why we haven't been shown by Apple the 12-core pricing is that Apple is still dealing with the throttling that was evident to me when Apple posted on Geekbench [
https://www.macrumors.com/2013/06/19/apples-new-mac-pro-begins-showing-up-in-benchmarks/ ] the performance of the E5-2697 V2 model. That Geekbench 2 score should have been much higher than 23,901. That's why I question whether Apple will be sticking with the E5-2697 V2 or should adopt the E5-2695 V2 because of its lower TDP. I've found that a GPU with lesser potential can outperform a higher potential one in Geekbench and other metrics because the GPU that doesn't find the need to throttle runs at full potential while the one with higher potential just overheats and slows down so much that you never see what it could do if it was tested in the proper environment. Being housed in a cylinder with just one fan for cooling everything w/the cylinder sitting on a desk just in front of the user who's running Logic loaded with plugins, an overheated CPU may soon yell, "I can't handle it - Its getting to hot. Fans, you better run at full tilt and let the goal of remaining quiet be damned." Apple takes great pride in the silent operation of its systems.