As CPUs become available with faster speeds, they also come with finer node architecture. Why is this? Could they not just make the CPU larger to accommodate more pathways? I understand that there are some power savings with finer architecture, but why would that stop a company from producing a larger and faster chip that was more power hungry, but intend that chip for wall-powered applications?
I just don't quite get it. Bandwidth, impedance… I'm not sure what other factors com into play…
I was a procurement and component engineer in IBM and Lenovo, but never had direct responsibility in microcomponents. I can give you some reasoning, but probably not the complete answer.
First, smaller grids in the architecture (the nm number) means two things.
It reduces the physical size for the required gate count to get the function. The smaller IC means more IC's per wafer directly reducing the cost to produce it. A wafer is a fixed size. Get more IC's on the wafer where the per wafer cost is the same means you reduce the cost as a divisor. For the same design IC, it is called a die shrink.
The second thing is that even as small as the wiring is, smaller distances mean lower resistance and a shorter timing delay between the gates in the gate array. All ICs are a set of transistors and memory.
The ICs used in laptops are designed to run at lower voltage for battery reasons, but another advantage of lower wattage is less heat, and the heat has to be removed from any system. They do run the smaller spacing at a lower voltage in general for reliability reasons. Note that the actual performance of an IC is more related to the design itself. Intel CPUs are designed better than AMDs. Intel gets more performance in a clock cycle. Coupled with the tighter spacing and lower voltages getting better performance, it is quite a combination of advanced technology at Intel.
Wider spaces are cheaper only to the extent of running on lower technology production lines to make them. As the line spacing is reduced, the lithography and processes and cleanliness needs to improve even further.
There's nothing that makes a physically larger IC faster. The power use is in its internal resistance and efficiency of the transistor gates. You are talking 1.4 billion transistors at 3rd gen:
http://www.legitreviews.com/intel-core-i7-3770k-3-5ghz-ivy-bridge-processor-review_1914
The design is about the transistors. The pathways need to be minimized. They need more transistors for higher functions.
Impedance is just another name for resistance.
The speed you refer to is a clock speed as in cycles per second and not the speed of electricity through the pathways. Smaller pathways are faster so tighter density is faster.
I hope there's enough here to explain what you ask.