What do nanometers (nm) refer to when talking about microprocessors? Why is smaller better?


I often hear about a new, smaller microprocessor that is, for example 5nm instead of 7nm.

In: Engineering

Basically this is the size of the “switches” inside the processor. The more of these you have the more calculations you can do and the faster you can get things done.

Smaller switches means more processing in the same amount of space.

A microprocessor (or any other chip) consists of millions of electronic elements (such as transistors) that are extremely tiny. These elements are created by a special and very expensive machine on a surface called a wafer. Multiple chips are created on a single wafer as the chips are tiny and the wafers are about as large as a small pizza.
The nm is a measure for the accuracy at which the circuits can be created by the machine. A nm is about 150 000 times smaller than the width of a hair. Developing machines that can manage this accuracy is extremely hard and it is also common to have some amount of produced chips on a wafer that are bad. The percentage of good chips on a wafer is called the yield.
But the benefit of smaller nm sizes is that the created circuits are smaller, allowing more chips on one wafer which reduces cost. Also, having smaller circuits means that they can be more power efficient when they run and this can allow a chip to run at faster speeds.
When manufacturers start producing with a lower nm technology, the yield is often low. But for microprocessors many “bad” chips can be salvaged by disabling some functions or memory and selling them as lower specced processors.

When talking about nm when it comes to microprocessors, it can mean two things.

The most common one that likely everyone has heard is in reference to the fabrication process name. For example TSMC’s 7nm, in the context of the first 7nm CPU in the world.

That meaning is nothing, it is just a marketing term. 20 to 30 years ago, the process name had meaning behind it, related to the second meaning I’m talking about. It used to relate to the physical size or distance of something on the actual CPU.

Smaller is better because it allows less distance to move elections between components. This means it can be faster and more energy efficient. Smaller also means you can fit more in the same amount of space.

The size of the all the little tiny switches and operators that let computers compute stuff! Smaller is better because the smaller things are the more you can fit in that space AND you reduce the distance that particles called electrons (which make up electric current flow) need to travel to make it “faster”.

It’s not too meaningful of a number nowadays and more of a marketing thing, but it was important some time ago.

A huge amount of time is wasted when you need billions of calculations to travel aross a 10nm distance compared to a 7nm distance. Getting processors to be smaller means less time is wasted on travel, more processors can fit in the same area, and heat becomes more concentrated which (when cooled properly) will be an advantage over spreading that heat over other modules of the processor that run more efficiently when cool.

Originally it mean the size of the transistors that made up the processor. Somehwere round 45nm that stopped applying as it got harder to make them smaller. It’s just a name for a particular generation nowadays and has little to do withthe size of features.