Okay, so for the first question.
Imagine you had a warehouse, and in that you were putting machines that do one specific thing, let’s say, make a pencil. If your warehouse is 1000 square feet, and your pencil machine takes up 10 square feet, it would follow that you can naturally fit 100 pencil machines. Now say you reduce the size of the pencil machine to 5 square feet, half the original size. It would naturally follow that you can fit twice as many machines in the same space.
For computer parts essentially scale that down to extremely small areas. Part manufacturers are trying to put as many “Machines” (transistors) in a small area, and reducing the scale of each individual “machine” means you can fit more
For the second question.
A CPU and a GPU are essentially machines that are specialized to do one specific thing.
Back to the pencil machine analogy, if you have a pencil machine, you may be able to make a pen using it, but it’s not what it’s designed to do, so it will be less efficient. Maybe it takes three times as long to make a pen with a pencil machine. So what do you do? You get a warehouse with pen machines in addition to your pencil machines. Now you can make pens much faster, and your pencil machines are able to do work that they’re more specialized for.
In terms of CPU VS GPU. Essentially CPUs are meant to calculate one “large” operation, say 123456789 x 987654321, whereas a GPU is meant to calculate a lot of smaller operations, for example it might calculate 2×2 5000 different times.
This is a massive simplification of hardware architecture overall, but in principle that’s how they’re meant to work.
Latest Answers