What’s the most hands-on way to learn how CPUs work?

If you want to see the physical hardware side, the 8-bit breadboard computer series by Ben Eater is excellent. It shows how registers, buses, memory, clocks, and control signals work together. For programming practice, pair that with a small assembly language or emulator project—building hardware explains the CPU, while writing programs teaches you how to use it.

Turing Complete is a good option if you want something game-like. Its progression has you build gates, memory cells, adders, decoders, and registers before using your own CPU to solve programming challenges with assembly. It provides a lot more immediate feedback than a textbook.

Another practical route is to write an emulator. Start with something small, such as CHIP-8 or the 6502 used in the NES. Implement memory, registers, instruction decoding, and the fetch-execute loop, then run test programs. The 6502 has a manageable instruction set and plenty of documentation, and an emulator forces you to understand what every instruction actually does.

I’d avoid jumping straight to modern x86. The instruction count is only part of the problem; x86 also has decades of legacy features, multiple addressing modes, privilege levels, caches, and other complexity. A small 8-bit CPU or emulator will make the fetch-decode-execute cycle much easier to understand. You only need a small subset of instructions for useful beginner programs anyway.

Try Nand2Tetris or a similar build-it-yourself project. You start with logic gates and gradually create adders, memory, registers, a CPU, an assembler, and eventually a working computer. Each step has an implementation task, so you’re not just reading about the theory—you immediately use it.