1. Two things to carry

A controller needs instructions to know what to do.

It also needs data to work with.

Two fundamental requirements: Instructions (what to do) and Data (what to work with)

Every decision, state change, and control signal inside an embedded machine comes down to these two kinds of information.

2. What if they could travel separately?

In a system where instructions and data travel across the same shared pathway, they must take turns.

The processor cannot fetch an instruction and read or write data across that path at the exact same instant.

What if they didn't have to share the same path?

What if instruction memory and data memory were kept completely separate — each with its own direct highway to the processor?

3. Harvard Architecture

This organization is known as Harvard Architecture.

Instructions and data have separate memory spaces and separate paths to the processor.

Harvard Architecture: Separate memory spaces and independent communication paths

Instead of a single shared pathway, the processor communicates across two independent channels.

4. A different rhythm

Because instructions and data have separate paths, the processor can access them independently.

Instruction access and data access don't have to compete for the same path.

While the processor is reading or writing data from one memory, it can simultaneously fetch its next instruction from the other.

They move in parallel.

5. Built into the chip

Inside a real microcontroller, this separation takes physical form.

Flash stores the program — the instructions the controller needs to execute.

SRAM provides the working space for data while the program runs.

They serve different purposes, and the processor accesses them through their respective paths.

The controller has a place for its program.

It has a place for the data that program works with.

But that still leaves the most important part of a controller unexplored.

How does it interact with the world around it?