Inside a computer: CPU, memory and buses

Computer organization is about the parts a computer is built from and how they are joined. Computer architecture is the view a programmer sees: the instructions, registers and memory model. This course covers both.

The von Neumann design

Almost every computer since 1945 follows the design described by John von Neumann and others:

• a CPU (processor), which runs instructions;
• a memory, which holds both the program and its data, as numbers (the stored program idea: a program can be loaded like any other data);
• input and output devices;
• buses connecting them.

Inside the CPU:

Buses

A bus is a set of shared wires.

• The address bus says which memory cell or device to use. With k address lines, 2ᵏ addresses are possible: 32 lines reach 4 GB.
• The data bus carries the value read or written. Its width (8, 32, 64 bits) is how much moves at once.
• The control bus says what to do: read, write, wait, interrupt.

memory = [0] * 16
program = ["LOADI 5", "ADD 10", "STORE 11", "HALT"]
memory[0:4] = program          # the program is stored in memory...
memory[10] = 37                # ...and so is its data

for address, value in enumerate(memory[:12]):
    if value != 0:
        print("address", address, ":", value)

Because program and data share one memory and one bus, the CPU often waits for memory: the von Neumann bottleneck. Caches (later in this course) exist to hide it. Some designs, especially small microcontrollers, keep instructions and data in separate memories with separate buses: the Harvard architecture.

for lines in [8, 16, 20, 32]:
    addresses = 2 ** lines
    print(lines, "address lines ->", addresses, "addresses")
print("32 lines reach", 2 ** 32 // 2 ** 30, "GB of bytes")

Mark lesson complete