CPX8The CPU, Von Neumann and the Fetch-Decode-Execute Cycle

Almost every computer today follows the von Neumann stored program concept.

• Program instructions and the data they use are both stored, as binary, in the same main memory.
• The CPU fetches an instruction from memory, decodes it and executes it, one instruction after another (sequentially).
• Because a program is just data in memory, a computer can run a different program simply by loading it into memory. Nothing has to be rewired.
• Instructions and data travel between the CPU and main memory along the same set of buses.

One main memory holds both instructions and data; the CPU reaches it over three buses.

Stored program concept

Program instructions and data are stored together in the same main memory; the CPU fetches, decodes and executes the instructions in sequence.

• Main memory (RAM) holds the instructions and data of the programs that are running, including the operating system.
• Every location in memory has a unique address, so the CPU can say exactly which one to read or write.
• The CPU can read and write RAM directly and quickly. Programs are copied into RAM from storage when they are opened.
• Volatile storage loses its contents when the power is turned off. RAM is volatile.
• Non-volatile storage keeps its contents without power: secondary storage such as a solid-state or magnetic drive, and the chip holding the start-up (boot) instructions.
• Why computers need both: RAM is fast enough for the CPU to work from, but it forgets everything at power-off. Non-volatile storage keeps files and programs, holds the instructions needed to start up, and has a much larger capacity.

Main memory (RAM)

Memory that holds the instructions and data currently in use, which the CPU can read and write directly. It is volatile.

Volatile

Loses its contents when the power is switched off.

Non-volatile

Keeps its contents when the power is switched off.

Memory address

A unique number that identifies one location in memory.

The central processing unit (CPU) fetches, decodes and executes instructions. It has three kinds of component you need to know.

• Control unit (CU): decodes each instruction and sends control signals to the other components to coordinate what they do and when. It controls the flow of data inside the CPU and between the CPU and memory.
• Arithmetic logic unit (ALU): carries out arithmetic (adding, subtracting) and logical operations (comparisons such as “is a greater than b?”, and AND, OR, NOT).
• Registers: small memory cells within the CPU used for temporary storage of data and interim results. They are the quickest memory to reach because they are inside the CPU.
• You don’t need to name or describe any particular register.
• Diagrams use symbols: the ALU is a V-notched shape (two inputs, one output) and a register is a box split into cells.

Learn to recognise these two symbols in a diagram.

Control unit

The part of the CPU that decodes instructions and sends control signals to coordinate the other components.

Arithmetic logic unit (ALU)

The part of the CPU that carries out arithmetic and logical operations.

Register

A small memory cell within the CPU used for temporary storage of data and interim results.

• The clock synchronises the actions of the CPU. Each tick of the clock triggers an operation, so every component works in step.
• In diagrams the clock connects to the control unit, which uses its ticks to time the control signals.
• Clock speed is measured in hertz (Hz): ticks per second. 1 GHz = 1000 MHz = 1 000 000 000 Hz (these prefixes are 1000s, not 1024s).
• A higher clock speed means more ticks per second, so more instructions can be fetched, decoded and executed each second. Programs run faster, as long as nothing else holds the CPU up.

Clock

Synchronises the actions of the CPU; each tick triggers an operation.

Clock speed

The number of ticks per second, measured in hertz (Hz).

A bus is a set of parallel wires through which data or signals are transmitted from one component to another.

• Address bus: carries the address of the memory location to be read or written. It is unidirectional: from the CPU to memory only, because only the CPU chooses which location to use. Memory never sends an address back.
• Data bus: carries data and instructions. It is bidirectional: from memory to the CPU when reading, and from the CPU to memory when writing.
• Control bus: carries control signals such as “read”, “write” and clock ticks. It is bidirectional: the CPU sends commands and other components send signals back (for example, that a transfer is complete).

Bus

A set of parallel wires through which data or signals are transmitted from one component to another.

Address bus

Carries the address of a memory location from the CPU to memory. Unidirectional.

Data bus

Carries data and instructions between the CPU and memory. Bidirectional.

Control bus

Carries control signals, such as read and write, between components. Bidirectional.

• The width of a bus is the number of parallel wires it has. Each wire carries one bit at a time.
• An address bus with n wires can carry 2ⁿ different addresses, so the CPU can reach 2ⁿ memory locations.
• Adding one wire doubles the number of locations. A wider address bus means more memory can be addressed (used).
• A wider data bus moves more bits in each transfer.

The CPU repeats the fetch-decode-execute cycle for every instruction, billions of times a second. Fetch brings the next instruction from main memory into the CPU.

• 1. The address of the memory location holding the next instruction is placed on the address bus.
• 2. The control unit sends a read signal on the control bus.
• 3. Main memory places the instruction stored at that address on the data bus, and it travels to the CPU.
• 4. The instruction is held in a register, ready to be decoded. The CPU then moves on to the address of the following instruction.

Fetch, decode, execute, repeat.

• Decode: the control unit decodes the instruction to work out what it has to do and what data it needs.
• Execute: the control unit sends control signals to the components that carry it out.
• If the instruction needs data from memory, the data is fetched the same way as an instruction: address on the address bus, read signal on the control bus, data back on the data bus.
• The ALU carries out any arithmetic or logical operation, such as adding two numbers or comparing them.
• Interim results are held in registers. A result may be written back to main memory (see below).
• Then the cycle starts again: the next instruction is fetched.

Every transfer between the CPU and main memory uses all three buses.

• Reading (fetching an instruction or loading data): the address goes on the address bus; the control unit sends a read signal on the control bus; memory puts the contents on the data bus to the CPU.
• Writing (storing a result): the address goes on the address bus; the control unit sends a write signal on the control bus; the CPU puts the data on the data bus to memory, which stores it at that address.
• So the data bus carries in both directions, but the address bus always carries from the CPU to memory.
• Labelling diagrams: name each part exactly (control unit, ALU, register, clock, address bus, data bus, control bus, main memory). The clock connects to the control unit. Draw only the lines the question asks for.