Von Neumann Architecture

Von Neumann (Princeton Architecture)

Design

Design:

John von Neumann (1945)
Stored-program computer concept
- Instruction and data stored in the same memory

Components:

Processing Unit
- Arithmetic Logic Unit (ALU) and Registers
Control Unit
- Instruction Register (IR) and Program Counter (PC)
Memory Unit
- Data and Instructions
External Mass Storage
Input and Output Mechanisms

Registers:

PC: Program Counter
- Address of next instruction to be executed
IR: Instruction Register
- Current instruction during processing
AC: Accumulator
- Stores intermediate arithmetic and logic results
MAR: Memory Address Register
- Stores address of memory to read into the MDR.
  - (To send data to the MDR you need to send a read signal)
MDR: Memory Data Register
- Where read data is stored.
GP: General Purpose Registers

Note: MAR and MDR can be referred to as the BIU (Bus Interface Unit)

System on a Chip (SOC): When the CPU and memory unit are on the same silicon.

Memory: - Address and content

Large array of locations where binary information is stored and fetched

Buses: Means by which data is transmitted from one part to another

Note: Registers used in the fetch-decode-execute cycle.
Fetch
PC
IR
Decode
IR
Execute
Every single register gets used!
REGS
IR
PC
MDR
MAR
(If there is a jump command)
# 1. Fetch Cycle
MAR <- PC # Move program counter into Memory Address Register
MR # Signal read through address bus, then data bus sends data to MDR
IR <- MDR # (We need to move instruction to IR for CU to work on it)

Example: C = A + B

Suppose we have three locations in memory: A, B, and C.

Q: How can we do C = A + B?

A: Place A in the MAR, then send the read signal through the address bus.

The data will be placed on the data bus, where the data will show up on the MDR.

In individual steps:

MAR <- @A
MR # memory read signal
MDR <- A
AC <- MDR
MAR <- @B
MR
MDR <- B
AC <- MDR
MAR <- C
MDR <- AC
MW

In assembly:

LD A // The first four lines of the above
ADD B
ST C // Store contents of AC in C

Example: A = A + B

LD A # Put A in the AC
ADD B # Add B to the contents of the AC
ST A # Store contents of AC in A

Example: Micro-stepping through AC <- AC + M1

# Fetch
MAR <- PC
MR
MDR < MMEM.MDR
IR <- MDR
# Decode
MAR <- @MI
MR
MDR <- MMEM.MDR
# Execute
ALU.lhs <- AC
ALU.rhs <- MDR
ALU.add
AC <- ALU.out

Harvard Architecture

Biggest Difference: Instruction and data memory are separate (two memory units).

Used for digital signal processing where the bandwidth and speed are needed.