4-Bit CPU
한국어
A fully functional 4-bit CPU designed from scratch using Logisim (visual circuit) and Verilog HDL (simulation).
Try it in your browser — write assembly, assemble, and step through execution with no install required.
This project was inspired by a circuit from Electronics Stack Exchange. The original circuit had issues when replicated, so it was fixed and improved. The instruction set and instruction decoder mapping were designed independently.
Architecture
The CPU follows a single-cycle design — each instruction completes in one clock cycle.
graph LR
INS["16-bit Instruction"] --> DEC["Instruction<br/>Decoder"]
DEC -- "Control Signals<br/>(SUB, AFS, RSS, DWE, WE)" --> MUX["4-to-1<br/>MUX"]
INS -- "Data (imm)" --> MUX
MUX -- "Write Data" --> REG["3-Port Register File<br/>(8 × 4-bit)"]
INS -- "Addr D, A, B" --> REG
REG -- "A" --> ALU["ALU<br/>XOR / AND / OR<br/>ADD / SUB"]
REG -- "B" --> ALU
ALU -- "ALU Result" --> MUX
REG -- "A (passthrough)" --> MUX
Components
| Component |
Description |
| Instruction Decoder |
Decodes 4-bit opcode into 7-bit control signals (SUB, AFS, RSS, DWE, WE) |
| 3-Port Register File |
8 general-purpose 4-bit registers (R0–R7), 2 read ports + 1 write port |
| ALU |
4 operations — XOR, AND, OR, ADD/SUB (two's complement subtraction) |
| 4-to-1 MUX |
Selects write-back source: immediate data, ALU result, or register value |
| 8-to-1 MUX |
Selects register output for read ports |
| 3-to-8 Decoder |
Converts register address to one-hot enable signal |
Instruction Set
16-bit instruction format:
[15:12] [11] [10:8] [7] [6:4] [3] [2:0]
opcode - Addr D - Addr A - Addr B / Data
| Opcode |
Mnemonic |
Operation |
Assembly Format |
Machine Code |
0000 |
NOP |
No operation |
NOP |
0000 0000 0000 0000 |
0001 |
LD |
Rd ← Data |
LD Rd, Data |
0001 0ddd 0000 Data |
0010 |
MOV |
Rd ← Ra |
MOV Rd, Ra |
0010 0ddd 0aaa 0000 |
0011 |
DISP |
Display (Ra, Rb) on ASCII display |
DISP Ra, Rb |
0011 0000 0aaa 0bbb |
0100 |
XOR |
Rd ← Ra XOR Rb |
XOR Rd, Ra, Rb |
0100 0ddd 0aaa 0bbb |
0101 |
AND |
Rd ← Ra AND Rb |
AND Rd, Ra, Rb |
0101 0ddd 0aaa 0bbb |
0110 |
OR |
Rd ← Ra OR Rb |
OR Rd, Ra, Rb |
0110 0ddd 0aaa 0bbb |
0111 |
ADD |
Rd ← Ra + Rb |
ADD Rd, Ra, Rb |
0111 0ddd 0aaa 0bbb |
1111 |
SUB |
Rd ← Ra - Rb |
SUB Rd, Ra, Rb |
1111 0ddd 0aaa 0bbb |
Instruction Decoder Control Signals
Each opcode maps to a 7-bit control word:
| Bit |
Signal |
Description |
| 6 |
SUB |
ALU subtraction flag |
| 5:4 |
AFS |
ALU function select (00=XOR, 01=AND, 10=OR, 11=ADD/SUB) |
| 3:2 |
RSS |
Register source MUX select |
| 1 |
DWE |
Display write enable |
| 0 |
WE |
Register write enable |
Project Structure
├── src/ # Verilog HDL source
│ ├── cpu.v # CPU top module (instruction decoder + datapath)
│ ├── alu.v # Arithmetic Logic Unit
│ ├── three_port_reg_file.v # 8 x 4-bit register file with 3 ports
│ ├── cpu_tb.v # Testbench with sample program
│ ├── mux_4to1.v # 4-to-1 multiplexer
│ ├── mux_8to1.v # 8-to-1 multiplexer
│ ├── decoder_3to8.v # 3-to-8 one-hot decoder
│ ├── fulladder.v # Full adder
│ └── Instruction.txt # Sample program listing
├── logisim/
│ └── 4-bit-cpu.circ # Logisim circuit file
├── docs/ # Web Playground (GitHub Pages)
│ ├── index.html
│ ├── style.css
│ └── cpu.js
├── images/
│ ├── Instruction-Set.png
│ ├── Instruction-Operation-Code(opcode).png
│ ├── logisim_main.png
│ └── verilog_cpu_tb.png
├── Makefile # Build automation (make run / make wave / make clean)
└── Dockerfile # Run simulation without local install
Logisim Circuit
The Logisim implementation includes a Program Counter (PC) and two PROMs — one storing the program (instruction codes) and one storing the instruction decoder mapping. This makes it a fully self-contained, clock-driven CPU.
Getting Started
Option 1: Web Playground (no install)
https://simdasong.github.io/4-bit-cpu/
Write assembly in the browser, assemble, and step through execution. Registers, program counter, and display output update in real time.
Option 2: Make (local)
Requires Icarus Verilog and optionally GTKWave.
make run # Compile and run simulation
make wave # Compile, run, and open waveform viewer
make clean # Remove build artifacts
Option 3: Docker (no local install)
docker build -t 4bit-cpu .
docker run --rm 4bit-cpu
Example Program
The testbench (cpu_tb.v) runs the following program:
LD $0, 0 # R0 = 0 (0x1000)
LD $1, 1 # R1 = 1 (0x1101)
LD $2, 3 # R2 = 3 (0x1203)
LD $3, 3 # R3 = 3 (0x1303)
DISP $2, $3 # Display R3 → 3 (0x3023)
MOV $4, $2 # R4 = R2 = 3 (0x2420)
SUB $4, $4, $1 # R4 = 3 - 1 = 2 (0xF441)
DISP $2, $4 # Display R4 → 2 (0x3024)
ADD $3, $3, $1 # R3 = 3 + 1 = 4 (0x7331)
DISP $2, $3 # Display R3 → 4 (0x3023)
XOR $5, $3, $4 # R5 = 4 ^ 2 = 6 (0x4534)
DISP $2, $5 # Display R5 → 6 (0x3025)
AND $5, $3, $4 # R5 = 4 & 2 = 0 (0x5534)
DISP $2, $5 # Display R5 → 0 (0x3025)
OR $5, $3, $4 # R5 = 4 | 2 = 6 (0x6534)
DISP $2, $5 # Display R5 → 6 (0x3025)
Expected display output: 3, 2, 4, 6, 0, 6
Simulation Waveform
References
License
MIT