Chip8.js
A Chip-8 emulator written in JavaScript.
Chip-8 is a simple, interpreted, programming language which was first used on some do-it-yourself computer systems in the late 1970s and early 1980s.
Table of Contents
Installation
This guide assumes you already have Node.js and npm installed.
Prior to installing Chip8.js, you must have CMake installed.
brew install cmake
Clone the repository and install.
git clone git@github.com:taniarascia/chip8.git
cd chip8
npm i
Usage
Chip8.js can be run on the web, in a terminal, or using native keybindings.
Web
Development
Spin up a local server during development.
npm run watch:web
cd web && http-server
Deployment
Build and bundle the code for the web.
npm run build:web
Deploy to GitHub.
git add web && git commit -m "update web version"
git subtree push --prefix web origin gh-pages
Terminal
Run Chip8.js in the terminal by selecting a ROM.
npm run play:terminal roms/<ROM>
Native
Run Chip8.js natively with raylib (experimental).
npm run play:native roms/<ROM>
Motivation
Chip8.js is a project to write a Chip-8 emulator in JavaScript. The main motivation is to learn lower level programming concepts and to increase familiarity with the Node.js environment.
Here are some of the concepts I learned while writing this program:
- The base system: specifically base 2 (binary), base 10 (decimal), base 16 (hexadecimal), how they interact with each other and the concept of abstract numbers in programming
- Bits, nibbles, bytes, ASCII encoding, and big and little endian values
- Bitwise operators - AND (
&
), OR (|
), XOR (^
), left shift (<<
), right shift (>>
) and how to use them for masking, setting, and testing values - Using the Node built-in file system (fs)
- The concept of a raw data buffer and how to work with it, how to convert an 8-bit buffer to a 16-bit big endian array
- Writing and understanding a 8-bit and 16-bit hex dump
- How to disassemble and decode an opcode into instructions a CPU can use
- How a CPU can utilize memory, stack, program counters, stack pointers, memory addresses, and registers
- How a CPU implements fetch, decode, and execute
And here are some articles I wrote based on those concepts:
Testing
The unit tests for Chip8.js use the Jest testing framework. You can run all test suites with or without displaying coverage.
npm run test
npm run test --coverage
Chip8.js has two suites of unit tests:
- Opcode instruction masks and arguments
- CPU implementation of instructions
Instruction tests
The instruction tests cover the INSTRUCTION_SET
found in data/instructionSet.js
. Each instruction has:
- A
key
: for internal use - An
id
: for a unique name - A
name
: for the type of instruction) - A
mask
: to filter out arguments from instruction signifiers) - A
pattern
: to match the mask to the specific instruction pattern arguments
, each of which contain:
- A
mask
: to filter the nibble(s) to arguments - A
shift
: to shift it by location - A
type
: to signify the type of argument
{
key: 6,
id: 'SE_VX_NN',
name: 'SE',
mask: 0xf000,
pattern: 0x3000,
arguments: [{ mask: 0x0f00, shift: 8, type: 'R' }, { mask: 0x00ff, shift: 0, type: 'NN' }],
}
Each unit test checks an opcode to an instruction and tests:
- The unique
id
to ensure the correct instruction is running for the mask/pattern - The number of arguments
- The value of the arguments
test('6: Expect disassembler to match opcode 3xnn to instruction SE_VX_NN', () => {
expect(Disassembler.disassemble(0x3abb).instruction).toHaveProperty('id', 'SE_VX_NN')
expect(Disassembler.disassemble(0x3abb).args).toHaveLength(2)
expect(Disassembler.disassemble(0x3abb).args[0]).toBe(0xa)
expect(Disassembler.disassemble(0x3abb).args[1]).toBe(0xbb)
})
There are 35 instruction tests for 35 opcodes (the first instruction, CLS
, is no longer implemented).
CPU tests
The CPU decodes the opcode and returns the instruction object from data/instructionSet.js
. Each instruction performs a specific, unique action in the case
. The CPU tests test the state of the CPU after an executing an instruction.
In the below example, the instruction is skipping an instruction if Vx === nn
, otherwise it's going to the next instruction as usual.
case 'SE_VX_NN':
if (this.registers[args[0]] === args[1]) {
this._skipInstruction()
} else {
this._nextInstruction()
}
break
Each CPU test:
- Loads a
RomBuffer
containing the data of a single opcode - Sets up the state to make the instruction testable (if necessary)
- Executes the
step
method - Tests all possible outcomes of an instruction and state updates
In this example, the instruction can either be skipped or not skipped depending on the arguments, and both cases are tested.
test('6: SE_VX_NN (3xnn) - Program counter should increment by two bytes if register x is not equal to nn argument', () => {
cpu.load({ data: [0x3abb] })
cpu.step()
expect(cpu.PC).toBe(0x202)
})
test('6: SE_VX_NN (3xnn) - Program counter should increment by four bytes if register x is equal to nn argument', () => {
cpu.load({ data: [0x3abb] })
cpu.registers[0xa] = 0xbb
cpu.step()
expect(cpu.PC).toBe(0x204)
})
Acknowledgements
Author
License
This project is open source and available under the MIT License.