Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/omarazizi/kaleidoscope

Toy Programming language & optimizing JIT Compiler built with LLVM compiler infrastructure and C++
https://github.com/omarazizi/kaleidoscope

compilers cplusplus llvm llvm-ir programming-language

Last synced: about 20 hours ago
JSON representation

Toy Programming language & optimizing JIT Compiler built with LLVM compiler infrastructure and C++

Host: GitHub
URL: https://github.com/omarazizi/kaleidoscope
Owner: OmarAzizi
Created: 2024-02-09T11:58:27.000Z (8 months ago)
Default Branch: main
Last Pushed: 2024-02-19T18:10:23.000Z (7 months ago)
Last Synced: 2024-09-21T05:02:52.207Z (4 days ago)
Topics: compilers, cplusplus, llvm, llvm-ir, programming-language
Language: C++
Homepage:
Size: 85.9 KB
Stars: 3
Watchers: 1
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

        # Kaleidoscope :telescope:

![C++](https://img.shields.io/badge/c++-%2300599C.svg?style=for-the-badge&logo=c%2B%2B&logoColor=white)

This is a toy programming language and its compiler made with [LLVM](https://llvm.org/) and C++. I have implemented all of the frontend components the language needs:

- Lexing

- Abstract Syntax Tree (AST)

- Parsing

Along with the code the generation to generate the [LLVM IR (LLVM Intermediate Representation)](https://llvm.org/docs/LangRef.html). Also I added a [JIT](https://en.wikipedia.org/wiki/Just-in-time_compilation) Driver so I can see the IR generated, and the evaluated results. 





And Finally I Added additional [Optimizers](https://en.wikipedia.org/wiki/Optimizing_compiler#:~:text=Compiler%20optimization%20is%20generally%20implemented,fewer%20resources%20or%20executes%20faster.) to the compiler such as:

- [Constant Folding &  Constant Propagation](https://en.wikipedia.org/wiki/Constant_folding)

- [Common Subexpression Elimination](https://en.wikipedia.org/wiki/Common_subexpression_elimination)

- [Peephole Optimization](https://en.wikipedia.org/wiki/Peephole_optimization)

- [Dead-Code Elimination](https://en.wikipedia.org/wiki/Dead-code_elimination)

## Optimization Example

### Before adding optimization

This screenshot shows the corresponding LLVM IR for the function I defined, the function basically computes $(x + 3)^2$, but notice in the IR that the compiler instead of calculating $(x + 3)$ once then squaring the result, what it actually does is computing it twice, the first time it stores it in **addtmp** and the other in **addtmp1**:

![Screenshot from 2024-02-10 23-47-22](https://github.com/OmarAzizi/Kaleidoscope/assets/110500643/7e7d9146-1a1b-44a8-bde7-313bd3290b54)

### After adding optimization

This screenshot shows the IR for the same function definition from earlier after I added the **Common Subexpression Elimination** optimization:

![Screenshot from 2024-02-11 00-18-45](https://github.com/OmarAzizi/Kaleidoscope/assets/110500643/6481a5c5-2db0-4a27-9eb0-7e8246a77823)

### Note

This is just something I made to teach myself more about compilers in general and the LLVM compiler infrastructure in particular, so I didn't follow software engineering best practices and its not production ready.