Welcome to the Language Benchmarks repository, a comprehensive benchmarking suite designed to measure and compare the performance of various programming languages in executing CPU-intensive tasks.
This repository hosts a set of multi-threaded benchmark programs that execute a busy loop, leveraging the full CPU power for a fixed duration of 10 seconds. The benchmarks are written in several popular programming languages, including C++, Go, Java, JavaScript (Node.js), PHP, Python, and Rust.
The benchmarking approach used here can be extended to a wide range of tasks such as sorting arrays, mathematical computations, string manipulations, file system operations, HTTP requests, and JSON encoding/decoding. This project serves as a template for developers to create their own benchmarks and contribute to a growing library of performance tests.
Before diving into the benchmarks, please consider starring and forking the repository on GitHub to make your own modifications:
https://github.com/discommand2/language_benchmarks
After forking, clone your forked repository to your local machine:
git clone https://github.com/YOUR_USERNAME/language_benchmarks.git
cd language_benchmarks
Run the make.sh script to compile all benchmark programs:
./make.sh
Execute the run.sh script to run all benchmarks:
./run.sh
Each program will utilize all available CPU cores (Threads / 2) to execute as many loops as possible within the 10-second window.
This is from my bare-metal linux dedicated server.
- CPU Model: Intel Xeon-E 2386G
- CPU Cores/Threads: 6 cores, 12 threads
- CPU Base Frequency: 3.5 GHz
- CPU Max Turbo Frequency: 4.7 GHz
- RAM: 64 GB ECC
- RAM Speed: 3200 MHz
Python 3.9.18 looped 5,295,000,000 times.
Python 3.11.5 looped 5,665,000,000 times.
Python 3.12.0 looped 6,015,000,000 times.
C++ 11.4.1 looped 123,145,000,000 times.
PHP 8.4.0-dev looped 257,035,000,000 times.
go1.21.3 looped 277,775,000,000 times.
Node.js v18.14.2 looped 277,595,000,000 times.
C#.NET 6.0.24 looped 278,550,000,000 times.
Java 11.0.18 looped 4,106,230,020,000,000 times.
Rust (Debug) 1.73.0 looped 18,585,000,000 times.
Rust (Release) 1.73.0 looped 4,626,430,415,000,000 times.
The current benchmarks focus on busy loops, but the possibilities for extension are vast. Here are some ideas:
- Sorting Algorithms: Compare the performance of different sorting algorithms across languages.
- Mathematical Computations: Benchmark operations like prime number generation or matrix multiplication.
- String Manipulations: Test operations such as concatenation, searching, and regex.
- File System Operations: Measure the speed of file I/O operations.
- HTTP Requests: Evaluate the performance of web API interactions.
- JSON Encoding/Decoding: Assess the speed of data serialization and deserialization.
Contributions are welcome! If you have a benchmark to add, an improvement to suggest, or a bug to report, please submit a pull request or open an issue on the GitHub repository.
Join the vibrant community of developers on the discommand2 Discord server for support, discussions, and collaboration:
This project is licensed under the MIT License. For more details, see the LICENSE file in the repository.
Dive into the world of performance benchmarking and unleash the full potential of your applications with Language Benchmarks!