ThreadBench is a powerful, modern stability and stress testing tool built for hardware enthusiasts, overclockers, and system tuners. Unlike conventional stress tests that simply apply a continuous blank load across all cores, ThreadBench is specifically engineered to detect instability that appears during variable loads, erratic core-hopping, and aggressive undervolting.

ThreadBench runs entirely locally through a browser-based command center, making it fully portable across Windows, macOS, and Linux.

Features

• Hardware Instability Detection: ThreadBench doesn't just burn your CPU; it calculates extreme math functions twice per cycle and compares the bits. If an undervolted CPU flips a bit due to instability, ThreadBench catches it and logs the hardware fault instantly in the terminal.
• Multiple Stress Algorithms:
  • INT: Linear Congruential Generator (High-Density Arithmetic)
  • FP32: Matrix Multiplication (Floating Point Unit Heavy)
  • BRANCH: Prime Sieve (Cache and Branch Prediction Heavy)
  • GPU WebGL: True native-resolution WebGL fractal burner designed to max out iGPUs and dedicated hardware.
• Dynamic Topology Map: A live visual grid of your CPU topology. See exactly which cores are under load, and simply click a core to explicitly exclude it from the active stress test.
• Live Telemetry Engine: Millisecond-accurate monitoring of CPU Frequency, Total Package Temperature, RAM Allocation, and per-core utilization.
• Integrated System Console: A full-width terminal console that tracks exact testing stages, time remaining, and spits out live warnings if instability logic faults are detected.
• Zero Bloat: Runs entirely off Python with no heavy desktop frameworks (like Electron). Just double click and open your browser.

Requirements

• Python 3.8 or higher
• A modern Web Browser (Chrome, Firefox, Edge, Safari)

Installation & Usage

1. Clone the repository and install the requirements:

   git clone https://github.com/williamracinal/ThreadBench.git
   cd ThreadBench
   pip install -r requirements.txt

2. Launch:

   python main.py

3. Engage: The application will automatically launch the Flask backend and open the Command Center dashboard http://localhost:5000 in your default browser.

Disclaimer

NO WARRANTY: ThreadBench is designed to push your processor and integrated graphics to their absolute thermal and architectural limits. By using this software, you acknowledge that the authors are not responsible for any hardware degradation, thermal throttling, system crashes, data loss, or silicon damage. Please monitor your thermals.

Architecture

• Backend: Python Flask providing a lightweight REST API and websocket-style polling. Python's multiprocessing library is used to bypass the GIL and directly pin exact worker threads to specific physical hardware cores.
• Hardware Interfacing: psutil handles raw telemetry polling and process affinity bindings.
• Frontend: High-performance Vanilla JavaScript, HTML5, and CSS3, engineered to utilize zero external bloated web frameworks. WebGL handles raw GPU compute shaders directly in the browser canvas.

License

This project is open-source. See the LICENSE file for details.