"Can machines think?" — Alan Turing, 1950.
This project says: Yes. Even on a 4GB Intel Core i3.
GPGPU AI Task Simulator is not just a simulator. It is a proof-of-concept, a whisper to the Turing Association and AI research community that intelligence can emerge even on minimal systems using open technologies and smart computation.
Crafted entirely in Python, powered by OpenCL, this tool simulates AI matrix workloads such as Matrix Multiplication, ReLU, Sigmoid, Softmax, and Tanh with GPU acceleration—no NVIDIA required.
This is AI for the edge, the curious, and the conscious engineer.
In the spirit of Alan Turing, who explored the very possibility of machine intelligence, this project explores its accessibility.
Why should AI require datacenters? Why can’t it run on what you already have?
This simulator is:
- A tribute to computational minimalism
- A sandbox for educational exploration
- A technical showcase for GPGPU parallelism in AI
| Task | Method | Acceleration |
|---|---|---|
| Matrix Multiplication | GPU via OpenCL | Yes ⚡ |
| ReLU | GPU via OpenCL | Yes ⚡ |
| Sigmoid | GPU via OpenCL | Yes ⚡ |
| Softmax | CPU (vectorized) | Partial ⚖ |
| Tanh | CPU (vectorized) | Partial ⚖ |
- Clean CustomTkinter GUI
- Console-based CLI option
- CSV input/output file support
- GPU vs CPU comparison (time + accuracy)
- Modular codebase: easy to extend
- Python 3.10
- PyOpenCL: GPU computation
- NumPy: Numerical validation
- CustomTkinter: GUI
- OpenCL kernels for core GPU ops
GPGPU_AI_Simulator/
├── gui.py
├── simulator.py
├── tasks/
│ ├── relu_task.py
│ ├── sigmoid_task.py
│ ├── matmul_task.py
│ └── ...
├── kernels/
│ ├── relu.cl
│ ├── sigmoid.cl
│ ├── matmul.cl
│ └── ...
├── input_files/
├── output_files/
└── README.md
python gui.py- Choose your task
- Select CSV files
- Click Run
python simulator.py- Menu-driven
- Terminal I/O
input.csv --> RELU (GPU) --> output.csv
inputA.csv + inputB.csv --> MATMUL --> result.csv
Results show:
- Input matrix
- GPU result
- CPU result
- GPU time vs CPU time
- Match/mismatch verification
This project is:
- A call to think small, compute big
- A blueprint for Edge AI without dependency
- A tool for AI education on low-resource devices
In a world obsessed with billion-parameter models and trillion-dollar data centers, this project is a Turing Test for practicality.
- ☑️ Conv2D kernel in OpenCL
- ☑️ On-GPU Softmax implementation
- ☑️ Batch processing + real-time graphs
- ☑️ Packaging as
.exeand.deb - ☑️ Web interface using Flask + React
- AI & ML Students
- GPGPU & Systems Programming Learners
- Professors teaching computational theory
- Researchers validating AI hardware abstraction
- Applicants to research programs / MS abroad
Karthick Anandh RJ
Deep Learning Explorer | GPGPU Hobbyist | Builder on the Edge
"I wanted to prove to myself that I didn’t need NVIDIA to simulate intelligence."
- GitHub: [https://github.com/karthick-anandh-13]
- Email: [[email protected]]
This project is licensed under the MIT License.
Use it. Fork it. Break it. Rebuild it better.
"Sometimes it is the people no one imagines anything of who do the things that no one can imagine."
— The Imitation Game
This project is a tribute to Turing’s spirit. If you're part of the Turing Association or a community devoted to practical intelligence, we invite you to explore, contribute, and critique.
This isn’t just about AI. It’s about access to intelligence.
Let the simulation begin.
⚡ GPGPU AI Task Simulator
"# gpgpu-ai-task-simulator"