← back to notes

GPU Computing 101

I created this repository to learn the fundamentals of GPU computation from scratch. The goal was to build intuition for parallel programming by starting with simple examples and gradually increasing complexity.

I used Claude Code to help me create simple, understandable examples. Each example includes detailed FAQ sections answering the questions I had while learning - documented directly in the source code comments for future reference.

CUDA Examples

I started with CUDA examples to understand GPU concepts in the "standard" library that most GPU programming is built on. The progression goes from basic kernel execution to real-world image processing:

  • Basic GPU kernel execution and memory management
  • Thread organization with blocks and grids
  • Parallel matrix multiplication
  • Image blur processing

WebGPU Examples

After understanding the core concepts in CUDA, I moved to WebGPU examples to apply the same principles in the browser. These include compute shaders, 2D/3D rendering, and a WGSL language reference.

The WebGPU examples are hosted as live interactive demos - you can run them directly in your browser.

Source

Full repository: github.com/shipurjan/gpu-playground