Write GPU kernels in your browser, run them on a real GPU

Write kernels in CUDA, Triton or PyTorch, then submit them to a real T4 for judging.

Courses

Learn CUDA one mechanism at a time, from your first kernel to tiled matrix multiplication: get it right first.

GPU kernel practice for LLMs in CUDA, Triton and PyTorch: get it right first, then get it fast.

3 stagesCUDA / Triton / PyTorchT4