5  Summary

In summary, we have looked at GPU architecture, threads, blocks, grids and programming with CUDA in C and Python.

To dive further into the world of accelerators, here are some suggested resources:

Nvidia courses

GPU Puzzles - very popular GitHub repo

Learn GPU Programming in Your Browser – from answer.ai, uses WebGPU Shading Language.

PyTorch CUDA backend – Look at some of the PyTorch CUDA backend code samples.