GPU Programming

CUDA.

GPU-accelerated computing with NVIDIA CUDA — parallel kernels, memory hierarchy, and deep learning acceleration.

Beginner

Start here — no prior experience needed

1 lesson

Learn CUDA’s programming model: kernels, threads/blocks/grid, device memory, and a first vector-add example.

For developers with core concepts down

1 lesson

Understand global vs shared vs registers, learn memory coalescing, and apply a tiled shared-memory matrix multiply.

Production-grade patterns for experienced engineers

1 lesson

Use Nsight Systems/Compute concepts, diagnose memory vs compute bottlenecks, apply occupancy and kernel launch optimizations.