GPU Programming

GPU Computing Fundamentals.

Core GPU architecture concepts — SIMT execution, memory coalescing, occupancy, and hardware parallelism.

Beginner

Start here — no prior experience needed

1 lesson

Learn SIMT execution, warp behavior, and the GPU memory hierarchy (registers, shared, L2, DRAM) with practical mapping tips.

For developers with core concepts down

1 lesson

Improve bandwidth using coalesced memory accesses, and understand occupancy to hide latency effectively.

Production-grade patterns for experienced engineers

1 lesson

Learn how to interpret GPU profiling signals (bandwidth, utilization) and apply a roofline-style mental model to optimize kernels.