todo.academy Try a free chapter

In development

GPU Programming with NVIDIA CUDA

Make GPU code correct first, then prove the performance work mattered.

Practice host/device flow, kernels, grids, blocks, threads, memory transfers, coalescing, timing, shared memory, occupancy, races, streams, and profiling.

Planned practice

Skills currently represented in the roadmap.

  • Explain GPU execution geometry and memory flow
  • Fix indexing, transfer, race, and memory-pattern bugs
  • Prove performance claims only after correctness
  • Reason about coalescing, shared memory, divergence, and profiling signals

Current roadmap

The detailed syllabus is not published yet.

The course is listed to collect launch interest without pretending the material is ready.

Course update

Ask for one launch notification.

Joining does not reserve a place, start a purchase, or promise a launch date.