Step-by-step interactive animations of CUDA kernels. Run them, single-step them, and watch data move between global and shared memory.