CUDA Kernel Programming

Use this note for CUDA's execution hierarchy, launch model, synchronization, memory access patterns, and the progression from naive to tiled kernels.