AI032
Programming Massively Parallel Processors: A Hands-on Approach
CUDA Program Structure and Memory Management
Learning Objectives
- Understand the CUDA thread hierarchy (grids, blocks, and threads)
- Distinguish between different GPU memory types (global, shared, local, constant, and texture)
- Learn the lifecycle of a CUDA kernel from host to device
- Implement basic memory allocation and transfer using cudaMalloc and cudaMemcpy