AI032
Programming Massively Parallel Processors: A Hands-on Approach
Introduction to Parallel Computing and GPU Architectures
Learning Objectives
- Differentiate between serial latency-oriented and parallel throughput-oriented processing.
- Understand the evolution of the Graphics Processing Unit (GPU) from a fixed-function pipe to a general-purpose processor.
- Explain the Flynn's Taxonomy with a focus on SIMD and SIMT models.
- Identify the core components of the NVIDIA GPU architecture, including Streaming Multiprocessors and the thread hierarchy.
- Evaluate the performance impact of memory bandwidth and latency in parallel applications.