课程视频链接： https://www.youtube.com/playlist?list=PLAwxTw4SYaPnFKojVQrmyOGFCqHTxfdv2
[TOC]
1. Unit 1
1.1 typical CUDA Program
1.2 parallel communication patterns
1.3 GPU allocate blocks to SMs
1.3 GPU memory hierarchy
1.4 high level strategies of optimizing performance
2. Unit 2
2.1 parallel communication patters - Map
2.2 parallel communication patters - Gather
2.3 parallel communication patters - Scatter
2.4 parallel communication patters - Stencil
2.5 parallel communication patters - Transpose
2.6 parallel communication patters recap
2.7 SM, Kernel, Thread Blocks, Thread
