playlists
CUDA GRAPHS
- Part 1: Intro & Basics
- Part 2: What is LLM Inference and How CUDA Graphs Improve It
- Part 3: Memory Management, Capture Safety, Graph Updates
TORCH.COMPILE
- Part 1: Why LLM Inference Frameworks Use torch.compile
- Part 2: How Does PyTorch Compile Regular Python Code?
- Part 3: How TorchDynamo Captures Python Execution
- Part 4: Understanding FX Graphs in PyTorch