Design of Hardware-Aware LLM Inference Engine: System-level Optimization from Architecture to Implementation

Introduction / Main Floor: Design of Hardware-Aware LLM Inference Engine: System-level Optimization from Architecture to Implementation

Delve into the design philosophy and implementation methods of hardware-aware LLM inference engines, covering system-level collaborative optimization strategies for key technologies such as GPU/CPU heterogeneous computing, memory hierarchy optimization, and operator fusion.