Fetching the paper…

Characterizing and Optimizing LLM Inference Workloads on CPU-GPU Coupled Architectures · Around