Shao-Nan Ma

PDScope: Breaking the Prefill-Decode Seesaw for Memory-Bound MoE Inference

Local deployment of MoE-based LLMs on CPU-GPU nodes is attractive for cost and privacy reasons, yet limited GPU memory forces most experts into CPU memory. Prefill is then compute-bound and benefits from aggressive GPU loading, while decode is I/O-bound and requires conservative PCIe transfers. These demands cannot be...