SMG: The Case for Disaggregating CPU from GPU in LLM Serving

(pytorch.org) by gmays | May 5, 2026 | 0 comments on HN