Optimizing vLLM for Intel CPUs and XPUs | Ray Summit 2024 | Anyscale | Podwise