View of When Does Disaggregated, Cache-Aware LLM Serving Pay Off? An Analytical Crossover Model Validated on Real Compute, Network, and Coordination Measurements

Return to Article Details When Does Disaggregated, Cache-Aware LLM Serving Pay Off? An Analytical Crossover Model Validated on Real Compute, Network, and Coordination Measurements Download Download PDF