FATE reduces normalized makespan and P95 latency in real LLM workflow DAGs to 0.675 and 0.677 by jointly preserving multiple future execution states, outperforming RoundRobin by 32.5% and the strongest baseline by 8.9%.
Distserve: Disaggregating prefill and decoding for goodput-optimized large language model serving
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
FATE: Future-State-Aware Scheduling for Heterogeneous LLM Workflows
FATE reduces normalized makespan and P95 latency in real LLM workflow DAGs to 0.675 and 0.677 by jointly preserving multiple future execution states, outperforming RoundRobin by 32.5% and the strongest baseline by 8.9%.