arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.11530cs.LGcs.AR
收藏

PASCAL:一种面向推理型大语言模型的服务调度算法

PASCAL: A Phase-Aware Scheduling Algorithm for Serving Reasoning-based Large Language Models

KAIST(韩国科学技术院)

Eunyeong Cho, Jehyeon Bang, Ranggi Hwang, Minsoo Rhu

AI总结:

PASCAL通过优先处理推理阶段和控制回答阶段的令牌节奏,有效降低了推理型大语言模型的TTFT并保持服务等级目标。

摘要:

推理型大语言模型的出现引入了新的服务挑战,因为它们的扩展推理阶段会延迟用户可见的输出并增加时间到第一个令牌(TTFT)。现有的LLM服务框架无法区分推理和回答阶段,导致在GPU内存限制下性能下降。我们提出了PASCAL,一种具有阶段意识的调度算法,该算法优先处理推理以减少TTFT,同时在回答阶段使用受控的抢占和令牌节奏以保持用户体验质量(QoE)。我们的分层调度器结合了实例级放置与实例内执行,并在阶段边界处启用动态迁移,以平衡负载并减少干扰。在使用DeepSeek-R1-Distill-Qwen-32B的基准测试中,PASCAL将尾部TTFT减少了高达72%,同时保持回答阶段的服务等级目标(SLO)的达成,证明了阶段意识调度对于推理型大语言模型部署的重要性。

英文摘要:

The emergence of reasoning-based LLMs leveraging Chain-of-Thought (CoT) inference introduces new serving challenges, as their extended reasoning phases delay user-visible output and inflate Time-To-First-Token (TTFT). Existing LLM serving frameworks fail to distinguish between reasoning and answering phases, leading to performance degradation under GPU memory constraints. We present PASCAL, a phase-aware scheduling algorithm that prioritizes reasoning to reduce TTFT while using controlled preemption and token pacing during answering to preserve Quality-of-Experience (QoE). Our hierarchical scheduler combines instance-level placement with intra-instance execution and enables dynamic migration at phase boundaries to balance load and reduce interference. Across benchmarks using DeepSeek-R1-Distill-Qwen-32B, PASCAL reduces tail TTFT by up to 72% while maintaining answering phase SLO attainment, demonstrating the importance of phase-aware scheduling for reasoning-based LLM deployment.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑