用图匹配优化深度学习任务调度，提升集群资源利用率。
Tesserae: Scalable Placement Policies for Deep Learning Workloads
- 将调度约束转为图匹配问题，设计可扩展的放置策略
- 相比现有调度器，平均任务完成时间缩短41%，整体完成时间缩短15%
- 适合大规模GPU集群中的深度学习工作负载调度
训练深度学习（DL）模型已成为数据中心的主要工作负载，提升资源利用率是DL集群调度器的核心目标。当前调度器通常采用启发式或复杂优化中的约束来决定任务部署位置，但普遍存在性能不佳或扩展性差的问题。本文提出关键洞察：多数放置约束可建模为图匹配问题，并据此设计了新型放置策略，以减少任务迁移开销和提高任务打包效率。这些策略被集成到Tesserae系统中，实现高效且可扩展的GPU集群调度。实验表明，与现有调度器相比，Tesserae在平均作业完成时间（JCT）上最多提升1.62倍，在总工期（Makespan）上最多提升1.15倍。
原文摘要 · Abstract (English)
Training deep learning (DL) models has become a dominant workload in data-centers and improving resource utilization is a key goal of DL cluster schedulers. In order to do this, schedulers typically incorporate placement policies that govern where jobs are placed on the cluster. Existing placement policies are either designed as ad-hoc heuristics or incorporated as constraints within a complex optimization problem and thus either suffer from suboptimal performance or poor scalability. Our key insight is that many placement constraints can be formulated as graph matching problems and based on that we design novel placement policies for minimizing job migration overheads and job packing. We integrate these policies into Tesserae and describe how our design leads to a scalable and effective GPU cluster scheduler. Our experimental results show that Tesserae improves average JCT by up to 1.62x and the Makespan by up to 1.15x compared with the existing schedulers.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。