arXivDaily arXiv每日学术速递 周一至周五更新
看起来今天可以划水了
arXiv 2201.03175cs.DCcs.AI
收藏

一个基于数千GPU的多租户机器学习服务仿真平台

A Simulation Platform for Multi-tenant Machine Learning Services on Thousands of GPUs

University of Toronto(多伦多大学);National University of Singapore(新加坡国立大学);SenseTime(商汤科技)

Ruofan Liang, Bingsheng He, Shengen Yan, Peng Sun

摘要:

多租户机器学习服务已经成为数据中心中新兴的数据密集型工作负载,其大量使用GPU资源。由于集群规模大、调整参数多、资源占用量大,在真实集群上对这些机器学习服务进行评估和基准测试通常是不切实际的。在这个演示中,我们提出了AnalySIM,一个集群模拟器,允许多租户机器学习服务的有效设计探索。具体来说,通过跟踪驱动的集群工作负载模拟,AnalySIM可以轻松测试和分析各种调度策略在GPU资源利用率等多个性能指标中的表现。AnalySIM基于物理拓扑和逻辑分区模拟集群计算资源。该工具已在SenseTime中使用,以了解不同调度策略的影响,并跟踪超过1000个GPU的实际生产集群。我们发现,抢占和迁移能够显着减少平均作业完成时间,减轻资源碎片化问题。

英文摘要:

Multi-tenant machine learning services have become emerging data-intensive workloads in data centers with heavy usage of GPU resources. Due to the large scale, many tuning parameters and heavy resource usage, it is usually impractical to evaluate and benchmark those machine learning services on real clusters. In this demonstration, we present AnalySIM, a cluster simulator that allows efficient design explorations for multi-tenant machine learning services. Specifically, by trace-driven cluster workload simulation, AnalySIM can easily test and analyze various scheduling policies in a number of performance metrics such as GPU resource utilization. AnalySIM simulates the cluster computational resource based on both physical topology and logical partition. The tool has been used in SenseTime to understand the impact of different scheduling policies with the trace from a real production cluster of over 1000 GPUs. We find that preemption and migration are able to significantly reduce average job completion time and mitigate the resource fragmentation problem.

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。

↑