一句话：提出非侵入式、可扩展且可解释的深度学习训练作业调度器 Lucid，通过可解释模型与惰性打包降低平均完成时间。
现有深度学习工作负载调度器性能优异，但存在侵入性强、集成与维护成本高、可扩展性受限及决策不透明等部署缺陷。本文设计并实现 Lucid，一种基于可解释模型的非侵入式调度器。其包含三个创新模块：二维优化 profiler 用于高效采集作业指标并反馈调试信息；惰性打包策略以规避干扰；基于估计作业优先级值和共享分数编排资源。此外，系统优化器支持模型性能维护与透明调整。评估表明，Lucid 相比先进抢占式调度器 Tiresias 将平均作业完成时间最多加速 1.3 倍，并提供显式系统解释和良好可扩展性。
While recent deep learning workload schedulers exhibit excellent performance, it is arduous to deploy them in practice due to some substantial defects, including inflexible intrusive manner, exorbitant integration and maintenance cost, limited scalability, as well as opaque decision processes. Motivated by these issues, we design and implement Lucid, a non-intrusive deep learning workload scheduler based on interpretable models. It consists of three innovative modules. First, a two-dimensional optimized profiler is introduced for efficient job metric collection and timely debugging job feedback. Second, Lucid utilizes an indolent packing strategy to circumvent interference. Third, Lucid orchestrates resources based on estimated job priority values and sharing scores to achieve efficient scheduling. Additionally, Lucid promotes model performance maintenance and system transparent adjustment via a well-designed system optimizer. Our evaluation shows that Lucid reduces the average job completion time by up to 1.3× compared with state-of-the-art preemptive scheduler Tiresias. Furthermore, it provides explicit system interpretations and excellent scalability for practical deployment.
@inproceedings{paperbot10033,
title = {Lucid: A Non-intrusive, Scalable and Interpretable Scheduler for Deep Learning Training Jobs},
author = {Qinghao Hu and Meng Zhang and Peng Sun and Yonggang Wen and Tianwei Zhang},
booktitle = {ASPLOS},
year = {2023},
doi = {10.1145/3575693.3575705}
}