Pollux: Co-adaptive Cluster Scheduling for Goodput-Optimized Deep Learning

• URL: http://arxiv.org/abs/2008.12260v2
• Date: Wed, 26 May 2021 06:08:21 GMT
• Title: Pollux: Co-adaptive Cluster Scheduling for Goodput-Optimized Deep Learning
• Authors: Aurick Qiao, Sang Keun Choe, Suhas Jayaram Subramanya, Willie Neiswanger, Qirong Ho, Hao Zhang, Gregory R. Ganger, Eric P. Xing
• Abstract summary: Pollux improves scheduling performance in deep learning (DL) clusters by adaptively co-optimizing inter-dependent factors. Pollux reduces average job completion times by 37-50% relative to state-of-the-art DL schedulers.
• Score: 61.29990368322931
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Pollux improves scheduling performance in deep learning (DL) clusters by adaptively co-optimizing inter-dependent factors both at the per-job level and at the cluster-wide level. Most existing schedulers expect users to specify the number of resources for each job, often leading to inefficient resource use. Some recent schedulers choose job resources for users, but do so without awareness of how DL training can be re-optimized to better utilize the provided resources. Pollux simultaneously considers both aspects. By monitoring the status of each job during training, Pollux models how their goodput (a novel metric we introduce that combines system throughput with statistical efficiency) would change by adding or removing resources. Leveraging these information, Pollux dynamically (re-)assigns resources to improve cluster-wide goodput, while respecting fairness and continually optimizing each DL job to better utilize those resources. In experiments with real DL jobs and with trace-driven simulations, Pollux reduces average job completion times by 37-50% relative to state-of-the-art DL schedulers, even when they are provided with ideal resource and training configurations for every job. Pollux promotes fairness among DL jobs competing for resources based on a more meaningful measure of useful job progress, and reveals a new opportunity for reducing DL cost in cloud environments. Pollux is implemented and publicly available as part of an open-source project at https://github.com/petuum/adaptdl.

Related papers

• StreamRL: Scalable, Heterogeneous, and Elastic RL for LLMs with Disaggregated Stream Generation [55.75008325187133]
  Reinforcement learning (RL) has become the core post-training technique for large language models (LLMs) StreamRL is designed with disaggregation from first principles to address two types of performance bottlenecks. Experiments show that StreamRL improves throughput by up to 2.66x compared to existing state-of-the-art systems.
  arXiv  Detail & Related papers  (2025-04-22T14:19:06Z)
• Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast, Scalable LLM Post-Training [71.16258800411696]
  Reinforcement learning (RL) is a critical component of large language model (LLM) post-training. Existing on-policy algorithms used for post-training are inherently incompatible with the use of experience replay buffers. We propose efficiently obtaining this benefit of replay buffers via Trajectory Balance with Asynchrony (TBA)
  arXiv  Detail & Related papers  (2025-03-24T17:51:39Z)
• OmniLearn: A Framework for Distributed Deep Learning over Heterogeneous Clusters [1.4131700241686853]
  We develop an adaptive batch-scaling framework called OmniLearn to mitigate the effects of heterogeneous resources. Our approach is inspired by proportional controllers to balance across heterogeneous servers, and works under varying resource availability.
  arXiv  Detail & Related papers  (2025-03-21T18:26:24Z)
• Resource Heterogeneity-Aware and Utilization-Enhanced Scheduling for Deep Learning Clusters [26.874684454125152]
  em Hadar is a task-level scheduler based on an optimization framework that can boost resource utilization. em HadarE exhibits considerable speed-ups in DL model training, reducing the total time duration by 50% (or 80%) on an Amazon's AWS (or our lab) cluster.
  arXiv  Detail & Related papers  (2025-03-13T22:13:20Z)
• When Does Visual Prompting Outperform Linear Probing for Vision-Language Models? A Likelihood Perspective [57.05315507519704]
  We propose a log-likelihood ratio (LLR) approach to analyze the comparative benefits of visual prompting and linear probing. Our measure attains up to a 100-fold reduction in run time compared to full training, while achieving prediction accuracies up to 91%.
  arXiv  Detail & Related papers  (2024-09-03T12:03:45Z)
• Prune at the Clients, Not the Server: Accelerated Sparse Training in Federated Learning [56.21666819468249]
  Resource constraints of clients and communication costs pose major problems for training large models in Federated Learning. We introduce Sparse-ProxSkip, which combines training and acceleration in a sparse setting. We demonstrate the good performance of Sparse-ProxSkip in extensive experiments.
  arXiv  Detail & Related papers  (2024-05-31T05:21:12Z)
• How to Train Data-Efficient LLMs [56.41105687693619]
  We study data-efficient approaches for pre-training language models (LLMs) We find that Ask-LLM and Density sampling are the best methods in their respective categories. In our comparison of 19 samplers, involving hundreds of evaluation tasks and pre-training runs, we find that Ask-LLM and Density are the best methods in their respective categories.
  arXiv  Detail & Related papers  (2024-02-15T02:27:57Z)
• Venn: Resource Management for Collaborative Learning Jobs [24.596584073531886]
  Collaborative learning (CL) has emerged as a promising approach for machine learning (ML) and data science across distributed edge devices. In this paper, we present Venn, a CL resource manager that efficiently schedules heterogeneous devices among multiple CL jobs. Our evaluation shows that, compared to the state-of-the-art CL resource managers, Venn improves the average JCT by up to 1.88x.
  arXiv  Detail & Related papers  (2023-12-13T17:13:08Z)
• Taming Resource Heterogeneity In Distributed ML Training With Dynamic Batching [1.047192732651018]
  Current techniques for distributed model training mostly assume that clusters are comprised of servers with a constant resource availability. We develop a dynamic technique for distributed data-parallel training that adjusts the mini-batch sizes on each worker based on availability and throughput.
  arXiv  Detail & Related papers  (2023-05-20T15:33:06Z)
• Learning to Optimize Permutation Flow Shop Scheduling via Graph-based Imitation Learning [70.65666982566655]
  Permutation flow shop scheduling (PFSS) is widely used in manufacturing systems. We propose to train the model via expert-driven imitation learning, which accelerates convergence more stably and accurately. Our model's network parameters are reduced to only 37% of theirs, and the solution gap of our model towards the expert solutions decreases from 6.8% to 1.3% on average.
  arXiv  Detail & Related papers  (2022-10-31T09:46:26Z)
• HeterPS: Distributed Deep Learning With Reinforcement Learning Based Scheduling in Heterogeneous Environments [37.55572042288321]
  Training process of neural networks (DNNs) generally handles large-scale input data with many sparse features. Paddle-HeterPS is composed of a distributed architecture and a Reinforcement Reinforcement (RL)-based scheduling method. We show that Paddle-HeterPS significantly outperforms state-of-the-art approaches in terms of throughput (14.5 times higher) and monetary cost (312.3% smaller)
  arXiv  Detail & Related papers  (2021-11-20T17:09:15Z)
• Synergy: Resource Sensitive DNN Scheduling in Multi-Tenant Clusters [10.38396444951436]
  Training Deep Neural Networks (DNNs) is a widely popular workload in both enterprises and cloud data centers. We propose Synergy, a resource-sensitive scheduler for shared GPU clusters. Our experiments show that workload-aware CPU and memory allocations can improve average JCT up to 3.4x when compared to traditional GPU-proportional scheduling.
  arXiv  Detail & Related papers  (2021-10-12T15:25:54Z)
• A Predictive Autoscaler for Elastic Batch Jobs [8.354712625979776]
  Large batch jobs such as Deep Learning, HPC and Spark require far more computational resources and higher cost than conventional online service. We propose a predictive autoscaler to provide an elastic interface for the customers and overprovision instances.
  arXiv  Detail & Related papers  (2020-10-10T17:35:55Z)
• Effective Elastic Scaling of Deep Learning Workloads [3.345876096131764]
  We examine the elastic scaling of Deep Learning (DL) jobs over large-scale training platforms. We propose a novel resource allocation strategy for DL training jobs, resulting in improved job run time performance as well as increased cluster utilization.
  arXiv  Detail & Related papers  (2020-06-24T17:01:09Z)
• Joint Parameter-and-Bandwidth Allocation for Improving the Efficiency of Partitioned Edge Learning [73.82875010696849]
  Machine learning algorithms are deployed at the network edge for training artificial intelligence (AI) models. This paper focuses on the novel joint design of parameter (computation load) allocation and bandwidth allocation.
  arXiv  Detail & Related papers  (2020-03-10T05:52:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.