COMET: A Comprehensive Cluster Design Methodology for Distributed Deep Learning Training

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kadiyala, Divya Kiran, Rashidi, Saeed, Heo, Taekyung, Bambhaniya, Abhimanyu Rajeshkumar, Krishna, Tushar, Daglis, Alexandros
Format: Preprint
Published: 2022
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910366335762432
author Kadiyala, Divya Kiran
Rashidi, Saeed
Heo, Taekyung
Bambhaniya, Abhimanyu Rajeshkumar
Krishna, Tushar
Daglis, Alexandros
author_facet Kadiyala, Divya Kiran
Rashidi, Saeed
Heo, Taekyung
Bambhaniya, Abhimanyu Rajeshkumar
Krishna, Tushar
Daglis, Alexandros
contents Modern Deep Learning (DL) models have grown to sizes requiring massive clusters of specialized, high-end nodes to train. Designing such clusters to maximize both performance and utilization--to amortize their steep cost--is a challenging task requiring careful balance of compute, memory, and network resources. Moreover, a plethora of each model's tuning knobs drastically affect the performance, with optimal values often depending on the underlying cluster's characteristics, which necessitates a complex cluster-workload co-design process. To facilitate the design space exploration of such massive DL training clusters, we introduce COMET, a holistic cluster design methodology and workflow to jointly study the impact of parallelization strategies and key cluster resource provisioning on the performance of distributed DL training. We develop a step-by-step process to establish a reusable and flexible methodology, and demonstrate its application with case studies of training large models on cluster configurations of variable compute, memory, and network resources. Our case studies demonstrate COMET's utility in identifying promising architectural optimization directions and guiding system designers in configuring key model and cluster parameters. To illustrate, cluster configuration comparisons identify performance differences of up to 7.7x and highlight performance optimization opportunities of up to 1.4x when employing memory expansion as an optimization technique.
format Preprint
id arxiv_https___arxiv_org_abs_2211_16648
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle COMET: A Comprehensive Cluster Design Methodology for Distributed Deep Learning Training
Kadiyala, Divya Kiran
Rashidi, Saeed
Heo, Taekyung
Bambhaniya, Abhimanyu Rajeshkumar
Krishna, Tushar
Daglis, Alexandros
Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Machine Learning
Modern Deep Learning (DL) models have grown to sizes requiring massive clusters of specialized, high-end nodes to train. Designing such clusters to maximize both performance and utilization--to amortize their steep cost--is a challenging task requiring careful balance of compute, memory, and network resources. Moreover, a plethora of each model's tuning knobs drastically affect the performance, with optimal values often depending on the underlying cluster's characteristics, which necessitates a complex cluster-workload co-design process. To facilitate the design space exploration of such massive DL training clusters, we introduce COMET, a holistic cluster design methodology and workflow to jointly study the impact of parallelization strategies and key cluster resource provisioning on the performance of distributed DL training. We develop a step-by-step process to establish a reusable and flexible methodology, and demonstrate its application with case studies of training large models on cluster configurations of variable compute, memory, and network resources. Our case studies demonstrate COMET's utility in identifying promising architectural optimization directions and guiding system designers in configuring key model and cluster parameters. To illustrate, cluster configuration comparisons identify performance differences of up to 7.7x and highlight performance optimization opportunities of up to 1.4x when employing memory expansion as an optimization technique.
title COMET: A Comprehensive Cluster Design Methodology for Distributed Deep Learning Training
topic Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2211.16648