Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service

This paper presents a solution to the challenge of mitigating carbon emissions from hosting large-scale machine learning (ML) inference services. ML inference is critical to modern technology products, but it is also a significant contributor to carbon footprint. We introduce, Clover, a carbon-frien...

Full description

Bibliographic Details
Main Authors: Li, Baolin, Samsi, Siddharth, Gadepally, Vijay, Tiwari, Devesh
Other Authors: Lincoln Laboratory
Format: Article
Language:English
Published: ACM|The International Conference for High Performance Computing, Networking, Storage and Analysis 2023
Online Access:https://hdl.handle.net/1721.1/153142
_version_ 1811089894670336000
author Li, Baolin
Samsi, Siddharth
Gadepally, Vijay
Tiwari, Devesh
author2 Lincoln Laboratory
author_facet Lincoln Laboratory
Li, Baolin
Samsi, Siddharth
Gadepally, Vijay
Tiwari, Devesh
author_sort Li, Baolin
collection MIT
description This paper presents a solution to the challenge of mitigating carbon emissions from hosting large-scale machine learning (ML) inference services. ML inference is critical to modern technology products, but it is also a significant contributor to carbon footprint. We introduce, Clover, a carbon-friendly ML inference service runtime system that balances performance, accuracy, and carbon emissions through mixed-quality models and GPU resource partitioning. Our experimental results demonstrate that Clover is effective in substantially reducing carbon emissions while maintaining high accuracy and meeting service level agreement (SLA) targets.
first_indexed 2024-09-23T14:26:48Z
format Article
id mit-1721.1/153142
institution Massachusetts Institute of Technology
language English
last_indexed 2024-09-23T14:26:48Z
publishDate 2023
publisher ACM|The International Conference for High Performance Computing, Networking, Storage and Analysis
record_format dspace
spelling mit-1721.1/1531422024-01-23T18:17:42Z Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service Li, Baolin Samsi, Siddharth Gadepally, Vijay Tiwari, Devesh Lincoln Laboratory This paper presents a solution to the challenge of mitigating carbon emissions from hosting large-scale machine learning (ML) inference services. ML inference is critical to modern technology products, but it is also a significant contributor to carbon footprint. We introduce, Clover, a carbon-friendly ML inference service runtime system that balances performance, accuracy, and carbon emissions through mixed-quality models and GPU resource partitioning. Our experimental results demonstrate that Clover is effective in substantially reducing carbon emissions while maintaining high accuracy and meeting service level agreement (SLA) targets. 2023-12-12T14:34:04Z 2023-12-12T14:34:04Z 2023-11-12 2023-12-01T08:46:33Z Article http://purl.org/eprint/type/ConferencePaper 979-8-4007-0109-2 https://hdl.handle.net/1721.1/153142 Li, Baolin, Samsi, Siddharth, Gadepally, Vijay and Tiwari, Devesh. 2023. "Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service." PUBLISHER_POLICY en https://doi.org/10.1145/3581784.3607034 Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. The author(s) application/pdf ACM|The International Conference for High Performance Computing, Networking, Storage and Analysis Association for Computing Machinery
spellingShingle Li, Baolin
Samsi, Siddharth
Gadepally, Vijay
Tiwari, Devesh
Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service
title Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service
title_full Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service
title_fullStr Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service
title_full_unstemmed Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service
title_short Clover: Toward Sustainable AI with Carbon-Aware Machine Learning Inference Service
title_sort clover toward sustainable ai with carbon aware machine learning inference service
url https://hdl.handle.net/1721.1/153142
work_keys_str_mv AT libaolin clovertowardsustainableaiwithcarbonawaremachinelearninginferenceservice
AT samsisiddharth clovertowardsustainableaiwithcarbonawaremachinelearninginferenceservice
AT gadepallyvijay clovertowardsustainableaiwithcarbonawaremachinelearninginferenceservice
AT tiwaridevesh clovertowardsustainableaiwithcarbonawaremachinelearninginferenceservice