Energy-Conscious LLM Text Generation :  Impact of Decoding Strategies on GPU Energy Usage

Energy-Conscious LLM Text Generation : Impact of Decoding Strategies on GPU Energy Usage | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Energy-Conscious LLM Text Generation : Impact of Decoding Strategies on GPU Energy Usage Alireza Nik, Michael Riegler, Pål Halvorsen This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7794007/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 11 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted 10 You are reading this latest preprint version Abstract Decoding strategies significantly influence the quality and diversity of the generated text in Large Language Models (LLMs), yet their impact on computational resources, particularly GPU energy consumption, is insufficiently studied. This paper investigates the relationship between text generation decoding techniques and energy efficiency, focusing on the trade-off between generation quality and GPU energy usage across diverse tasks and decoding configurations. By benchmarking multiple strategies across various tasks, including Translation, Math Problem Solving, Coding, and Open-ended text generation, we reveal how selecting appropriate decoding techniques with their tuned hyperparameters affects text quality and has measurable implications for energy consumption. Our findings show that the choice of decoding strategy can greatly impact GPU energy usage, even when it has a minimal effect on output quality. Different strategies also involve trade-offs between quality and energy efficiency, and no single decoding method is best in all cases across every metric. To the best of our knowledge, this is one of the first studies to examine decoding strategies in LLMs from the perspective of energy consumption, providing useful insights for building energy-efficient applications without compromising text generation quality. Physical sciences/Engineering Physical sciences/Mathematics and computing Full Text Additional Declarations No competing interests reported. Supplementary Files Supplementary.pdf Cite Share Download PDF Status: Published Journal Publication published 11 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 27 Oct, 2025 Reviews received at journal 20 Oct, 2025 Reviews received at journal 20 Oct, 2025 Reviewers agreed at journal 15 Oct, 2025 Reviewers agreed at journal 14 Oct, 2025 Reviewers invited by journal 13 Oct, 2025 Editor invited by journal 10 Oct, 2025 Editor assigned by journal 09 Oct, 2025 Submission checks completed at journal 09 Oct, 2025 First submitted to journal 06 Oct, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7794007","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":534142945,"identity":"2cd22fa0-65f1-4448-af68-4fe45b2062b7","order_by":0,"name":"Alireza Nik","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAw0lEQVRIiWNgGAWjYDACdsYGECXHwAwkHxClhRmixRisJYE4LRAqEayRKC38zcxtHz78sUuf386dJpFQw5DHT0iLxGHG5pkzeJJzNxzm3SaRcIyhWLKBkB6gFmYeCebcDcwgLWwMiRsOENAhD9Lyx6A+Xb4ZpOUfEVoMQFoYEg4nMIAclthGhBZDoBbGngPHDYF+2WyR2CeROJOQX+SOtz9m+PGnWl6+/+zGGx++2ST2E9CBASRI1TAKRsEoGAWjABsAAOznPEnUoozoAAAAAElFTkSuQmCC","orcid":"","institution":"Simula Metropolitan Center for Digital Engineering","correspondingAuthor":true,"prefix":"","firstName":"Alireza","middleName":"","lastName":"Nik","suffix":""},{"id":534142946,"identity":"d728c2b3-6df4-4dcb-8612-5c16a44ead8b","order_by":1,"name":"Michael Riegler","email":"","orcid":"","institution":"Simula Research Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"","lastName":"Riegler","suffix":""},{"id":534142947,"identity":"bf0b4216-ca00-4237-89b2-8750511de403","order_by":2,"name":"Pål Halvorsen","email":"","orcid":"","institution":"Simula Metropolitan Center for Digital Engineering","correspondingAuthor":false,"prefix":"","firstName":"Pål","middleName":"","lastName":"Halvorsen","suffix":""}],"badges":[],"createdAt":"2025-10-06 19:53:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7794007/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7794007/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-31896-0","type":"published","date":"2025-12-11T15:57:39+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":94471046,"identity":"f2fbfbec-824d-432c-a3d2-fc55ceb53135","added_by":"auto","created_at":"2025-10-27 15:35:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1161316,"visible":true,"origin":"","legend":"","description":"","filename":"ScientificReportssubmissionv.1.2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7794007/v1/7ab504135b97bbe0c242a63b.pdf"},{"id":94471035,"identity":"7b82e5b1-0c21-48e5-9579-eb4ee84f396a","added_by":"auto","created_at":"2025-10-27 15:35:17","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5426,"visible":true,"origin":"","legend":"","description":"","filename":"71f7695255ea4edb8b545d6236639fd0.json","url":"https://assets-eu.researchsquare.com/files/rs-7794007/v1/0cb0646a83c7ee776fc1cf5d.json"},{"id":94470978,"identity":"1b48faf8-3d4f-4a1b-a15c-d5bd4a27fb83","added_by":"auto","created_at":"2025-10-27 15:35:08","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1191644,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7794007/v1/4f4019b3a4bca257ac8325ae.pdf"},{"id":98244041,"identity":"1b3d8cf4-b5d1-4062-bbaf-bc133f6edb62","added_by":"auto","created_at":"2025-12-15 16:12:37","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1373153,"visible":true,"origin":"","legend":"","description":"","filename":"ScientificReportssubmissionv.1.2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7794007/v1_covered_49c0ce38-db23-4bf6-aa1c-5ba756534e99.pdf"},{"id":94470989,"identity":"9aeb1466-7e2a-4421-b9cf-e0a0b4a0570b","added_by":"auto","created_at":"2025-10-27 15:35:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1191644,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7794007/v1/197e6bd2c0b65c7e1616f7e1.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Energy-Conscious LLM Text Generation : Impact of Decoding Strategies on GPU Energy Usage","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7794007/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7794007/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Decoding strategies significantly influence the quality and diversity of the generated text in Large Language Models (LLMs), yet their impact on computational resources, particularly GPU energy consumption, is insufficiently studied. This paper investigates the relationship between text generation decoding techniques and energy efficiency, focusing on the trade-off between generation quality and GPU energy usage across diverse tasks and decoding configurations. By benchmarking multiple strategies across various tasks, including Translation, Math Problem Solving, Coding, and Open-ended text generation, we reveal how selecting appropriate decoding techniques with their tuned hyperparameters affects text quality and has measurable implications for energy consumption. Our findings show that the choice of decoding strategy can greatly impact GPU energy usage, even when it has a minimal effect on output quality. Different strategies also involve trade-offs between quality and energy efficiency, and no single decoding method is best in all cases across every metric. To the best of our knowledge, this is one of the first studies to examine decoding strategies in LLMs from the perspective of energy consumption, providing useful insights for building energy-efficient applications without compromising text generation quality.","manuscriptTitle":"Energy-Conscious LLM Text Generation : Impact of Decoding Strategies on GPU Energy Usage","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-27 14:06:47","doi":"10.21203/rs.3.rs-7794007/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-27T07:37:28+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-21T02:48:08+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-21T00:41:53+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"185403630890177509941317475943318511043","date":"2025-10-15T16:28:16+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"273223333819652739679473862009557327648","date":"2025-10-14T20:17:57+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-10-13T19:39:21+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-10-10T19:08:00+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-10-09T08:49:50+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-10-09T08:49:43+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-10-06T19:38:47+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"56af37a0-a074-4ed8-8ee6-6317b4afea89","owner":[],"postedDate":"October 27th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":56796574,"name":"Physical sciences/Engineering"},{"id":56796575,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2025-12-15T16:05:36+00:00","versionOfRecord":{"articleIdentity":"rs-7794007","link":"https://doi.org/10.1038/s41598-025-31896-0","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-12-11 15:57:39","publishedOnDateReadable":"December 11th, 2025"},"versionCreatedAt":"2025-10-27 14:06:47","video":"","vorDoi":"10.1038/s41598-025-31896-0","vorDoiUrl":"https://doi.org/10.1038/s41598-025-31896-0","workflowStages":[]},"version":"v1","identity":"rs-7794007","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7794007","identity":"rs-7794007","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}