Stress-Testing USB Accelerators for Efficient Edge Inference

Stress-Testing USB Accelerators for Efficient Edge Inference | 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 Research Article Stress-Testing USB Accelerators for Efficient Edge Inference Raphael Fischer, Alexander van der Staay, Sebastian Buschjäger This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3793927/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Several manufacturers sell specialized USB devices for accelerating machine learning (ML) on the edge. While being generally promoted as a versatile solution for more efficient edge inference with deep learning models, extensive practical insights on their usability and performance are hard to find. In order to make ML deployment on the edge more sustainable, our work investigates how resource efficient these USB accelerators really are. For that, we first introduce a novel and theoretically sound methodology. It allows for comparing intricate model performance in terms of quality and resource consumption across different execution environments. We then put it into practice by studying the usability and efficiency of Google's Coral edge tensor processing unit (TPU) and Intel's neural compute stick 2 (NCS). In total, we benchmark over 30 models across nine hardware configurations, which reveals intricate trade-offs. Our work demonstrates that USB accelerators are indeed capable of reducing the energy consumption by a factor up to ten, however this improvement cannot be observed for all configurations - more than 50% of the investigated models cannot be run on accelerator hardware, and in several other cases, the power draw is only marginally improved. Our experiments show that the NCS improves efficiency in a more stable way, while the TPU shows further benefits in specific cases but performs less predictable. We hope that our paper provides valuable insights for practitioners that want to deploy ML on the edge in the most efficient and sustainable way. USB accelerators sustainability benchmarking edge learning resource-awareness Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-3793927","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":264782751,"identity":"75f697d7-9d21-45d0-947b-6e41470b1ebf","order_by":0,"name":"Raphael Fischer","email":"data:image/png;base64,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","orcid":"","institution":"TU Dortmund University","correspondingAuthor":true,"prefix":"","firstName":"Raphael","middleName":"","lastName":"Fischer","suffix":""},{"id":264782752,"identity":"13021987-2fec-46ff-8cbc-325179f70013","order_by":1,"name":"Alexander van der Staay","email":"","orcid":"","institution":"TU Dortmund University","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"van der","lastName":"Staay","suffix":""},{"id":264782753,"identity":"12f175e7-3667-428f-ba2c-b94dcff56d03","order_by":2,"name":"Sebastian Buschjäger","email":"","orcid":"","institution":"TU Dortmund University","correspondingAuthor":false,"prefix":"","firstName":"Sebastian","middleName":"","lastName":"Buschjäger","suffix":""}],"badges":[],"createdAt":"2023-12-22 21:44:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3793927/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3793927/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":51313354,"identity":"66f565c2-6ec4-4d38-a681-ca0d74af7d64","added_by":"auto","created_at":"2024-02-19 11:54:15","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1206214,"visible":true,"origin":"","legend":"","description":"","filename":"24gsai.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3793927/v1_covered_fe282c0a-e040-4ecc-a917-289bebaf8d58.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Stress-Testing USB Accelerators for Efficient Edge Inference","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"USB accelerators, sustainability, benchmarking, edge learning, resource-awareness","lastPublishedDoi":"10.21203/rs.3.rs-3793927/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3793927/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Several manufacturers sell specialized USB devices for accelerating machine learning (ML) on the edge. While being generally promoted as a versatile solution for more efficient edge inference with deep learning models, extensive practical insights on their usability and performance are hard to find. In order to make ML deployment on the edge more sustainable, our work investigates how resource efficient these USB accelerators really are. For that, we first introduce a novel and theoretically sound methodology. It allows for comparing intricate model performance in terms of quality and resource consumption across different execution environments. We then put it into practice by studying the usability and efficiency of Google's Coral edge tensor processing unit (TPU) and Intel's neural compute stick 2 (NCS). In total, we benchmark over 30 models across nine hardware configurations, which reveals intricate trade-offs. Our work demonstrates that USB accelerators are indeed capable of reducing the energy consumption by a factor up to ten, however this improvement cannot be observed for all configurations - more than 50% of the investigated models cannot be run on accelerator hardware, and in several other cases, the power draw is only marginally improved. Our experiments show that the NCS improves efficiency in a more stable way, while the TPU shows further benefits in specific cases but performs less predictable. We hope that our paper provides valuable insights for practitioners that want to deploy ML on the edge in the most efficient and sustainable way.","manuscriptTitle":"Stress-Testing USB Accelerators for Efficient Edge Inference","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-01-04 13:03:15","doi":"10.21203/rs.3.rs-3793927/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5c2f61f0-a7a7-47ad-a3a8-c62b754bb096","owner":[],"postedDate":"January 4th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-02-19T11:46:02+00:00","versionOfRecord":[],"versionCreatedAt":"2024-01-04 13:03:15","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3793927","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3793927","identity":"rs-3793927","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}