A Fully Hardware-Managed Scheduling Architecture for AI Accelerators

A Fully Hardware-Managed Scheduling Architecture for AI Accelerators | 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 A Fully Hardware-Managed Scheduling Architecture for AI Accelerators Libo Cheng, Liang Yang, Jian Shao, Xinyi Gu, Rong Qian, Xinwei Zhang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8183074/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 14 You are reading this latest preprint version Abstract The proliferation of AI applications across diverse domains has driven the evolution of AI accelerators toward higher performance and energy efficiency. This paper addresses the critical challenge of task scheduling in AI accelerators by introducing a fully hardware-managed scheduling system. Our approach leverages Operator Completion Status Registers (OCSRs) and a novel computation-scheduling instruction set to minimize software overhead and maximize execution parallelism. The co-designed hardware-software solution comprises: (1) a dedicated hardware scheduling unit with a complete instruction pipeline, (2) a compiler that maps operators to scheduling instructions while managing OCSR allocation, and (3) a lightweight runtime for efficient task dispatch. Experimental results demonstrate that our system significantly reduces scheduling latency and improves overall throughput, achieving an average performance gain of approximately 30% across multiple CNN models while maintaining minimal area overhead of only 7.41%. The proposed architecture establishes a new paradigm for high-efficiency AI accelerator design. AI Accelerators Hardware-Managed Scheduling OCSR Computation-Scheduling Instructions Scheduling Latency Throughput Optimization Neural Network Inference Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 24 Dec, 2025 Reviews received at journal 17 Dec, 2025 Reviews received at journal 16 Dec, 2025 Reviews received at journal 10 Dec, 2025 Reviewers agreed at journal 09 Dec, 2025 Reviewers agreed at journal 09 Dec, 2025 Reviews received at journal 08 Dec, 2025 Reviewers agreed at journal 08 Dec, 2025 Reviewers agreed at journal 08 Dec, 2025 Reviewers agreed at journal 07 Dec, 2025 Reviewers invited by journal 06 Dec, 2025 Editor assigned by journal 26 Nov, 2025 Submission checks completed at journal 26 Nov, 2025 First submitted to journal 22 Nov, 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-8183074","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":558986119,"identity":"81767c79-9a18-45f8-ad71-1a86f7a6e5f5","order_by":0,"name":"Libo Cheng","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzElEQVRIiWNgGAWjYDACCRBRcICHn72x8eEH4rUYHJCT7DncbCxBihZjgxvpbQI8xOjgn9187OEXgzuJG24+bAPqt5PTbSBkyZ1j6cYyBs8SZ95ObHtQwJBsbHaAgBYDiRwzaQmDw4l9txPbDSQYDiRuI6wl/xtYS8PNg20SPMRpyWGT/GBw2FjgBiORWiRupJlJMxg8AwZyIjCQDYjwC/+M5GeSPyruAKPy+MOHHyrs5AhqAQFmRHQYEKEcBBh/EKlwFIyCUTAKRigAAHlSReoD1OjxAAAAAElFTkSuQmCC","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Libo","middleName":"","lastName":"Cheng","suffix":""},{"id":558986120,"identity":"498fc6c1-326d-4500-be99-91ffacbb4275","order_by":1,"name":"Liang Yang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Liang","middleName":"","lastName":"Yang","suffix":""},{"id":558986124,"identity":"b6a1b897-babb-4301-803d-3679751cc6e7","order_by":2,"name":"Jian Shao","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Jian","middleName":"","lastName":"Shao","suffix":""},{"id":558986128,"identity":"bcd97dd4-b2ec-4599-a46d-d18fe0659884","order_by":3,"name":"Xinyi Gu","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Xinyi","middleName":"","lastName":"Gu","suffix":""},{"id":558986132,"identity":"14b29beb-d899-400e-995f-5397bef7be1f","order_by":4,"name":"Rong Qian","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Rong","middleName":"","lastName":"Qian","suffix":""},{"id":558986134,"identity":"a2b2f633-b2ea-4cef-b546-a87496c0ae42","order_by":5,"name":"Xinwei Zhang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Xinwei","middleName":"","lastName":"Zhang","suffix":""},{"id":558986141,"identity":"13849e06-e309-4d1b-be81-6085dd6e4357","order_by":6,"name":"Donghao Li","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Donghao","middleName":"","lastName":"Li","suffix":""},{"id":558986142,"identity":"9c108314-c87f-4ddf-8159-74a2a3f91238","order_by":7,"name":"Hongbin Wang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Hongbin","middleName":"","lastName":"Wang","suffix":""},{"id":558986145,"identity":"3cf1d9f6-616d-4a22-a506-093718e0bcbb","order_by":8,"name":"Xiaoqi Xia","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Xiaoqi","middleName":"","lastName":"Xia","suffix":""}],"badges":[],"createdAt":"2025-11-23 02:38:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8183074/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8183074/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98076956,"identity":"176e9a16-0ab7-48cf-95b0-aa095d8d92d8","added_by":"auto","created_at":"2025-12-12 13:55:07","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9785,"visible":true,"origin":"","legend":"","description":"","filename":"d5ee0918cc994a79911d5af275a6c145.json","url":"https://assets-eu.researchsquare.com/files/rs-8183074/v1/21fdfce3f7b0dc72e95fd59a.json"},{"id":98429536,"identity":"192bf193-cb2f-4bc9-92eb-cc0c861afa09","added_by":"auto","created_at":"2025-12-17 16:43:39","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":796329,"visible":true,"origin":"","legend":"","description":"","filename":"fullyhardwarescheduler.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8183074/v1_covered_27385c6c-e60c-4985-a161-78ab2fa4c2ad.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Fully Hardware-Managed Scheduling Architecture for AI Accelerators","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":"journal-of-king-saud-university-computer-and-information-sciences","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Journal of King Saud University Computer and Information Sciences](https://link.springer.com/journal/44443)","snPcode":"44443","submissionUrl":"https://submission.springernature.com/new-submission/44443/3","title":"Journal of King Saud University Computer and Information Sciences","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Open","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"AI Accelerators, Hardware-Managed Scheduling, OCSR, Computation-Scheduling Instructions, Scheduling Latency, Throughput Optimization, Neural Network Inference","lastPublishedDoi":"10.21203/rs.3.rs-8183074/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8183074/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The proliferation of AI applications across diverse domains has driven the evolution of AI accelerators toward higher performance and energy efficiency. This paper addresses the critical challenge of task scheduling in AI accelerators by introducing a fully hardware-managed scheduling system. Our approach leverages Operator Completion Status Registers (OCSRs) and a novel computation-scheduling instruction set to minimize software overhead and maximize execution parallelism. The co-designed hardware-software solution comprises: (1) a dedicated hardware scheduling unit with a complete instruction pipeline, (2) a compiler that maps operators to scheduling instructions while managing OCSR allocation, and (3) a lightweight runtime for efficient task dispatch. Experimental results demonstrate that our system significantly reduces scheduling latency and improves overall throughput, achieving an average performance gain of approximately 30\\% across multiple CNN models while maintaining minimal area overhead of only 7.41\\%. The proposed architecture establishes a new paradigm for high-efficiency AI accelerator design.","manuscriptTitle":"A Fully Hardware-Managed Scheduling Architecture for AI Accelerators","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-12 13:55:01","doi":"10.21203/rs.3.rs-8183074/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-12-24T08:20:49+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-17T16:47:37+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-16T13:46:43+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-10T17:18:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"339231169766349682120880800379517228101","date":"2025-12-09T14:28:53+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"25671349077734066113640006581609497661","date":"2025-12-09T13:07:15+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-08T16:18:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"241368006284868862288760937892354866097","date":"2025-12-08T15:25:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"321708404846242609260508435796056000331","date":"2025-12-08T09:58:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"109866593059762874497687393262326754457","date":"2025-12-07T07:08:41+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-12-06T16:32:37+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-11-26T09:40:33+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-26T09:39:43+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of King Saud University Computer and Information Sciences","date":"2025-11-23T02:34:25+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-king-saud-university-computer-and-information-sciences","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Journal of King Saud University Computer and Information Sciences](https://link.springer.com/journal/44443)","snPcode":"44443","submissionUrl":"https://submission.springernature.com/new-submission/44443/3","title":"Journal of King Saud University Computer and Information Sciences","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Open","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"65d01efd-ae46-4f0f-b3f4-97efca74ed7f","owner":[],"postedDate":"December 12th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-01-22T14:13:12+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-12 13:55:01","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8183074","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8183074","identity":"rs-8183074","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}