{"id":"https://openalex.org/W4402897415","doi":"https://doi.org/10.1109/iwqos61813.2024.10682900","title":"COS: Cross-Processor Operator Scheduling for Multi-Tenant Deep Learning Inference","display_name":"COS: Cross-Processor Operator Scheduling for Multi-Tenant Deep Learning Inference","publication_year":2024,"publication_date":"2024-06-19","ids":{"openalex":"https://openalex.org/W4402897415","doi":"https://doi.org/10.1109/iwqos61813.2024.10682900"},"language":"en","primary_location":{"id":"doi:10.1109/iwqos61813.2024.10682900","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/iwqos61813.2024.10682900","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 IEEE/ACM 32nd International Symposium on Quality of Service (IWQoS)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5065564945","display_name":"Changyao Lin","orcid":"https://orcid.org/0000-0001-6805-2649"},"institutions":[{"id":"https://openalex.org/I204983213","display_name":"Harbin Institute of Technology","ror":"https://ror.org/01yqg2h08","country_code":"CN","type":"education","lineage":["https://openalex.org/I204983213"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Changyao Lin","raw_affiliation_strings":["Harbin Institute of Technology,Harbin,China"],"affiliations":[{"raw_affiliation_string":"Harbin Institute of Technology,Harbin,China","institution_ids":["https://openalex.org/I204983213"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5115592259","display_name":"Jie Liu","orcid":"https://orcid.org/0009-0006-5054-4550"},"institutions":[{"id":"https://openalex.org/I204983213","display_name":"Harbin Institute of Technology","ror":"https://ror.org/01yqg2h08","country_code":"CN","type":"education","lineage":["https://openalex.org/I204983213"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jie Liu","raw_affiliation_strings":["Harbin Institute of Technology,Shenzhen,China"],"affiliations":[{"raw_affiliation_string":"Harbin Institute of Technology,Shenzhen,China","institution_ids":["https://openalex.org/I204983213"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5065564945"],"corresponding_institution_ids":["https://openalex.org/I204983213"],"apc_list":null,"apc_paid":null,"fwci":0.9829,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.7548137,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":96},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"10"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9355000257492065,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9355000257492065,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.929099977016449,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11801","display_name":"Reservoir Engineering and Simulation Methods","score":0.9068999886512756,"subfield":{"id":"https://openalex.org/subfields/2212","display_name":"Ocean Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7697274684906006},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.5914702415466309},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5405203104019165},{"id":"https://openalex.org/keywords/operator","display_name":"Operator (biology)","score":0.49846625328063965},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4811040759086609},{"id":"https://openalex.org/keywords/processor-scheduling","display_name":"Processor scheduling","score":0.4587538242340088},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4294770359992981},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.39241528511047363},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.24391934275627136},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.1324334740638733},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.09773552417755127}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7697274684906006},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5914702415466309},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5405203104019165},{"id":"https://openalex.org/C17020691","wikidata":"https://www.wikidata.org/wiki/Q139677","display_name":"Operator (biology)","level":5,"score":0.49846625328063965},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4811040759086609},{"id":"https://openalex.org/C2984822820","wikidata":"https://www.wikidata.org/wiki/Q1123036","display_name":"Processor scheduling","level":3,"score":0.4587538242340088},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4294770359992981},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.39241528511047363},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.24391934275627136},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.1324334740638733},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.09773552417755127},{"id":"https://openalex.org/C68387754","wikidata":"https://www.wikidata.org/wiki/Q7271585","display_name":"Schedule","level":2,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C86339819","wikidata":"https://www.wikidata.org/wiki/Q407384","display_name":"Transcription factor","level":3,"score":0.0},{"id":"https://openalex.org/C158448853","wikidata":"https://www.wikidata.org/wiki/Q425218","display_name":"Repressor","level":4,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/iwqos61813.2024.10682900","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/iwqos61813.2024.10682900","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 IEEE/ACM 32nd International Symposium on Quality of Service (IWQoS)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":21,"referenced_works":["https://openalex.org/W2183341477","https://openalex.org/W2194775991","https://openalex.org/W3011667710","https://openalex.org/W3014810041","https://openalex.org/W3212184265","https://openalex.org/W3212621679","https://openalex.org/W3215253865","https://openalex.org/W3217445637","https://openalex.org/W4283020086","https://openalex.org/W4308450151","https://openalex.org/W4320060064","https://openalex.org/W4372271902","https://openalex.org/W4394670654","https://openalex.org/W6637373629","https://openalex.org/W6687681856","https://openalex.org/W6738796088","https://openalex.org/W6751349269","https://openalex.org/W6752515464","https://openalex.org/W6766978945","https://openalex.org/W6785429063","https://openalex.org/W6809979497"],"related_works":["https://openalex.org/W4375867731","https://openalex.org/W2055243143","https://openalex.org/W2611989081","https://openalex.org/W2128410848","https://openalex.org/W2102390841","https://openalex.org/W2118368532","https://openalex.org/W2126232624","https://openalex.org/W2434525066","https://openalex.org/W3140149227","https://openalex.org/W2130555437"],"abstract_inverted_index":{"Multi-tenant":[0],"inference,":[1],"as":[2,33],"a":[3,97,124],"prevalent":[4],"inference":[5,21,55],"paradigm":[6],"nowadays,":[7],"requires":[8,86],"deploying":[9],"multiple":[10],"deep":[11],"learning":[12],"models":[13],"on":[14,149],"the":[15,47,57,92,109,116,132,176],"hardware":[16,152],"platform":[17],"to":[18,74,128],"concurrently":[19],"process":[20],"tasks.":[22],"Modern":[23],"platforms":[24,153],"are":[25],"typically":[26],"equipped":[27],"with":[28],"various":[29,150],"heterogeneous":[30,151],"processors,":[31],"such":[32],"CPU-GPU":[34],"platform.":[35],"To":[36],"reduce":[37],"resource":[38,69],"contention":[39],"and":[40,59,77,89,112,130,135,145,154,163,168,186],"improve":[41],"Quality":[42],"of":[43,91],"Service":[44],"(QoS)":[45],"in":[46,175],"multi-tenant":[48,177],"scenario,":[49],"existing":[50],"work":[51,84],"has":[52],"studied":[53],"cross-processor":[54,100],"at":[56],"model-":[58],"layer-level.":[60],"However,":[61],"coarse-grained":[62],"scheduling":[63,101,133],"cannot":[64],"flexibly":[65],"account":[66],"for":[67,115],"subtle":[68],"fluctuations,":[70],"which":[71,104],"may":[72],"lead":[73],"task":[75],"blockages":[76],"incur":[78],"significant":[79],"processor":[80],"switching":[81,113],"overheads.":[82],"Such":[83],"usually":[85],"extensive":[87],"modification":[88],"retraining":[90],"models.":[93],"Therefore,":[94],"we":[95,156],"propose":[96,136],"finer-grained":[98],"operator-level":[99],"framework":[102],"COS,":[103],"can":[105],"more":[106,161],"precisely":[107],"divide":[108],"computational":[110],"workloads":[111],"overheads":[114],"tenants,":[117],"without":[118],"modifying":[119],"or":[120],"retraining.":[121],"We":[122],"introduce":[123],"novel":[125],"intermediate":[126],"representation":[127],"abstract":[129],"simplify":[131],"problem,":[134],"an":[137,182],"efficient":[138],"two-phase":[139],"search":[140],"algorithm.":[141],"COS":[142,159,180],"is":[143,160,181,189],"automated":[144],"easy-to-scale,":[146],"through":[147],"experiments":[148],"models,":[155],"demonstrate":[157],"that":[158],"flexible":[162],"effective":[164],"than":[165,172],"layer-level":[166],"scheduling,":[167],"achieves":[169],"higher":[170],"throughput":[171],"single-processor":[173],"processing":[174],"scenario.":[178],"Furthermore,":[179],"offline":[183],"optimization":[184],"method,":[185],"its":[187],"overhead":[188],"highly":[190],"acceptable.":[191]},"counts_by_year":[{"year":2025,"cited_by_count":2}],"updated_date":"2026-03-25T13:04:00.132906","created_date":"2025-10-10T00:00:00"}