{"id":"https://openalex.org/W7154072029","doi":"https://doi.org/10.1145/3800690","title":"Enabling Low-Latency, GPU-Efficient Serverless Inference with Model Swapping","display_name":"Enabling Low-Latency, GPU-Efficient Serverless Inference with Model Swapping","publication_year":2026,"publication_date":"2026-04-13","ids":{"openalex":"https://openalex.org/W7154072029","doi":"https://doi.org/10.1145/3800690"},"language":"en","primary_location":{"id":"doi:10.1145/3800690","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3800690","pdf_url":null,"source":{"id":"https://openalex.org/S26056741","display_name":"ACM Transactions on Architecture and Code Optimization","issn_l":"1544-3566","issn":["1544-3566","1544-3973"],"is_oa":true,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Architecture and Code Optimization","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://doi.org/10.1145/3800690","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5051893788","display_name":"Minchen Yu","orcid":"https://orcid.org/0000-0002-6797-9028"},"institutions":[{"id":"https://openalex.org/I4210116924","display_name":"Chinese University of Hong Kong, Shenzhen","ror":"https://ror.org/02d5ks197","country_code":"CN","type":"education","lineage":["https://openalex.org/I177725633","https://openalex.org/I180726961","https://openalex.org/I4210116924"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Minchen Yu","raw_affiliation_strings":["The Chinese University of Hong Kong, Shenzhen"],"raw_orcid":"https://orcid.org/0000-0002-6797-9028","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Shenzhen","institution_ids":["https://openalex.org/I4210116924"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Ao Wang","orcid":"https://orcid.org/0009-0003-1018-0957"},"institutions":[{"id":"https://openalex.org/I4210095624","display_name":"Alibaba Group (United States)","ror":"https://ror.org/00rn0m335","country_code":"US","type":"company","lineage":["https://openalex.org/I4210095624","https://openalex.org/I45928872"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ao Wang","raw_affiliation_strings":["Alibaba Group"],"raw_orcid":"https://orcid.org/0009-0003-1018-0957","affiliations":[{"raw_affiliation_string":"Alibaba Group","institution_ids":["https://openalex.org/I4210095624"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133517969","display_name":"Bohui Wu","orcid":"https://orcid.org/0009-0006-5210-564X"},"institutions":[{"id":"https://openalex.org/I4210116924","display_name":"Chinese University of Hong Kong, Shenzhen","ror":"https://ror.org/02d5ks197","country_code":"CN","type":"education","lineage":["https://openalex.org/I177725633","https://openalex.org/I180726961","https://openalex.org/I4210116924"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bohui Wu","raw_affiliation_strings":["The Chinese University of Hong Kong, Shenzhen"],"raw_orcid":"https://orcid.org/0009-0006-5210-564X","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Shenzhen","institution_ids":["https://openalex.org/I4210116924"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yuxuan Liu","orcid":"https://orcid.org/0009-0007-6785-5053"},"institutions":[{"id":"https://openalex.org/I4210116924","display_name":"Chinese University of Hong Kong, Shenzhen","ror":"https://ror.org/02d5ks197","country_code":"CN","type":"education","lineage":["https://openalex.org/I177725633","https://openalex.org/I180726961","https://openalex.org/I4210116924"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuxuan Liu","raw_affiliation_strings":["The Chinese University of Hong Kong, Shenzhen"],"raw_orcid":"https://orcid.org/0009-0007-6785-5053","affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong, Shenzhen","institution_ids":["https://openalex.org/I4210116924"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Dong Chen","orcid":"https://orcid.org/0009-0005-0167-8432"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Dong Chen","raw_affiliation_strings":["Hong Kong University of Science and Technology"],"raw_orcid":"https://orcid.org/0009-0005-0167-8432","affiliations":[{"raw_affiliation_string":"Hong Kong University of Science and Technology","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Haoxuan Yu","orcid":"https://orcid.org/0009-0008-9677-7960"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Haoxuan Yu","raw_affiliation_strings":["Hong Kong University of Science and Technology"],"raw_orcid":"https://orcid.org/0009-0008-9677-7960","affiliations":[{"raw_affiliation_string":"Hong Kong University of Science and Technology","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5026164915","display_name":"W. WANG","orcid":null},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Wei Wang","raw_affiliation_strings":["Hong Kong University of Science and Technology"],"raw_orcid":"https://orcid.org/0000-0002-4585-4152","affiliations":[{"raw_affiliation_string":"Hong Kong University of Science and Technology","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125558839","display_name":"Ruichuan Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ruichuan Chen","raw_affiliation_strings":["Nokia Bell Labs"],"raw_orcid":"https://orcid.org/0009-0006-5060-8411","affiliations":[{"raw_affiliation_string":"Nokia Bell Labs","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112932558","display_name":"D.X. Nie","orcid":null},"institutions":[{"id":"https://openalex.org/I4210095624","display_name":"Alibaba Group (United States)","ror":"https://ror.org/00rn0m335","country_code":"US","type":"company","lineage":["https://openalex.org/I4210095624","https://openalex.org/I45928872"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dapeng Nie","raw_affiliation_strings":["Alibaba Group"],"raw_orcid":"https://orcid.org/0009-0003-1381-6834","affiliations":[{"raw_affiliation_string":"Alibaba Group","institution_ids":["https://openalex.org/I4210095624"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Haoran Yang","orcid":"https://orcid.org/0009-0003-8889-4132"},"institutions":[{"id":"https://openalex.org/I4210095624","display_name":"Alibaba Group (United States)","ror":"https://ror.org/00rn0m335","country_code":"US","type":"company","lineage":["https://openalex.org/I4210095624","https://openalex.org/I45928872"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Haoran Yang","raw_affiliation_strings":["Alibaba Group"],"raw_orcid":"https://orcid.org/0009-0003-8889-4132","affiliations":[{"raw_affiliation_string":"Alibaba Group","institution_ids":["https://openalex.org/I4210095624"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5133510318","display_name":"Yu Ding","orcid":"https://orcid.org/0009-0001-1179-9888"},"institutions":[{"id":"https://openalex.org/I4210095624","display_name":"Alibaba Group (United States)","ror":"https://ror.org/00rn0m335","country_code":"US","type":"company","lineage":["https://openalex.org/I4210095624","https://openalex.org/I45928872"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yu Ding","raw_affiliation_strings":["Alibaba Group"],"raw_orcid":"https://orcid.org/0009-0001-1179-9888","affiliations":[{"raw_affiliation_string":"Alibaba Group","institution_ids":["https://openalex.org/I4210095624"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.47900024,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"23","issue":"2","first_page":"1","last_page":"27"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10101","display_name":"Cloud Computing and Resource Management","score":0.7531999945640564,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10101","display_name":"Cloud Computing and Resource Management","score":0.7531999945640564,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.05660000070929527,"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/T14347","display_name":"Big Data and Digital Economy","score":0.0430000014603138,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6826000213623047},{"id":"https://openalex.org/keywords/cloud-computing","display_name":"Cloud computing","score":0.6255999803543091},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.6191999912261963},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.5867999792098999},{"id":"https://openalex.org/keywords/provisioning","display_name":"Provisioning","score":0.5569999814033508},{"id":"https://openalex.org/keywords/asynchronous-communication","display_name":"Asynchronous communication","score":0.5109000205993652},{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.49799999594688416}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.9075999855995178},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6826000213623047},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.6255999803543091},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.6191999912261963},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5867999792098999},{"id":"https://openalex.org/C172191483","wikidata":"https://www.wikidata.org/wiki/Q1071806","display_name":"Provisioning","level":2,"score":0.5569999814033508},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.5346999764442444},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.5109000205993652},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.49799999594688416},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.4878999888896942},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4327000081539154},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.39959999918937683},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.33719998598098755},{"id":"https://openalex.org/C34165917","wikidata":"https://www.wikidata.org/wiki/Q188267","display_name":"Programming paradigm","level":2,"score":0.31790000200271606},{"id":"https://openalex.org/C83283714","wikidata":"https://www.wikidata.org/wiki/Q121117","display_name":"Supercomputer","level":2,"score":0.2775999903678894},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.2736999988555908},{"id":"https://openalex.org/C7366592","wikidata":"https://www.wikidata.org/wiki/Q1255620","display_name":"Dram","level":2,"score":0.27140000462532043},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.2685000002384186},{"id":"https://openalex.org/C2777851325","wikidata":"https://www.wikidata.org/wiki/Q7094102","display_name":"Online model","level":2,"score":0.25949999690055847}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3800690","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3800690","pdf_url":null,"source":{"id":"https://openalex.org/S26056741","display_name":"ACM Transactions on Architecture and Code Optimization","issn_l":"1544-3566","issn":["1544-3566","1544-3973"],"is_oa":true,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Architecture and Code Optimization","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1145/3800690","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3800690","pdf_url":null,"source":{"id":"https://openalex.org/S26056741","display_name":"ACM Transactions on Architecture and Code Optimization","issn_l":"1544-3566","issn":["1544-3566","1544-3973"],"is_oa":true,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Architecture and Code Optimization","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":25,"referenced_works":["https://openalex.org/W1996027765","https://openalex.org/W2005574683","https://openalex.org/W2139875443","https://openalex.org/W2489529491","https://openalex.org/W2767239597","https://openalex.org/W2963341956","https://openalex.org/W2981449041","https://openalex.org/W2982157693","https://openalex.org/W3010640493","https://openalex.org/W3011293047","https://openalex.org/W3012479151","https://openalex.org/W3012514909","https://openalex.org/W3125961627","https://openalex.org/W3130689885","https://openalex.org/W3152809665","https://openalex.org/W3204216247","https://openalex.org/W4214764640","https://openalex.org/W4285504009","https://openalex.org/W4308399059","https://openalex.org/W4372263604","https://openalex.org/W4387321091","https://openalex.org/W4394944658","https://openalex.org/W4407218580","https://openalex.org/W4407218922","https://openalex.org/W4414735816"],"related_works":[],"abstract_inverted_index":{"Serverless":[0],"computing":[1],"offers":[2],"a":[3,34,46,133,151,172,177],"compelling":[4],"cloud":[5,181],"model":[6,72,86,99,140,167],"for":[7,18,37,120,194],"online":[8],"inference":[9,51,122,148],"services.":[10],"However,":[11],"existing":[12],"serverless":[13,35,180],"platforms":[14],"lack":[15],"efficient":[16,43,89],"support":[17],"GPUs,":[19,156],"hindering":[20],"their":[21],"ability":[22],"to":[23,93,114,162],"deliver":[24],"high-performance":[25],"inference.":[26,40],"In":[27],"this":[28],"article,":[29],"we":[30,102],"present":[31],"Torpor":[32,53,74,127,142,184],",":[33],"platform":[36],"GPU-efficient,":[38],"low-latency":[39],"To":[41],"enable":[42],"sharing":[44],"of":[45,147],"node\u2019s":[47],"GPUs":[48,64],"among":[49],"numerous":[50],"functions,":[52],"maintains":[54],"models":[55],"in":[56,132,176],"main":[57],"memory":[58,91],"and":[59,88,128,139,192,196],"dynamically":[60],"swaps":[61],"them":[62],"onto":[63],"upon":[65],"request":[66,106],"arrivals":[67],"(i.e.,":[68],"late":[69,137],"binding":[70,138],"with":[71,154],"swapping).":[73],"uses":[75],"various":[76],"techniques,":[77],"including":[78],"asynchronous":[79],"API":[80],"redirection,":[81],"GPU":[82,90,112,187],"runtime":[83],"sharing,":[84],"pipelined":[85],"execution,":[87,164],"management,":[92],"minimize":[94],"latency":[95,116,159],"overhead":[96],"caused":[97],"by":[98,190],"swapping.":[100],"Additionally,":[101],"design":[103],"an":[104],"interference-aware":[105],"scheduling":[107],"algorithm":[108],"that":[109,183],"utilizes":[110],"high-speed":[111],"interconnects":[113],"meet":[115],"service-level":[117],"objectives":[118],"(SLOs)":[119],"individual":[121],"functions.":[123],"We":[124],"have":[125],"implemented":[126],"evaluated":[129],"its":[130],"performance":[131,160],"production":[134],"environment.":[135],"Utilizing":[136],"swapping,":[141],"can":[143],"concurrently":[144],"serve":[145],"hundreds":[146],"functions":[149],"on":[150,171],"worker":[152],"node":[153],"4":[155],"while":[157],"achieving":[158],"comparable":[161],"native":[163],"where":[165],"each":[166],"is":[168],"cached":[169],"exclusively":[170],"GPU.":[173],"Pilot":[174],"deployment":[175],"leading":[178],"commercial":[179],"shows":[182],"reduces":[185],"the":[186,197],"provisioning":[188],"cost":[189],"70%":[191],"65%":[193],"users":[195],"platform,":[198],"respectively.":[199]},"counts_by_year":[],"updated_date":"2026-06-26T06:17:10.115597","created_date":"2026-04-14T00:00:00"}
