{"id":"https://openalex.org/W4415748216","doi":"https://doi.org/10.1109/tpds.2025.3626974","title":"Efficient KV Cache Spillover Management on Memory-Constrained GPU for LLM Inference","display_name":"Efficient KV Cache Spillover Management on Memory-Constrained GPU for LLM Inference","publication_year":2025,"publication_date":"2025-10-31","ids":{"openalex":"https://openalex.org/W4415748216","doi":"https://doi.org/10.1109/tpds.2025.3626974"},"language":null,"primary_location":{"id":"doi:10.1109/tpds.2025.3626974","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2025.3626974","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Parallel and Distributed Systems","raw_type":"journal-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/A5048148542","display_name":"Jiazhi Jiang","orcid":"https://orcid.org/0000-0002-1417-3012"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]},{"id":"https://openalex.org/I25254941","display_name":"Beijing Normal University","ror":"https://ror.org/022k4wk35","country_code":"CN","type":"education","lineage":["https://openalex.org/I25254941"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiazhi Jiang","raw_affiliation_strings":["Beijing Normal University, Zhuhai, China","Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0002-1417-3012","affiliations":[{"raw_affiliation_string":"Beijing Normal University, Zhuhai, China","institution_ids":["https://openalex.org/I25254941"]},{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100394256","display_name":"Yao Chen","orcid":"https://orcid.org/0000-0002-5798-2282"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]},{"id":"https://openalex.org/I4210151723","display_name":"Institute of Refrigeration","ror":"https://ror.org/046jmy298","country_code":"GB","type":"nonprofit","lineage":["https://openalex.org/I4210151723"]}],"countries":["GB","SG"],"is_corresponding":false,"raw_author_name":"Yao Chen","raw_affiliation_strings":["4Paradigm Inc, Singapore","National University of Singapore, Singapore"],"raw_orcid":"https://orcid.org/0000-0002-5798-2282","affiliations":[{"raw_affiliation_string":"4Paradigm Inc, Singapore","institution_ids":["https://openalex.org/I4210151723"]},{"raw_affiliation_string":"National University of Singapore, Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030460543","display_name":"Zining Zhang","orcid":"https://orcid.org/0000-0002-7992-426X"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Zining Zhang","raw_affiliation_strings":["National University of Singapore, Singapore"],"raw_orcid":"https://orcid.org/0000-0002-7992-426X","affiliations":[{"raw_affiliation_string":"National University of Singapore, Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039946576","display_name":"Bingsheng He","orcid":"https://orcid.org/0000-0001-8618-4581"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Bingsheng He","raw_affiliation_strings":["National University of Singapore, Singapore"],"raw_orcid":"https://orcid.org/0000-0001-8618-4581","affiliations":[{"raw_affiliation_string":"National University of Singapore, Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5057157090","display_name":"Ping Luo","orcid":"https://orcid.org/0000-0001-6171-3811"},"institutions":[{"id":"https://openalex.org/I4210151723","display_name":"Institute of Refrigeration","ror":"https://ror.org/046jmy298","country_code":"GB","type":"nonprofit","lineage":["https://openalex.org/I4210151723"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Pingyi Luo","raw_affiliation_strings":["4Paradigm Inc, Singapore","4Paradigm Inc., China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"4Paradigm Inc, Singapore","institution_ids":["https://openalex.org/I4210151723"]},{"raw_affiliation_string":"4Paradigm Inc., China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103114808","display_name":"Mian Lu","orcid":"https://orcid.org/0009-0002-4853-8881"},"institutions":[{"id":"https://openalex.org/I4210151723","display_name":"Institute of Refrigeration","ror":"https://ror.org/046jmy298","country_code":"GB","type":"nonprofit","lineage":["https://openalex.org/I4210151723"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Mian Lu","raw_affiliation_strings":["4Paradigm Inc, Singapore","4Paradigm Inc., China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"4Paradigm Inc, Singapore","institution_ids":["https://openalex.org/I4210151723"]},{"raw_affiliation_string":"4Paradigm Inc., China","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yuqiang Chen","orcid":"https://orcid.org/0009-0005-9339-2510"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]},{"id":"https://openalex.org/I4210151723","display_name":"Institute of Refrigeration","ror":"https://ror.org/046jmy298","country_code":"GB","type":"nonprofit","lineage":["https://openalex.org/I4210151723"]}],"countries":["GB","SG"],"is_corresponding":false,"raw_author_name":"Yuqiang Chen","raw_affiliation_strings":["4Paradigm Inc, Singapore","National University of Singapore, Singapore","4Paradigm Inc., China"],"raw_orcid":"https://orcid.org/0009-0005-9339-2510","affiliations":[{"raw_affiliation_string":"4Paradigm Inc, Singapore","institution_ids":["https://openalex.org/I4210151723"]},{"raw_affiliation_string":"National University of Singapore, Singapore","institution_ids":["https://openalex.org/I165932596"]},{"raw_affiliation_string":"4Paradigm Inc., China","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"Hongbing Zhang","orcid":"https://orcid.org/0009-0007-9145-3919"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hongbing Zhang","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0009-0007-9145-3919","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024085713","display_name":"Jiangsu Du","orcid":"https://orcid.org/0000-0003-4707-9492"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiangsu Du","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0003-4707-9492","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5041534890","display_name":"Dan Huang","orcid":"https://orcid.org/0000-0001-5582-1031"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Dan Huang","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-5582-1031","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101633465","display_name":"Yutong Lu","orcid":"https://orcid.org/0000-0001-5315-3375"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yutong Lu","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-5315-3375","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":11,"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":0.30584647,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"37","issue":"1","first_page":"90","last_page":"105"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9330000281333923,"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":0.9330000281333923,"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/T10904","display_name":"Embedded Systems Design Techniques","score":0.011300000362098217,"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/T11181","display_name":"Advanced Data Storage Technologies","score":0.010400000028312206,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/cache","display_name":"Cache","score":0.8439000248908997},{"id":"https://openalex.org/keywords/cache-algorithms","display_name":"Cache algorithms","score":0.6068000197410583},{"id":"https://openalex.org/keywords/cache-coloring","display_name":"Cache coloring","score":0.5310999751091003},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.49549999833106995},{"id":"https://openalex.org/keywords/cache-pollution","display_name":"Cache pollution","score":0.489300012588501},{"id":"https://openalex.org/keywords/smart-cache","display_name":"Smart Cache","score":0.4350999891757965},{"id":"https://openalex.org/keywords/swap","display_name":"Swap (finance)","score":0.4138999879360199},{"id":"https://openalex.org/keywords/cache-oblivious-algorithm","display_name":"Cache-oblivious algorithm","score":0.3921000063419342}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8899000287055969},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.8439000248908997},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.6638000011444092},{"id":"https://openalex.org/C38556500","wikidata":"https://www.wikidata.org/wiki/Q13404475","display_name":"Cache algorithms","level":4,"score":0.6068000197410583},{"id":"https://openalex.org/C201148951","wikidata":"https://www.wikidata.org/wiki/Q5015976","display_name":"Cache coloring","level":4,"score":0.5310999751091003},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.49549999833106995},{"id":"https://openalex.org/C113166858","wikidata":"https://www.wikidata.org/wiki/Q5015981","display_name":"Cache pollution","level":5,"score":0.489300012588501},{"id":"https://openalex.org/C167713795","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"Smart Cache","level":5,"score":0.4350999891757965},{"id":"https://openalex.org/C99821215","wikidata":"https://www.wikidata.org/wiki/Q1136583","display_name":"Swap (finance)","level":2,"score":0.4138999879360199},{"id":"https://openalex.org/C59687516","wikidata":"https://www.wikidata.org/wiki/Q5015938","display_name":"Cache-oblivious algorithm","level":5,"score":0.3921000063419342},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.35040000081062317},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.3264999985694885},{"id":"https://openalex.org/C36340418","wikidata":"https://www.wikidata.org/wiki/Q7124288","display_name":"Page cache","level":5,"score":0.3237000107765198},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.31529998779296875},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.311599999666214},{"id":"https://openalex.org/C25536678","wikidata":"https://www.wikidata.org/wiki/Q5015977","display_name":"Cache invalidation","level":5,"score":0.28760001063346863},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.2639999985694885},{"id":"https://openalex.org/C195053848","wikidata":"https://www.wikidata.org/wiki/Q7894141","display_name":"Universal memory","level":5,"score":0.2612000107765198},{"id":"https://openalex.org/C49154492","wikidata":"https://www.wikidata.org/wiki/Q5300","display_name":"Central processing unit","level":2,"score":0.25429999828338623},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.2531000077724457},{"id":"https://openalex.org/C2776221188","wikidata":"https://www.wikidata.org/wiki/Q21072556","display_name":"Design space exploration","level":2,"score":0.25119999051094055}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tpds.2025.3626974","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2025.3626974","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Parallel and Distributed Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":18,"referenced_works":["https://openalex.org/W2002555321","https://openalex.org/W2489529491","https://openalex.org/W3012479151","https://openalex.org/W4285356434","https://openalex.org/W4287199907","https://openalex.org/W4289533987","https://openalex.org/W4289646389","https://openalex.org/W4310282800","https://openalex.org/W4321636575","https://openalex.org/W4364322902","https://openalex.org/W4366957282","https://openalex.org/W4378697133","https://openalex.org/W4385245566","https://openalex.org/W4387321091","https://openalex.org/W4389576338","https://openalex.org/W4401211704","https://openalex.org/W4404401018","https://openalex.org/W4409963643"],"related_works":[],"abstract_inverted_index":{"The":[0,209],"rapid":[1],"growth":[2],"of":[3,43,75,98,111,144,154,233],"model":[4,57,148],"parameters":[5],"presents":[6],"a":[7,121,145,159,182,193,220],"significant":[8],"challenge":[9],"when":[10,204],"deploying":[11],"large":[12],"generative":[13],"models":[14],"on":[15,49,132],"GPU.":[16],"Existing":[17],"LLM":[18,77,82,130],"runtime":[19,79],"memory":[20,53,64,84],"management":[21,85,125],"solutions":[22,86],"tend":[23],"to":[24,28,37,66,108,128,164,171,185,197,222],"maximize":[25],"batch":[26],"size":[27],"saturate":[29],"GPU":[30,63,99,134],"device":[31],"utilization.":[32],"Nevertheless,":[33],"this":[34,116],"practice":[35],"leads":[36],"situations":[38],"where":[39],"the":[40,56,76,151,167,199,230,234],"KV":[41,68,123,138,160,174,205],"Cache":[42,69,124,139,175,206],"certain":[44],"sequences":[45],"cannot":[46],"be":[47],"accommodated":[48],"GPUs":[50],"with":[51],"limited":[52],"capacity":[54],"during":[55],"inference,":[58],"requiring":[59],"temporary":[60],"eviction":[61],"from":[62],"(referred":[65],"as":[67],"spillover).":[70],"However,":[71],"without":[72],"careful":[73],"consideration":[74],"inference's":[78],"pattern,":[80],"current":[81],"inference":[83,131,231],"face":[87],"issues":[88],"like":[89],"one-size-fits-all":[90],"spillover":[91,146,155,207,235],"handling":[92,137,156],"approach":[93],"for":[94,179],"different":[95],"platforms,":[96],"under-utilization":[97],"in":[100,225],"prefill":[101],"stage,":[102],"and":[103,192],"suboptimal":[104],"sequence":[105,201],"selection":[106,202],"due":[107],"direct":[109],"employment":[110],"swap":[112,162,169],"or":[113],"recomputation.":[114],"In":[115],"paper,":[117],"we":[118],"introduce":[119],"FuseSpill,":[120],"holistic":[122],"solution":[126],"designed":[127],"boost":[129],"memory-constrained":[133],"by":[135],"efficiently":[136],"spillover.":[140],"Specifically,":[141],"FuseSpill":[142],"consists":[143],"cost":[147,153],"that":[149,213],"analyzes":[150],"system":[152],"techniques":[157],"quantitatively,":[158],"cache":[161],"orchestrator":[163],"further":[165],"refine":[166],"basic":[168],"technique":[170],"sophisticated":[172],"disaggregate":[173],"across":[176,190],"heterogeneous":[177],"devices":[178],"decoding":[180],"iterations,":[181],"multi-executor":[183],"scheduler":[184],"effectively":[186],"coordinate":[187],"task":[188],"executors":[189],"devices,":[191],"response":[194],"length":[195],"predictor":[196],"exploit":[198],"length-aware":[200],"strategy":[203],"occurs.":[208],"experimental":[210],"results":[211],"demonstrate":[212],"our":[214],"implementation":[215],"outperforms":[216],"existing":[217],"solutions,":[218],"delivering":[219],"20%":[221],"40%":[223],"increase":[224],"throughput":[226],"while":[227],"simultaneously":[228],"reducing":[229],"latency":[232],"sequences.":[236]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-31T00:00:00"}
