{"id":"https://openalex.org/W4416034208","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.957","title":"Distributed LLM Serving on Consumer-Grade GPUs by Reconciling Computation and Communication","display_name":"Distributed LLM Serving on Consumer-Grade GPUs by Reconciling Computation and Communication","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034208","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.957"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.957","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.957","pdf_url":"https://aclanthology.org/2025.findings-emnlp.957.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-emnlp.957.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Lewei Jin","orcid":null},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Lewei Jin","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030094911","display_name":"Kui Zhang","orcid":"https://orcid.org/0000-0002-1978-8753"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kui Zhang","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113780435","display_name":"Yuhao Chen","orcid":"https://orcid.org/0000-0001-7317-5897"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yongqi Chen","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120308517","display_name":"Zhuoyifan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhuoyifan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100442486","display_name":"Renjie Li","orcid":"https://orcid.org/0000-0002-3268-3956"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Renjie Li","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5007598993","display_name":"Yi Gao","orcid":"https://orcid.org/0000-0001-6343-4138"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yi Gao","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058494954","display_name":"Bowei Yang","orcid":"https://orcid.org/0000-0002-9349-3259"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bowei Yang","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101181699","display_name":"Zhengong Cai","orcid":null},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhengong Cai","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5071857387","display_name":"Wei Dong","orcid":"https://orcid.org/0000-0003-1119-8369"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wei Dong","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"17633","last_page":"17642"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10772","display_name":"Distributed systems and fault tolerance","score":0.07580000162124634,"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"}},"topics":[{"id":"https://openalex.org/T10772","display_name":"Distributed systems and fault tolerance","score":0.07580000162124634,"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"}},{"id":"https://openalex.org/T10080","display_name":"Energy Efficient Wireless Sensor Networks","score":0.06859999895095825,"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"}},{"id":"https://openalex.org/T12203","display_name":"Mobile Agent-Based Network Management","score":0.033900000154972076,"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/computation","display_name":"Computation","score":0.47929999232292175},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.29989999532699585},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.2526000142097473},{"id":"https://openalex.org/keywords/field","display_name":"Field (mathematics)","score":0.2493000030517578}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7117999792098999},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.47929999232292175},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.31709998846054077},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.3025999963283539},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.29989999532699585},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.2809000015258789},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.25870001316070557},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2526000142097473},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.2493000030517578},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.24560000002384186}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.957","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.957","pdf_url":"https://aclanthology.org/2025.findings-emnlp.957.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.957","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.957","pdf_url":"https://aclanthology.org/2025.findings-emnlp.957.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G4037956315","display_name":null,"funder_award_id":"62272407","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034208.pdf","grobid_xml":"https://content.openalex.org/works/W4416034208.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2,8],"are":[3,88],"reshaping":[4],"internet":[5],"services.Serving":[6],"these":[7],"is":[9],"often":[10],"costly,":[11],"as":[12,48],"it":[13],"requires":[14],"multiple":[15],"high-end":[16],"GPUs.Consumer-grade":[17],"GPUs":[18],"offer":[19],"cheaper":[20],"computational":[21],"power,":[22],"providing":[23],"an":[24,100],"opportunity":[25],"for":[26,92,152],"more":[27,76],"cost-efficient":[28],"LLM":[29,103,170],"serving.Prior":[30],"efforts":[31],"have":[32],"explored":[33],"distributed":[34,102,169],"serving":[35,104,171],"at":[36],"scale,":[37],"primarily":[38],"focusing":[39],"on":[40],"model":[41],"deployment":[42],"strategies.However,":[43],"communication":[44],"efficiency":[45],"has":[46],"emerged":[47],"a":[49,127,143,174],"challenge":[50],"due":[51],"to":[52,73,82,135,166,179],"the":[53,60,107,149,167],"imbalance":[54],"in":[55],"data":[56,77],"transfer":[57],"volumes":[58],"between":[59],"two":[61,124],"phases":[62],"of":[63,111,123,177],"inference:":[64],"prefill":[65,112,137,153],"and":[66,117,141,163],"decode.Prefill":[67],"requests":[68,84,113,154],"can":[69],"involve":[70],"transmitting":[71],"up":[72,178],"1000":[74],"times":[75],"than":[78],"decode":[79,83,93,139],"requests,":[80,140],"leading":[81],"being":[85],"delayed.Consequently,":[86],"servers":[87],"underutilized":[89],"while":[90],"waiting":[91],"requests.In":[94],"this":[95],"paper,":[96],"we":[97],"present":[98],"MoLink,":[99],"efficient":[101],"system.It":[105],"splits":[106],"prolonged":[108],"transmission":[109,128],"volume":[110,151],"into":[114],"smaller":[115],"chunks":[116],"carefully":[118],"scheduling":[119,129],"their":[120],"transmission.It":[121],"consists":[122],"parts:":[125],"(i)":[126],"algorithm":[130,146],"that":[131,147,158],"fairly":[132],"determines":[133,148],"whether":[134],"transmit":[136,150],"or":[138],"(ii)":[142],"chunking":[144],"determination":[145],"just-in-time.Our":[155],"evaluation":[156],"demonstrates":[157],"MoLink":[159],"reduces":[160],"TTFT,":[161],"TPOT,":[162],"latency":[164],"compared":[165],"state-ofthe-art":[168],"system,":[172],"with":[173],"maximum":[175],"reduction":[176],"46%.":[180]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-11-08T00:00:00"}