{"id":"https://openalex.org/W4413785379","doi":"https://doi.org/10.1109/tsc.2025.3603576","title":"Serving LLM in Distributed GPU Cluster with Fine-Grain Pipeline Constraints","display_name":"Serving LLM in Distributed GPU Cluster with Fine-Grain Pipeline Constraints","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4413785379","doi":"https://doi.org/10.1109/tsc.2025.3603576"},"language":"en","primary_location":{"id":"doi:10.1109/tsc.2025.3603576","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tsc.2025.3603576","pdf_url":null,"source":{"id":"https://openalex.org/S204223317","display_name":"IEEE Transactions on Services Computing","issn_l":"1939-1374","issn":["1939-1374","2372-0204"],"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 Services Computing","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/A5033311175","display_name":"Yanying Lin","orcid":"https://orcid.org/0000-0002-4809-9543"},"institutions":[{"id":"https://openalex.org/I4210145761","display_name":"Shenzhen Institutes of Advanced Technology","ror":"https://ror.org/04gh4er46","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210145761"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Yanying Lin","raw_affiliation_strings":["Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences"],"affiliations":[{"raw_affiliation_string":"Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences","institution_ids":["https://openalex.org/I4210145761"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102014822","display_name":"Shijie Peng","orcid":"https://orcid.org/0009-0005-1776-4773"},"institutions":[{"id":"https://openalex.org/I4210145761","display_name":"Shenzhen Institutes of Advanced Technology","ror":"https://ror.org/04gh4er46","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210145761"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shijie Peng","raw_affiliation_strings":["Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences"],"affiliations":[{"raw_affiliation_string":"Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences","institution_ids":["https://openalex.org/I4210145761"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040227806","display_name":"Shuaipeng Wu","orcid":"https://orcid.org/0009-0002-4764-2601"},"institutions":[{"id":"https://openalex.org/I4210145761","display_name":"Shenzhen Institutes of Advanced Technology","ror":"https://ror.org/04gh4er46","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210145761"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shuaipeng Wu","raw_affiliation_strings":["Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences"],"affiliations":[{"raw_affiliation_string":"Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences","institution_ids":["https://openalex.org/I4210145761"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100652703","display_name":"Yanbo Li","orcid":"https://orcid.org/0000-0002-7324-7712"},"institutions":[{"id":"https://openalex.org/I4210145761","display_name":"Shenzhen Institutes of Advanced Technology","ror":"https://ror.org/04gh4er46","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210145761"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yanbo Li","raw_affiliation_strings":["Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences"],"affiliations":[{"raw_affiliation_string":"Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences","institution_ids":["https://openalex.org/I4210145761"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103176699","display_name":"Chengzhi Lu","orcid":"https://orcid.org/0000-0002-6421-3828"},"institutions":[{"id":"https://openalex.org/I4210145761","display_name":"Shenzhen Institutes of Advanced Technology","ror":"https://ror.org/04gh4er46","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210145761"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chengzhi Lu","raw_affiliation_strings":["Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences"],"affiliations":[{"raw_affiliation_string":"Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences","institution_ids":["https://openalex.org/I4210145761"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5012757772","display_name":"Kejiang Ye","orcid":"https://orcid.org/0000-0001-6133-407X"},"institutions":[{"id":"https://openalex.org/I4210145761","display_name":"Shenzhen Institutes of Advanced Technology","ror":"https://ror.org/04gh4er46","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210145761"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kejiang Ye","raw_affiliation_strings":["Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences"],"affiliations":[{"raw_affiliation_string":"Shenzhen Institute of Advanced TechnologyChinese Academy of Sciences","institution_ids":["https://openalex.org/I4210145761"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5012773300","display_name":"Chengzhong Xu","orcid":"https://orcid.org/0000-0001-9480-0356"},"institutions":[{"id":"https://openalex.org/I204512498","display_name":"University of Macau","ror":"https://ror.org/01r4q9n85","country_code":"MO","type":"education","lineage":["https://openalex.org/I204512498"]}],"countries":["MO"],"is_corresponding":false,"raw_author_name":"Chengzhong Xu","raw_affiliation_strings":["University of Macau"],"affiliations":[{"raw_affiliation_string":"University of Macau","institution_ids":["https://openalex.org/I204512498"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":7,"corresponding_author_ids":["https://openalex.org/A5033311175"],"corresponding_institution_ids":["https://openalex.org/I4210145761"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.2154607,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"14"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11451","display_name":"Advanced Machining and Optimization Techniques","score":0.954200029373169,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11451","display_name":"Advanced Machining and Optimization Techniques","score":0.954200029373169,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T13341","display_name":"Belt Conveyor Systems Engineering","score":0.9225999712944031,"subfield":{"id":"https://openalex.org/subfields/2210","display_name":"Mechanical 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":0.8405708074569702},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.6507022380828857},{"id":"https://openalex.org/keywords/cluster","display_name":"Cluster (spacecraft)","score":0.6052830815315247},{"id":"https://openalex.org/keywords/gpu-cluster","display_name":"GPU cluster","score":0.6030233502388},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.5771597623825073},{"id":"https://openalex.org/keywords/computational-science","display_name":"Computational science","score":0.49816083908081055},{"id":"https://openalex.org/keywords/general-purpose-computing-on-graphics-processing-units","display_name":"General-purpose computing on graphics processing units","score":0.4916256070137024},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.40370461344718933},{"id":"https://openalex.org/keywords/computer-graphics","display_name":"Computer graphics (images)","score":0.36473381519317627},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.3339385688304901},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.21072223782539368},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.17730796337127686},{"id":"https://openalex.org/keywords/graphics","display_name":"Graphics","score":0.16257673501968384}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8405708074569702},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6507022380828857},{"id":"https://openalex.org/C164866538","wikidata":"https://www.wikidata.org/wiki/Q367351","display_name":"Cluster (spacecraft)","level":2,"score":0.6052830815315247},{"id":"https://openalex.org/C2781335571","wikidata":"https://www.wikidata.org/wiki/Q2633544","display_name":"GPU cluster","level":3,"score":0.6030233502388},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5771597623825073},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.49816083908081055},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.4916256070137024},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.40370461344718933},{"id":"https://openalex.org/C121684516","wikidata":"https://www.wikidata.org/wiki/Q7600677","display_name":"Computer graphics (images)","level":1,"score":0.36473381519317627},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.3339385688304901},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.21072223782539368},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.17730796337127686},{"id":"https://openalex.org/C21442007","wikidata":"https://www.wikidata.org/wiki/Q1027879","display_name":"Graphics","level":2,"score":0.16257673501968384}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tsc.2025.3603576","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tsc.2025.3603576","pdf_url":null,"source":{"id":"https://openalex.org/S204223317","display_name":"IEEE Transactions on Services Computing","issn_l":"1939-1374","issn":["1939-1374","2372-0204"],"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 Services Computing","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":0,"referenced_works":[],"related_works":["https://openalex.org/W1963859303","https://openalex.org/W2364044215","https://openalex.org/W2389600408","https://openalex.org/W240129890","https://openalex.org/W3048701459","https://openalex.org/W2149078538","https://openalex.org/W2370314112","https://openalex.org/W1912958759","https://openalex.org/W2792081825","https://openalex.org/W2893308117"],"abstract_inverted_index":{"As":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"continue":[5],"to":[6,140],"advance,":[7],"their":[8],"parameter":[9],"sizes":[10],"are":[11],"growing":[12],"exponentially-far":[13],"outpacing":[14],"hardware":[15],"capabilities.":[16],"This":[17],"widening":[18],"gap":[19],"necessitates":[20],"distributed":[21,61],"computing":[22],"through":[23,70,98],"pipeline":[24,36,106,154],"parallelism":[25],"for":[26,60],"efficient":[27],"inference.":[28,63],"However,":[29],"the":[30,99,144],"uneven":[31],"distribution":[32],"of":[33],"requests":[34],"across":[35,105,153],"stages":[37],"creates":[38],"significant":[39],"performance":[40,55,90],"bottlenecks":[41,94],"in":[42,125],"real-world":[43],"deployments.":[44],"To":[45],"address":[46],"this":[47],"challenge,":[48],"we":[49],"present":[50],"<sc":[51,64,108,131],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[52,65,109,132],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">P</small>lanck,":[53],"a":[54,74],"optimization":[56],"framework":[57],"specifically":[58],"designed":[59],"LLM":[62],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">P</small>lanck":[66,110,133],"implements":[67],"fine-grained":[68],"control":[69],"two":[71],"key":[72],"mechanisms:":[73],"progressive":[75],"SLO":[76],"allocation":[77],"strategy":[78],"that":[79,92,130],"dynamically":[80],"adjusts":[81],"time":[82],"constraints":[83],"based":[84],"on":[85],"workload":[86],"patterns,":[87],"and":[88,142,161],"stage-specific":[89],"controllers":[91],"prevent":[93],"before":[95,118],"they":[96],"cascade":[97],"system.":[100],"By":[101],"intelligently":[102],"balancing":[103],"resources":[104],"stages,":[107,155],"effectively":[111],"eliminates":[112],"queue":[113,146],"buildup-essentially":[114],"preventing":[115],"traffic":[116],"congestion":[117],"it":[119],"forms.":[120],"Evaluation":[121],"using":[122],"diverse":[123],"workloads":[124],"real":[126],"cloud":[127],"environments":[128],"demonstrates":[129],"reduces":[134],"P99":[135],"tail":[136],"latency":[137],"by":[138,148],"up":[139],"18%":[141],"decreases":[143],"longest":[145],"lengths":[147],"as":[149,151],"much":[150],"47.8%":[152],"significantly":[156],"improving":[157],"both":[158],"system":[159],"responsiveness":[160],"resource":[162],"utilization.":[163]},"counts_by_year":[],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-10-10T00:00:00"}
