{"id":"https://openalex.org/W7166800250","doi":"https://doi.org/10.18653/v1/2026.acl-long.619","title":"SpiderFlow: Efficient Topology-Aware Scheduling for LLM Training Across Decentralized GPU Clusters","display_name":"SpiderFlow: Efficient Topology-Aware Scheduling for LLM Training Across Decentralized GPU Clusters","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166800250","doi":"https://doi.org/10.18653/v1/2026.acl-long.619"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.619","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.619","pdf_url":"https://aclanthology.org/2026.acl-long.619.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.619.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139768197","display_name":"Zihan Chang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zihan Chang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004997843","display_name":"Shuibing He","orcid":"https://orcid.org/0000-0002-7075-4153"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shuibing He","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139814021","display_name":"Bo Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bo Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139751121","display_name":"Sheng Xiao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sheng Xiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5043851369","display_name":"Siling Yang","orcid":"https://orcid.org/0000-0002-2208-559X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Siling Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139771966","display_name":"Rui Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rui Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5060678493","display_name":"Zhe Pan","orcid":"https://orcid.org/0000-0002-0355-8099"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhe Pan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.840818,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"13603","last_page":"13615"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10715","display_name":"Distributed and Parallel Computing Systems","score":0.2393999993801117,"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/T10715","display_name":"Distributed and Parallel Computing Systems","score":0.2393999993801117,"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/T10101","display_name":"Cloud Computing and Resource Management","score":0.14720000326633453,"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":0.10980000346899033,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.5719000101089478},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.5317999720573425},{"id":"https://openalex.org/keywords/software","display_name":"Software","score":0.2603999972343445},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.2597000002861023},{"id":"https://openalex.org/keywords/field","display_name":"Field (mathematics)","score":0.2526000142097473}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.692799985408783},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5719000101089478},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.5317999720573425},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.35920000076293945},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.3352000117301941},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.30889999866485596},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.2603999972343445},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.2597000002861023},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.2526000142097473},{"id":"https://openalex.org/C55416958","wikidata":"https://www.wikidata.org/wiki/Q6206757","display_name":"Job shop scheduling","level":3,"score":0.25200000405311584},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.2502000033855438}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.619","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.619","pdf_url":"https://aclanthology.org/2026.acl-long.619.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.619","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.619","pdf_url":"https://aclanthology.org/2026.acl-long.619.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2746217201","display_name":null,"funder_award_id":"2025SSYS0005","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6212169838","display_name":null,"funder_award_id":"U25A20423","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320338469","display_name":"Science and Technology Program of Zhejiang Province","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166800250.pdf","grobid_xml":"https://content.openalex.org/works/W7166800250.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In":[0],"response":[1],"to":[2,59,64,170,196],"increasing":[3],"demands":[4],"for":[5,87,109],"largescale":[6],"machine":[7],"learning":[8],"training":[9,28,144],"jobs,":[10],"many":[11],"organizations":[12],"have":[13],"deployed":[14],"GPU":[15,89],"clusters":[16],"across":[17],"geographically":[18],"distributed":[19],"regions.However,":[20],"existing":[21],"Integer":[22],"Linear":[23],"Programming":[24],"(ILP)-or":[25],"genetic-based":[26],"cross-cluster":[27,56,92],"approaches":[29],"largely":[30],"overlook":[31],"the":[32,65,128,138],"topology":[33],"of":[34],"decentralized":[35,88,159,172],"clusters,":[36,135,160],"lacking":[37],"both":[38],"topology-aware":[39,105],"task":[40,93],"scheduling":[41,61,83,94,106,121,149,189,198],"mechanisms":[42],"and":[43,100],"automated":[44,110],"model":[45,111],"parallelization":[46],"strategies.As":[47],"a":[48,81,96,103,119,154,165],"result,":[49],"naively":[50],"applying":[51],"these":[52,76],"optimization-based":[53],"methods":[54,126],"in":[55],"settings":[57],"leads":[58],"prohibitive":[60],"overhead,":[62],"due":[63],"drastically":[66],"enlarged":[67],"search":[68,139],"space":[69,140],"induced":[70],"by":[71,183,186,191],"complex":[72],"inter-cluster":[73,129],"topologies.To":[74],"address":[75],"challenges,":[77],"we":[78,113],"propose":[79,114],"SpiderFlow,":[80],"topologyaware":[82],"system":[84],"specifically":[85],"designed":[86],"clusters.We":[90],"formulate":[91],"as":[95,161,163],"graph":[97],"optimization":[98,133],"problem":[99],"introduce":[101],"SpinSearch,":[102],"low-overhead":[104],"algorithm.In":[107],"addition,":[108],"parallelization,":[112],"Topology-aware":[115],"Parallelism":[116],"Automation":[117],"(TPA),":[118],"two-level":[120],"framework":[122],"that":[123,176],"combines":[124],"heuristic":[125],"at":[127],"level":[130],"with":[131,146,168],"ILP-based":[132],"within":[134],"effectively":[136],"reducing":[137],"while":[141],"maintaining":[142],"high":[143],"throughput":[145,185],"substantially":[147],"lower":[148],"overhead.We":[150],"evaluate":[151],"SpiderFlow":[152,177],"on":[153,164,193],"physical":[155],"platform":[156,167],"comprising":[157],"8":[158],"well":[162],"simulation":[166],"up":[169],"64":[171],"clusters.Experimental":[173],"results":[174],"demonstrate":[175],"reduces":[178,188],"job":[179],"completion":[180],"time":[181],"(JCT)":[182],"1.2-1.3\u00d7,improves":[184],"1.12-1.25\u00d7,and":[187],"overhead":[190],"20-90\u00d7":[192],"average":[194],"compared":[195],"state-of-the-art":[197],"systems.":[199]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}