{"id":"https://openalex.org/W7133532395","doi":"https://doi.org/10.1109/hpca68181.2026.11408533","title":"AutoHAAP: Automated Heterogeneity-Aware Asymmetric Partitioning for LLM Training","display_name":"AutoHAAP: Automated Heterogeneity-Aware Asymmetric Partitioning for LLM Training","publication_year":2026,"publication_date":"2026-01-31","ids":{"openalex":"https://openalex.org/W7133532395","doi":"https://doi.org/10.1109/hpca68181.2026.11408533"},"language":null,"primary_location":{"id":"doi:10.1109/hpca68181.2026.11408533","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408533","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","raw_type":"proceedings-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/A5128057510","display_name":"Yuanyuan Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Yuanyuan Wang","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128081061","display_name":"Nana Tang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Nana Tang","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128092570","display_name":"Yuyang Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuyang Wang","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128078367","display_name":"Shu Pan","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shu Pan","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5007508057","display_name":"Dingding Yu","orcid":"https://orcid.org/0000-0001-6352-2007"},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Dingding Yu","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Zeyue Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zeyue Wang","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127399095","display_name":"Mou Sun","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Mou Sun","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050236956","display_name":"Kejie Fu","orcid":"https://orcid.org/0000-0002-6764-6813"},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kejie Fu","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128124863","display_name":"Fangyu Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Fangyu Wang","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yunchuan Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yunchuan Chen","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128041476","display_name":"Ning Sun","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ning Sun","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5128092707","display_name":"Fei Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Fei Yang","raw_affiliation_strings":["Zhejiang Lab"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab","institution_ids":["https://openalex.org/I4210123185"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":12,"corresponding_author_ids":["https://openalex.org/A5128057510"],"corresponding_institution_ids":["https://openalex.org/I4210123185"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.53563156,"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":"17"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12535","display_name":"Machine Learning and Data Classification","score":0.29190000891685486,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T12535","display_name":"Machine Learning and Data Classification","score":0.29190000891685486,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.0778999999165535,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T12292","display_name":"Graph Theory and Algorithms","score":0.04500000178813934,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/training","display_name":"Training (meteorology)","score":0.4986000061035156},{"id":"https://openalex.org/keywords/automation","display_name":"Automation","score":0.3273000121116638},{"id":"https://openalex.org/keywords/troubleshooting","display_name":"Troubleshooting","score":0.3140999972820282},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.30709999799728394},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.2849999964237213}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5669999718666077},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4986000061035156},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3935999870300293},{"id":"https://openalex.org/C115901376","wikidata":"https://www.wikidata.org/wiki/Q184199","display_name":"Automation","level":2,"score":0.3273000121116638},{"id":"https://openalex.org/C147494362","wikidata":"https://www.wikidata.org/wiki/Q2078905","display_name":"Troubleshooting","level":2,"score":0.3140999972820282},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.30709999799728394},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.2849999964237213},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.2775000035762787},{"id":"https://openalex.org/C2776904630","wikidata":"https://www.wikidata.org/wiki/Q356336","display_name":"Adept","level":3,"score":0.24469999969005585},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.24369999766349792}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/hpca68181.2026.11408533","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408533","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G119839963","display_name":null,"funder_award_id":"U22A6001","funder_id":"https://openalex.org/F4320320885","funder_display_name":"European Research Consortium for Informatics and Mathematics"}],"funders":[{"id":"https://openalex.org/F4320320885","display_name":"European Research Consortium for Informatics and Mathematics","ror":"https://ror.org/055zrhj18"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":38,"referenced_works":["https://openalex.org/W2612026221","https://openalex.org/W2911382970","https://openalex.org/W2955013293","https://openalex.org/W2969388332","https://openalex.org/W3081168214","https://openalex.org/W3177828909","https://openalex.org/W3194592686","https://openalex.org/W3204998121","https://openalex.org/W3206832494","https://openalex.org/W4283379731","https://openalex.org/W4310282800","https://openalex.org/W4312060029","https://openalex.org/W4327694855","https://openalex.org/W4360831977","https://openalex.org/W4383197616","https://openalex.org/W4383218913","https://openalex.org/W4385597806","https://openalex.org/W4386709668","https://openalex.org/W4387321109","https://openalex.org/W4387500346","https://openalex.org/W4392207935","https://openalex.org/W4393406906","https://openalex.org/W4394922822","https://openalex.org/W4395020669","https://openalex.org/W4395106447","https://openalex.org/W4395117922","https://openalex.org/W4396721167","https://openalex.org/W4401408735","https://openalex.org/W4401568760","https://openalex.org/W4401671778","https://openalex.org/W4406152279","https://openalex.org/W4406472463","https://openalex.org/W4408858386","https://openalex.org/W4408925916","https://openalex.org/W4411486137","https://openalex.org/W4412888297","https://openalex.org/W4413169625","https://openalex.org/W4413411773"],"related_works":[],"abstract_inverted_index":{"Heterogeneous":[0],"clusters":[1,190],"with":[2,101],"diverse":[3,249],"devices":[4],"mitigate":[5],"computational":[6,107],"and":[7,26,36,82,117,167,179,188,223],"memory":[8],"burdens":[9],"in":[10,64,79,127,220,236,244],"large":[11,66],"language":[12],"model":[13],"(LLM)":[14],"training,":[15],"yet":[16],"their":[17],"inherent":[18],"resource":[19,83],"heterogeneity,":[20],"characterized":[21],"by":[22],"divergent":[23],"computation,":[24],"memory,":[25],"bandwidth":[27],"capabilities,":[28],"renders":[29],"manual":[30],"parallelization":[31,39],"strategy":[32,89],"optimization":[33],"both":[34,186],"challenging":[35],"time-intensive.":[37],"Automatic":[38],"is":[40],"critical":[41],"for":[42,99],"scaling":[43],"complex":[44],"workloads":[45],"across":[46,177],"heterogeneous":[47,189,237],"architectures.":[48],"However,":[49],"previous":[50],"methodologies":[51],"face":[52],"significant":[53],"inefficiencies.":[54],"First,":[55],"insufficient":[56,110],"pruning":[57],"of":[58,182],"the":[59,70,124],"parameter":[60],"initialization":[61,150],"space":[62,112],"results":[63,240],"impractically":[65],"search":[67,74,111,155,202],"spaces.":[68],"Second,":[69],"prevailing":[71],"automatic":[72],"parallel":[73,88],"strategies":[75],"exhibit":[76],"suboptimal":[77,114],"performance":[78],"load":[80],"balancing":[81],"constraint":[84],"adaptation.":[85],"Third,":[86],"dynamic":[87],"tuning":[90],"incurs":[91],"substantial":[92],"overhead":[93],"due":[94],"to":[95,105,151],"redundant":[96,118,173],"latency":[97,119,174],"calculations":[98],"operators":[100],"unchanged":[102],"configurations,":[103],"leading":[104],"unnecessary":[106],"costs.":[108],"Therefore,":[109],"pruning,":[113],"load/resource":[115],"adaptation,":[116],"computation":[120],"are":[121],"identified":[122],"as":[123],"major":[125],"bottlenecks":[126],"our":[128],"research.":[129],"To":[130],"address":[131],"these":[132],"challenges,":[133],"we":[134],"propose":[135],"AutoHAAP":[136,193],"(Automated":[137],"Heterogeneity-Aware":[138],"Asymmetric":[139],"Partitioning),":[140],"a":[141,158],"novel":[142],"framework":[143],"incorporating":[144],"three":[145],"core":[146],"innovations:":[147],"(1)":[148],"memory-aware":[149],"drastically":[152],"reduce":[153],"viable":[154],"spaces;":[156],"(2)":[157],"heterogeneity-aware":[159],"load-balancing":[160],"estimator":[161],"that":[162,192],"guides":[163],"resource-efficient":[164],"configuration":[165],"search;":[166],"(3)":[168],"state":[169],"caching":[170],"mechanisms":[171],"eliminating":[172],"calculations.":[175],"Evaluations":[176],"GPT3":[178],"Llama3":[180],"models":[181],"varying":[183],"scales":[184],"on":[185,248],"homogeneous":[187,221],"demonstrate":[191],"achieves":[194],"<tex":[195,205,224],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[196,206,225],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$\\mathbf{0.":[197],"6":[198],"8}-\\mathbf{9":[199],"8}":[200],"\\times$</tex>":[201,233],"efficiency":[203],"gains,":[204],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$\\mathbf{6.":[207],"5":[208],"7":[209],"\\%}":[210,216,229],"\\boldsymbol{-}":[211,230],"\\mathbf{1":[212],"0":[213],"6.":[214],"9":[215],"\\boldsymbol{\\times}$</tex>":[217],"throughput":[218,234],"improvements":[219],"environments,":[222],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$\\mathbf{1":[226],"0.":[227],"1":[228],"22.28":[231],"\\%":[232],"enhancements":[235],"setups.":[238],"These":[239],"validate":[241],"AutoHAAP's":[242],"effectiveness":[243],"distributed":[245],"LLM":[246],"training":[247],"hardware.":[250]},"counts_by_year":[],"updated_date":"2026-04-09T08:11:56.329763","created_date":"2026-03-05T00:00:00"}