{"id":"https://openalex.org/W7084116360","doi":"https://doi.org/10.1109/infocom55648.2025.11044693","title":"Espresso: Cost-Efficient Large Model Training by Exploiting GPU Heterogeneity in the Cloud","display_name":"Espresso: Cost-Efficient Large Model Training by Exploiting GPU Heterogeneity in the Cloud","publication_year":2025,"publication_date":"2025-05-19","ids":{"openalex":"https://openalex.org/W7084116360","doi":"https://doi.org/10.1109/infocom55648.2025.11044693"},"language":"en","primary_location":{"id":"doi:10.1109/infocom55648.2025.11044693","is_oa":false,"landing_page_url":"https://doi.org/10.1109/infocom55648.2025.11044693","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE INFOCOM 2025 - IEEE Conference on Computer Communications","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":null,"display_name":"Qiannan Zhou","orcid":null},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qiannan Zhou","raw_affiliation_strings":["East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing","institution_ids":["https://openalex.org/I66867065"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Fei Xu","orcid":null},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Fei Xu","raw_affiliation_strings":["East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing","institution_ids":["https://openalex.org/I66867065"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Lingxuan Weng","orcid":null},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Lingxuan Weng","raw_affiliation_strings":["East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing","institution_ids":["https://openalex.org/I66867065"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Ruixing Li","orcid":null},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ruixing Li","raw_affiliation_strings":["East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing","institution_ids":["https://openalex.org/I66867065"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Xudong Wu","orcid":null},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xudong Wu","raw_affiliation_strings":["East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University.,Shanghai Key Laboratory of Multidimensional Information Processing","institution_ids":["https://openalex.org/I66867065"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Li Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I79516672","display_name":"University of Louisiana at Lafayette","ror":"https://ror.org/01x8rc503","country_code":"US","type":"education","lineage":["https://openalex.org/I2799628689","https://openalex.org/I79516672"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Li Chen","raw_affiliation_strings":["University of Louisiana,Lafayette"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Louisiana,Lafayette","institution_ids":["https://openalex.org/I79516672"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Zhi Zhou","orcid":null},"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":"Zhi Zhou","raw_affiliation_strings":["Sun Yat-sen University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Sun Yat-sen University","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"last","author":{"id":null,"display_name":"Fangming Liu","orcid":null},"institutions":[{"id":"https://openalex.org/I4210136793","display_name":"Peng Cheng Laboratory","ror":"https://ror.org/03qdqbt06","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210136793"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Fangming Liu","raw_affiliation_strings":["Peng Cheng Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Peng Cheng Laboratory","institution_ids":["https://openalex.org/I4210136793"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":8,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.9349,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.82248813,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"10"},"is_retracted":false,"is_paratext":false,"is_xpac":true,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.3741999864578247,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.3741999864578247,"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/T10101","display_name":"Cloud Computing and Resource Management","score":0.25130000710487366,"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/T14347","display_name":"Big Data and Digital Economy","score":0.10369999706745148,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/cloud-computing","display_name":"Cloud computing","score":0.8525000214576721},{"id":"https://openalex.org/keywords/provisioning","display_name":"Provisioning","score":0.8500000238418579},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.5429999828338623},{"id":"https://openalex.org/keywords/resource-allocation","display_name":"Resource allocation","score":0.4659000039100647},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4652999937534332},{"id":"https://openalex.org/keywords/computation-offloading","display_name":"Computation offloading","score":0.4020000100135803},{"id":"https://openalex.org/keywords/supercomputer","display_name":"Supercomputer","score":0.3937999904155731},{"id":"https://openalex.org/keywords/resource","display_name":"Resource (disambiguation)","score":0.3840999901294708},{"id":"https://openalex.org/keywords/homogeneous","display_name":"Homogeneous","score":0.37130001187324524}],"concepts":[{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.8525000214576721},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8508999943733215},{"id":"https://openalex.org/C172191483","wikidata":"https://www.wikidata.org/wiki/Q1071806","display_name":"Provisioning","level":2,"score":0.8500000238418579},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.6547999978065491},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.5429999828338623},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.4659000039100647},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4652999937534332},{"id":"https://openalex.org/C2781041963","wikidata":"https://www.wikidata.org/wiki/Q18348618","display_name":"Computation offloading","level":4,"score":0.4020000100135803},{"id":"https://openalex.org/C83283714","wikidata":"https://www.wikidata.org/wiki/Q121117","display_name":"Supercomputer","level":2,"score":0.3937999904155731},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.3840999901294708},{"id":"https://openalex.org/C66882249","wikidata":"https://www.wikidata.org/wiki/Q169336","display_name":"Homogeneous","level":2,"score":0.37130001187324524},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.36070001125335693},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.3594000041484833},{"id":"https://openalex.org/C172430144","wikidata":"https://www.wikidata.org/wiki/Q17111997","display_name":"Symmetric multiprocessor system","level":2,"score":0.3508000075817108},{"id":"https://openalex.org/C11644782","wikidata":"https://www.wikidata.org/wiki/Q15401790","display_name":"Cost efficiency","level":2,"score":0.3246000111103058},{"id":"https://openalex.org/C111696304","wikidata":"https://www.wikidata.org/wiki/Q2303697","display_name":"Sorting","level":2,"score":0.3140999972820282},{"id":"https://openalex.org/C2780609101","wikidata":"https://www.wikidata.org/wiki/Q17156588","display_name":"Resource management (computing)","level":2,"score":0.303600013256073},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.28870001435279846},{"id":"https://openalex.org/C20556612","wikidata":"https://www.wikidata.org/wiki/Q4469374","display_name":"Volume (thermodynamics)","level":2,"score":0.28700000047683716},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.27900001406669617},{"id":"https://openalex.org/C93996380","wikidata":"https://www.wikidata.org/wiki/Q44127","display_name":"Server","level":2,"score":0.2782000005245209},{"id":"https://openalex.org/C2780757406","wikidata":"https://www.wikidata.org/wiki/Q465837","display_name":"Skyline","level":2,"score":0.2743000090122223},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.2687000036239624},{"id":"https://openalex.org/C138959212","wikidata":"https://www.wikidata.org/wiki/Q1806783","display_name":"Load balancing (electrical power)","level":3,"score":0.2651999890804291},{"id":"https://openalex.org/C2777958785","wikidata":"https://www.wikidata.org/wiki/Q17120940","display_name":"Resource efficiency","level":2,"score":0.25619998574256897},{"id":"https://openalex.org/C34165917","wikidata":"https://www.wikidata.org/wiki/Q188267","display_name":"Programming paradigm","level":2,"score":0.2551000118255615}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/infocom55648.2025.11044693","is_oa":false,"landing_page_url":"https://doi.org/10.1109/infocom55648.2025.11044693","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE INFOCOM 2025 - IEEE Conference on Computer Communications","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G4114173854","display_name":null,"funder_award_id":"OIA-2327452","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0],"Transformer-based":[1],"models":[2,9,170],"deepen":[3],"and":[4,31,62,99,130,150,162],"datasets":[5],"expand,":[6],"training":[7,42,50,76,108,186],"large":[8,48,106,169,184],"demands":[10,64],"numerous":[11],"accelerators,":[12],"particularly":[13],"GPUs,":[14],"bringing":[15],"high":[16],"cloud":[17,29],"expenses.":[18],"However,":[19],"conventional":[20],"homogeneous":[21],"resource":[22],"provisioning":[23,40,89,120],"is":[24],"inefficient":[25],"due":[26],"to":[27,122,147,192,196],"limited":[28],"resources":[30],"low":[32],"GPU":[33,39,88,95,114],"utilization.":[34],"This":[35,82],"challenge":[36],"necessitates":[37],"heterogeneous":[38,71,94],"for":[41,75,105,183],"in":[43,109,171],"clouds.":[44,173],"Current":[45],"research":[46],"on":[47,53,166],"model":[49,107,185],"often":[51],"focuses":[52],"load":[54],"balancing":[55],"of":[56,70,160],"stages,":[57],"neglecting":[58],"the":[59,68,93,110,113,181],"varying":[60],"computing":[61],"memory":[63],"across":[65],"stages.":[66],"Additionally,":[67],"allocation":[69,96,125],"G":[72],"PU":[73],"s":[74],"has":[77],"surprisingly":[78],"received":[79],"little":[80],"attention.":[81],"paper":[83],"introduces":[84],"Espresso,":[85],"a":[86,117,144,158],"cost-efficient":[87],"framework":[90],"that":[91,178],"unifies":[92],"(GPU":[97],"allocator)":[98],"adequate":[100],"stage":[101,140],"placement":[102],"(stage":[103],"placer)":[104],"cloud.":[111],"Specifically,":[112],"allocator":[115],"proposes":[116],"cost":[118],"tree-based":[119],"strategy":[121],"prioritize":[123],"searching":[124],"plans":[126],"with":[127,200],"lower":[128],"costs":[129,189],"reduce":[131],"unnecessary":[132],"branches":[133],"by":[134,190],"multi-dimensional":[135],"pruning":[136],"methods.":[137],"The":[138],"resource-aware":[139],"placer":[141],"further":[142],"devises":[143],"compute-memory":[145],"ratio":[146],"optimize":[148],"communication":[149],"computation":[151],"efficiency":[152],"during":[153],"training.":[154],"We":[155],"have":[156],"open-sourced":[157],"prototype":[159,164],"Espresso":[161,179],"conducted":[163],"experiments":[165],"four":[167],"representative":[168],"public":[172],"Extensive":[174],"experiment":[175],"results":[176],"demonstrate":[177],"guarantees":[180],"performance":[182],"while":[187],"saving":[188],"up":[191],"49.8":[193],"%":[194],"compared":[195],"state-of-the-art":[197],"solutions,":[198],"yet":[199],"acceptable":[201],"runtime":[202],"overhead.":[203]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}