{"id":"https://openalex.org/W4408715930","doi":"https://doi.org/10.1109/tpds.2025.3553137","title":"IceFrog: A Layer-Elastic Scheduling System for Deep Learning Training in GPU Clusters","display_name":"IceFrog: A Layer-Elastic Scheduling System for Deep Learning Training in GPU Clusters","publication_year":2025,"publication_date":"2025-03-21","ids":{"openalex":"https://openalex.org/W4408715930","doi":"https://doi.org/10.1109/tpds.2025.3553137"},"language":"en","primary_location":{"id":"doi:10.1109/tpds.2025.3553137","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2025.3553137","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"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 Parallel and Distributed Systems","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/A5100732619","display_name":"Wei Gao","orcid":"https://orcid.org/0000-0002-7048-1722"},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Wei Gao","raw_affiliation_strings":["College of Computational and Data Science, Nanyang Technological University, Singapore","College of Computational and Data Science, Nanyang Technological University, China"],"raw_orcid":"https://orcid.org/0000-0002-7048-1722","affiliations":[{"raw_affiliation_string":"College of Computational and Data Science, Nanyang Technological University, Singapore","institution_ids":["https://openalex.org/I172675005"]},{"raw_affiliation_string":"College of Computational and Data Science, Nanyang Technological University, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001910697","display_name":"Zhuoyuan Ouyang","orcid":null},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Zhuoyuan Ouyang","raw_affiliation_strings":["College of Computational and Data Science, Nanyang Technological University, Singapore","College of Computational and Data Science, Nanyang Technological University, China"],"raw_orcid":"https://orcid.org/0009-0009-8998-8628","affiliations":[{"raw_affiliation_string":"College of Computational and Data Science, Nanyang Technological University, Singapore","institution_ids":["https://openalex.org/I172675005"]},{"raw_affiliation_string":"College of Computational and Data Science, Nanyang Technological University, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"Peng Sun","orcid":"https://orcid.org/0000-0001-8456-0491"},"institutions":[{"id":"https://openalex.org/I161346416","display_name":"Jimei University","ror":"https://ror.org/03hknyb50","country_code":"CN","type":"education","lineage":["https://openalex.org/I161346416"]},{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]},{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":null,"type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Peng Sun","raw_affiliation_strings":["Shanghai AI Lab &#x0026; Sensetime, Shanghai, China","Shanghai Artificial Intelligence Laboratory, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0001-8456-0491","affiliations":[{"raw_affiliation_string":"Shanghai AI Lab &#x0026; Sensetime, Shanghai, China","institution_ids":["https://openalex.org/I161346416"]},{"raw_affiliation_string":"Shanghai Artificial Intelligence Laboratory, Shanghai, China","institution_ids":["https://openalex.org/I4210100255","https://openalex.org/I4391012619"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101591101","display_name":"Tianwei Zhang","orcid":"https://orcid.org/0000-0001-6595-6650"},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Tianwei Zhang","raw_affiliation_strings":["College of Computational and Data Science, Nanyang Technological University, Singapore","College of Computational and Data Science, Nanyang Technological University, China"],"raw_orcid":"https://orcid.org/0000-0001-6595-6650","affiliations":[{"raw_affiliation_string":"College of Computational and Data Science, Nanyang Technological University, Singapore","institution_ids":["https://openalex.org/I172675005"]},{"raw_affiliation_string":"College of Computational and Data Science, Nanyang Technological University, China","institution_ids":[]}]},{"author_position":"last","author":{"id":null,"display_name":"Yonggang Wen","orcid":"https://orcid.org/0000-0002-2751-5114"},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Yonggang Wen","raw_affiliation_strings":["College of Computational and Data Science, Nanyang Technological University, Singapore"],"raw_orcid":"https://orcid.org/0000-0002-2751-5114","affiliations":[{"raw_affiliation_string":"College of Computational and Data Science, Nanyang Technological University, Singapore","institution_ids":["https://openalex.org/I172675005"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.9389,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.7335137,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":"36","issue":"6","first_page":"1071","last_page":"1086"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10715","display_name":"Distributed and Parallel Computing Systems","score":0.8133000135421753,"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":0.8133000135421753,"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/computer-science","display_name":"Computer science","score":0.8706362247467041},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.604776918888092},{"id":"https://openalex.org/keywords/processor-scheduling","display_name":"Processor scheduling","score":0.5290775895118713},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.5014166831970215},{"id":"https://openalex.org/keywords/gpu-cluster","display_name":"GPU cluster","score":0.4632958173751831},{"id":"https://openalex.org/keywords/layer","display_name":"Layer (electronics)","score":0.42227500677108765},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.3513154089450836},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.34880977869033813},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.25274229049682617},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.2233525812625885},{"id":"https://openalex.org/keywords/schedule","display_name":"Schedule","score":0.09806233644485474},{"id":"https://openalex.org/keywords/composite-material","display_name":"Composite material","score":0.09138420224189758},{"id":"https://openalex.org/keywords/materials-science","display_name":"Materials science","score":0.07684621214866638}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8706362247467041},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.604776918888092},{"id":"https://openalex.org/C2984822820","wikidata":"https://www.wikidata.org/wiki/Q1123036","display_name":"Processor scheduling","level":3,"score":0.5290775895118713},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5014166831970215},{"id":"https://openalex.org/C2781335571","wikidata":"https://www.wikidata.org/wiki/Q2633544","display_name":"GPU cluster","level":3,"score":0.4632958173751831},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.42227500677108765},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3513154089450836},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.34880977869033813},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.25274229049682617},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.2233525812625885},{"id":"https://openalex.org/C68387754","wikidata":"https://www.wikidata.org/wiki/Q7271585","display_name":"Schedule","level":2,"score":0.09806233644485474},{"id":"https://openalex.org/C159985019","wikidata":"https://www.wikidata.org/wiki/Q181790","display_name":"Composite material","level":1,"score":0.09138420224189758},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.07684621214866638},{"id":"https://openalex.org/C21547014","wikidata":"https://www.wikidata.org/wiki/Q1423657","display_name":"Operations management","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/tpds.2025.3553137","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2025.3553137","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"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 Parallel and Distributed Systems","raw_type":"journal-article"},{"id":"pmh:oai:dr.ntu.edu.sg:10356/218218","is_oa":false,"landing_page_url":"https://hdl.handle.net/10356/218218","pdf_url":null,"source":{"id":"https://openalex.org/S4306402609","display_name":"DR-NTU (Nanyang Technological University)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I172675005","host_organization_name":"Nanyang Technological University","host_organization_lineage":["https://openalex.org/I172675005"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"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":52,"referenced_works":["https://openalex.org/W2031489346","https://openalex.org/W2108598243","https://openalex.org/W2194775991","https://openalex.org/W2798515322","https://openalex.org/W2896457183","https://openalex.org/W2963037989","https://openalex.org/W2963163009","https://openalex.org/W2963323070","https://openalex.org/W2969388332","https://openalex.org/W2981943644","https://openalex.org/W2991040477","https://openalex.org/W3011083606","https://openalex.org/W3084357044","https://openalex.org/W3129831491","https://openalex.org/W3134991928","https://openalex.org/W3137147200","https://openalex.org/W3157306683","https://openalex.org/W3203600045","https://openalex.org/W3206418100","https://openalex.org/W3209503812","https://openalex.org/W4246193833","https://openalex.org/W4308426163","https://openalex.org/W4310282800","https://openalex.org/W4318541676","https://openalex.org/W4372260356","https://openalex.org/W4386768656","https://openalex.org/W4387321109","https://openalex.org/W4392309097","https://openalex.org/W4394947044","https://openalex.org/W4399282175","https://openalex.org/W4399282536","https://openalex.org/W6637373629","https://openalex.org/W6691459498","https://openalex.org/W6727099177","https://openalex.org/W6739369274","https://openalex.org/W6745682157","https://openalex.org/W6752726010","https://openalex.org/W6756009870","https://openalex.org/W6758283263","https://openalex.org/W6759814162","https://openalex.org/W6779823529","https://openalex.org/W6781728138","https://openalex.org/W6782839094","https://openalex.org/W6785115463","https://openalex.org/W6787972765","https://openalex.org/W6790064496","https://openalex.org/W6793903029","https://openalex.org/W6796414051","https://openalex.org/W6845705509","https://openalex.org/W6850380837","https://openalex.org/W6850428050","https://openalex.org/W7067822191"],"related_works":["https://openalex.org/W2082176405","https://openalex.org/W2238531340","https://openalex.org/W2067902980","https://openalex.org/W2017587301","https://openalex.org/W1827514603","https://openalex.org/W2965037703","https://openalex.org/W1809394610","https://openalex.org/W2972376311","https://openalex.org/W3215286539","https://openalex.org/W2023938924"],"abstract_inverted_index":{"The":[0],"high":[1],"resource":[2,159],"demand":[3],"of":[4,13,41,83,143,178,192,206],"deep":[5],"learning":[6],"training":[7,119],"(DLT)":[8],"workloads":[9,22,145],"necessitates":[10],"the":[11,81,90,131,141,166,174,190],"design":[12,183],"efficient":[14,112],"schedulers.":[15,230],"While":[16],"most":[17],"existing":[18],"schedulers":[19],"expedite":[20],"DLT":[21,68,144],"by":[23,224],"considering":[24],"GPU":[25,87,147],"sharing":[26],"and":[27,121,157,169,197,211],"elastic":[28],"training,":[29],"they":[30],"neglect":[31],"<italic":[32,63],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[33,64,129,200,217],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">layer":[34,65],"elasticity</i>,":[35],"which":[36,115],"dynamically":[37],"freezes":[38],"certain":[39],"layers":[40],"a":[42,184,203],"network.":[43],"This":[44],"technique":[45],"has":[46],"been":[47],"shown":[48],"to":[49,61,71,99,109,139,172,187,227],"significantly":[50],"speed":[51,122],"up":[52],"individual":[53],"workloads.":[54,105,180],"In":[55,161],"this":[56,151],"paper,":[57],"we":[58,164],"explore":[59],"how":[60],"incorporate":[62],"elasticity</i>":[66],"into":[67],"scheduler":[69,186],"designs":[70,156],"achieve":[72],"higher":[73],"cluster-wide":[74],"efficiency.":[75],"A":[76],"key":[77],"factor":[78],"that":[79,135,215],"hinders":[80],"application":[82],"layer":[84,101,124,137,193],"elasticity":[85,102,138],"in":[86,93,146,202],"clusters":[88],"is":[89,107],"potential":[91],"loss":[92],"model":[94,165],"accuracy,":[95],"making":[96],"users":[97],"reluctant":[98],"enable":[100],"for":[103,123],"their":[104],"It":[106,149],"necessary":[108],"have":[110],"an":[111],"layer-elastic":[113,179],"system,":[114],"can":[116],"well":[117],"balance":[118],"accuracy":[120],"elasticity.":[125,194],"We":[126,182,195],"introduce":[127],"<sc":[128,199,216],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">IceFrog</small>,":[130],"first":[132],"scheduling":[133],"system":[134],"utilizes":[136],"improve":[140,189],"efficiency":[142,191],"clusters.":[148],"achieves":[150],"goal":[152],"with":[153],"superior":[154],"algorithmic":[155],"intelligent":[158],"management.":[160],"particular,":[162],"(1)":[163],"frozen":[167],"penalty":[168],"layer-aware":[170],"throughput":[171],"measure":[173],"effective":[175],"progress":[176],"metric":[177],"(2)":[181],"novel":[185],"further":[188],"implement":[196],"deploy":[198],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">IceFrog</small>":[201,218],"physical":[204],"cluster":[205],"48":[207],"GPUs.":[208],"Extensive":[209],"evaluations":[210],"large-scale":[212],"simulations":[213],"show":[214],"reduces":[219],"average":[220],"job":[221],"completion":[222],"times":[223],"36-48%":[225],"relative":[226],"state-of-the-art":[228],"DL":[229]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2025-12-21T01:58:51.020947","created_date":"2025-10-10T00:00:00"}