{"id":"https://openalex.org/W4394745249","doi":"https://doi.org/10.1145/3597503.3639232","title":"An Empirical Study on Low GPU Utilization of Deep Learning Jobs","display_name":"An Empirical Study on Low GPU Utilization of Deep Learning Jobs","publication_year":2024,"publication_date":"2024-04-12","ids":{"openalex":"https://openalex.org/W4394745249","doi":"https://doi.org/10.1145/3597503.3639232"},"language":"en","primary_location":{"id":"doi:10.1145/3597503.3639232","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3597503.3639232","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the IEEE/ACM 46th International Conference on Software Engineering","raw_type":"proceedings-article"},"type":"conference-paper","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/A5031774552","display_name":"Yanjie Gao","orcid":"https://orcid.org/0000-0003-1899-8561"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yanjie Gao","raw_affiliation_strings":["Microsoft Research, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-1899-8561","affiliations":[{"raw_affiliation_string":"Microsoft Research, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yichen He","orcid":"https://orcid.org/0009-0009-7357-9221"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yichen He","raw_affiliation_strings":["Microsoft Research, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0009-7357-9221","affiliations":[{"raw_affiliation_string":"Microsoft Research, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050807351","display_name":"Xinze Li","orcid":"https://orcid.org/0009-0002-8466-9336"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xinze Li","raw_affiliation_strings":["Peking University, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-8466-9336","affiliations":[{"raw_affiliation_string":"Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015457479","display_name":"Bo Zhao","orcid":"https://orcid.org/0000-0002-2967-5963"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bo Zhao","raw_affiliation_strings":["Microsoft Research, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-2967-5963","affiliations":[{"raw_affiliation_string":"Microsoft Research, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5013547786","display_name":"Haoxiang Lin","orcid":"https://orcid.org/0000-0002-9148-5861"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haoxiang Lin","raw_affiliation_strings":["Microsoft Research, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-9148-5861","affiliations":[{"raw_affiliation_string":"Microsoft Research, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044145016","display_name":"Y. F. Liang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yoyo Liang","raw_affiliation_strings":["Microsoft, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0007-1661-0013","affiliations":[{"raw_affiliation_string":"Microsoft, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5053813342","display_name":"Jing Zhong","orcid":null},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jing Zhong","raw_affiliation_strings":["Microsoft, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0004-2760-9024","affiliations":[{"raw_affiliation_string":"Microsoft, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100412598","display_name":"Hongyu Zhang","orcid":"https://orcid.org/0000-0002-3063-9425"},"institutions":[{"id":"https://openalex.org/I158842170","display_name":"Chongqing University","ror":"https://ror.org/023rhb549","country_code":"CN","type":"education","lineage":["https://openalex.org/I158842170"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hongyu Zhang","raw_affiliation_strings":["Chongqing University, Chongqing, China"],"raw_orcid":"https://orcid.org/0000-0002-3063-9425","affiliations":[{"raw_affiliation_string":"Chongqing University, Chongqing, China","institution_ids":["https://openalex.org/I158842170"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102003048","display_name":"Jingzhou Wang","orcid":"https://orcid.org/0009-0002-2172-588X"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jingzhou Wang","raw_affiliation_strings":["Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-2172-588X","affiliations":[{"raw_affiliation_string":"Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103665533","display_name":"Yonghua Zeng","orcid":null},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yonghua Zeng","raw_affiliation_strings":["Microsoft, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0004-1304-9265","affiliations":[{"raw_affiliation_string":"Microsoft, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5095376992","display_name":"Keli Gui","orcid":"https://orcid.org/0009-0002-1298-5321"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Keli Gui","raw_affiliation_strings":["Microsoft, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-1298-5321","affiliations":[{"raw_affiliation_string":"Microsoft, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022958107","display_name":"Jiaye Tong","orcid":"https://orcid.org/0009-0001-9812-0187"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jie Tong","raw_affiliation_strings":["Microsoft, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0001-9812-0187","affiliations":[{"raw_affiliation_string":"Microsoft, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100438310","display_name":"Mao Yang","orcid":"https://orcid.org/0009-0009-6455-3898"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Mao Yang","raw_affiliation_strings":["Microsoft Research, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0009-6455-3898","affiliations":[{"raw_affiliation_string":"Microsoft Research, Beijing, China","institution_ids":["https://openalex.org/I4210113369"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":23,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"13"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10101","display_name":"Cloud Computing and Resource Management","score":0.9986000061035156,"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"}},"topics":[{"id":"https://openalex.org/T10101","display_name":"Cloud Computing and Resource Management","score":0.9986000061035156,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9966999888420105,"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/T10273","display_name":"IoT and Edge/Fog Computing","score":0.995199978351593,"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":0.8622952699661255},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.7973923087120056},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.5928266048431396},{"id":"https://openalex.org/keywords/general-purpose-computing-on-graphics-processing-units","display_name":"General-purpose computing on graphics processing units","score":0.5833567976951599},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5699094533920288},{"id":"https://openalex.org/keywords/cloud-computing","display_name":"Cloud computing","score":0.5396734476089478},{"id":"https://openalex.org/keywords/scripting-language","display_name":"Scripting language","score":0.5261042714118958},{"id":"https://openalex.org/keywords/empirical-research","display_name":"Empirical research","score":0.5134257078170776},{"id":"https://openalex.org/keywords/metadata","display_name":"Metadata","score":0.4638180136680603},{"id":"https://openalex.org/keywords/graphics-processing-unit","display_name":"Graphics processing unit","score":0.4412495195865631},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.437522828578949},{"id":"https://openalex.org/keywords/graphics","display_name":"Graphics","score":0.4312064051628113},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.2950936555862427},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.18490204215049744}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8622952699661255},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.7973923087120056},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.5928266048431396},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.5833567976951599},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5699094533920288},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.5396734476089478},{"id":"https://openalex.org/C61423126","wikidata":"https://www.wikidata.org/wiki/Q187432","display_name":"Scripting language","level":2,"score":0.5261042714118958},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.5134257078170776},{"id":"https://openalex.org/C93518851","wikidata":"https://www.wikidata.org/wiki/Q180160","display_name":"Metadata","level":2,"score":0.4638180136680603},{"id":"https://openalex.org/C2779851693","wikidata":"https://www.wikidata.org/wiki/Q183484","display_name":"Graphics processing unit","level":2,"score":0.4412495195865631},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.437522828578949},{"id":"https://openalex.org/C21442007","wikidata":"https://www.wikidata.org/wiki/Q1027879","display_name":"Graphics","level":2,"score":0.4312064051628113},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.2950936555862427},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.18490204215049744},{"id":"https://openalex.org/C111472728","wikidata":"https://www.wikidata.org/wiki/Q9471","display_name":"Epistemology","level":1,"score":0.0},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3597503.3639232","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3597503.3639232","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the IEEE/ACM 46th International Conference on Software Engineering","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Decent work and economic growth","id":"https://metadata.un.org/sdg/8","score":0.46000000834465027}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":44,"referenced_works":["https://openalex.org/W2053154970","https://openalex.org/W2095839586","https://openalex.org/W2145458045","https://openalex.org/W2146854584","https://openalex.org/W2155524176","https://openalex.org/W2489529491","https://openalex.org/W2497580656","https://openalex.org/W2521352760","https://openalex.org/W2542459869","https://openalex.org/W2799144475","https://openalex.org/W2800817893","https://openalex.org/W2850992922","https://openalex.org/W2892181857","https://openalex.org/W2933138175","https://openalex.org/W2954916854","https://openalex.org/W2968594320","https://openalex.org/W2997591727","https://openalex.org/W3035078899","https://openalex.org/W3081168214","https://openalex.org/W3088735455","https://openalex.org/W3109585842","https://openalex.org/W3134991928","https://openalex.org/W3135013702","https://openalex.org/W3138516171","https://openalex.org/W3164944955","https://openalex.org/W3171093681","https://openalex.org/W3188321821","https://openalex.org/W3196239222","https://openalex.org/W3196442060","https://openalex.org/W3197816522","https://openalex.org/W3210465286","https://openalex.org/W4220891756","https://openalex.org/W4226064176","https://openalex.org/W4241104364","https://openalex.org/W4244284331","https://openalex.org/W4245987560","https://openalex.org/W4284692938","https://openalex.org/W4287214056","https://openalex.org/W4307926351","https://openalex.org/W4372261197","https://openalex.org/W6600349396","https://openalex.org/W6601323341","https://openalex.org/W6610640258","https://openalex.org/W6637031373"],"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/W2080146221","https://openalex.org/W2370314112","https://openalex.org/W1912958759","https://openalex.org/W2792081825"],"abstract_inverted_index":{"Deep":[0],"learning":[1,17,46,81,102,143,177],"plays":[2],"a":[3,35,71,197],"critical":[4],"role":[5],"in":[6,60,220],"numerous":[7],"intelligent":[8],"software":[9],"applications.":[10],"Enterprise":[11],"developers":[12,216],"submit":[13],"and":[14,26,64,120,129,150,179,185],"run":[15],"deep":[16,45,80,101,142,176],"jobs":[18,50,87,144],"on":[19,75,84,204],"shared,":[20],"multi-tenant":[21],"platforms":[22,30],"to":[23,43,165,175],"efficiently":[24],"train":[25],"test":[27],"models.":[28],"These":[29],"are":[31,163,173],"typically":[32],"equipped":[33],"with":[34,196],"large":[36],"number":[37,199],"of":[38,55,79,93,112,141,160,170,200],"graphics":[39],"processing":[40],"units":[41],"(GPUs)":[42],"expedite":[44],"computations.":[47],"However,":[48],"certain":[49],"exhibit":[51],"rather":[52],"low":[53,76],"utilization":[54,78,92,140],"the":[56,125,161,171,205],"allocated":[57],"GPUs,":[58],"resulting":[59],"substantial":[61],"resource":[62],"waste":[63],"reduced":[65],"development":[66],"productivity.":[67],"This":[68],"paper":[69],"presents":[70],"comprehensive":[72],"empirical":[73],"study":[74,206],"GPU":[77,91,139,148],"jobs,":[82],"based":[83],"400":[85],"real":[86],"(with":[88],"an":[89],"average":[90],"50%":[94],"or":[95],"less)":[96],"collected":[97],"from":[98,146],"Microsoft's":[99],"internal":[100],"platform.":[103],"We":[104],"discover":[105],"706":[106],"low-GPU-utilization":[107,191],"issues":[108,162,172,192],"through":[109],"meticulous":[110],"examination":[111],"job":[113],"metadata,":[114],"execution":[115],"logs,":[116],"runtime":[117],"metrics,":[118],"scripts,":[119],"programs.":[121],"Furthermore,":[122],"we":[123,208],"identify":[124],"common":[126],"root":[127],"causes":[128],"propose":[130,209],"corresponding":[131],"fixes.":[132],"Our":[133],"main":[134],"findings":[135],"include:":[136],"(1)":[137],"Low":[138],"stems":[145],"insufficient":[147],"computations":[149],"interruptions":[151],"caused":[152],"by":[153],"non-GPU":[154],"tasks;":[155],"(2)":[156],"Approximately":[157],"half":[158],"(46.03%)":[159],"attributed":[164],"data":[166],"operations;":[167],"(3)":[168],"45.18%":[169],"related":[174],"models":[178],"manifest":[180],"during":[181],"both":[182],"model":[183],"training":[184],"evaluation":[186],"stages;":[187],"(4)":[188],"Most":[189],"(84.99%)":[190],"could":[193,214],"be":[194],"fixed":[195],"small":[198],"code/script":[201],"modifications.":[202],"Based":[203],"results,":[207],"potential":[210],"research":[211],"directions":[212],"that":[213],"help":[215],"utilize":[217],"GPUs":[218],"better":[219],"cloud-based":[221],"platforms.":[222]},"counts_by_year":[{"year":2026,"cited_by_count":9},{"year":2025,"cited_by_count":11},{"year":2024,"cited_by_count":3}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}