{"id":"https://openalex.org/W4409060008","doi":"https://doi.org/10.1145/3721146.3721955","title":"Understanding Oversubscribed Memory Management for Deep Learning Training","display_name":"Understanding Oversubscribed Memory Management for Deep Learning Training","publication_year":2025,"publication_date":"2025-03-30","ids":{"openalex":"https://openalex.org/W4409060008","doi":"https://doi.org/10.1145/3721146.3721955"},"language":"en","primary_location":{"id":"doi:10.1145/3721146.3721955","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3721146.3721955","pdf_url":null,"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 5th Workshop on Machine Learning and Systems","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3721146.3721955","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5103667561","display_name":"Mao Lin","orcid":"https://orcid.org/0000-0002-1460-0766"},"institutions":[{"id":"https://openalex.org/I156087764","display_name":"University of California, Merced","ror":"https://ror.org/00d9ah105","country_code":"US","type":"education","lineage":["https://openalex.org/I156087764"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Mao Lin","raw_affiliation_strings":["University of California, Merced, Merced, California, USA"],"raw_orcid":"https://orcid.org/0000-0002-1460-0766","affiliations":[{"raw_affiliation_string":"University of California, Merced, Merced, California, USA","institution_ids":["https://openalex.org/I156087764"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5081531196","display_name":"Hyeran Jeon","orcid":"https://orcid.org/0000-0002-1767-8198"},"institutions":[{"id":"https://openalex.org/I156087764","display_name":"University of California, Merced","ror":"https://ror.org/00d9ah105","country_code":"US","type":"education","lineage":["https://openalex.org/I156087764"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hyeran Jeon","raw_affiliation_strings":["University of California, Merced, Merced, California, USA"],"raw_orcid":"https://orcid.org/0000-0002-1767-8198","affiliations":[{"raw_affiliation_string":"University of California, Merced, Merced, California, USA","institution_ids":["https://openalex.org/I156087764"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I156087764"],"apc_list":null,"apc_paid":null,"fwci":6.0602,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.96104928,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"46","last_page":"55"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9933000206947327,"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"}},"topics":[{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9933000206947327,"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"}},{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9930999875068665,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9908999800682068,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6472399234771729},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.6305602788925171},{"id":"https://openalex.org/keywords/management-training","display_name":"Management training","score":0.4132128357887268},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.32076790928840637},{"id":"https://openalex.org/keywords/management","display_name":"Management","score":0.0852046012878418}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6472399234771729},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.6305602788925171},{"id":"https://openalex.org/C2988483808","wikidata":"https://www.wikidata.org/wiki/Q191701","display_name":"Management training","level":2,"score":0.4132128357887268},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.32076790928840637},{"id":"https://openalex.org/C187736073","wikidata":"https://www.wikidata.org/wiki/Q2920921","display_name":"Management","level":1,"score":0.0852046012878418},{"id":"https://openalex.org/C153294291","wikidata":"https://www.wikidata.org/wiki/Q25261","display_name":"Meteorology","level":1,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"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":1,"locations":[{"id":"doi:10.1145/3721146.3721955","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3721146.3721955","pdf_url":null,"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 5th Workshop on Machine Learning and Systems","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3721146.3721955","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3721146.3721955","pdf_url":null,"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 5th Workshop on Machine Learning and Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":21,"referenced_works":["https://openalex.org/W2194775991","https://openalex.org/W2618530766","https://openalex.org/W2912918068","https://openalex.org/W2949486320","https://openalex.org/W3012479151","https://openalex.org/W3012514909","https://openalex.org/W3042369493","https://openalex.org/W3086105743","https://openalex.org/W3173778203","https://openalex.org/W3186076368","https://openalex.org/W3204998121","https://openalex.org/W4214753036","https://openalex.org/W4242874873","https://openalex.org/W4295312788","https://openalex.org/W4301361180","https://openalex.org/W4318541551","https://openalex.org/W4327911463","https://openalex.org/W4360897687","https://openalex.org/W4386768656","https://openalex.org/W4393407138","https://openalex.org/W4408847715"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2018366657","https://openalex.org/W2330966084","https://openalex.org/W3178569862","https://openalex.org/W2372932417","https://openalex.org/W604613010","https://openalex.org/W591768373","https://openalex.org/W4404261161"],"abstract_inverted_index":{"This":[0],"paper":[1],"analyzes":[2],"the":[3,27,43,47,52,80,115,126,147,179],"performance":[4,83,108,116,166],"impact":[5,117],"of":[6,32,82,118,181],"unified":[7],"virtual":[8],"memory":[9,23,77,89,153,186],"(UVM)":[10],"while":[11,134],"running":[12],"large":[13,138],"deep":[14],"learning":[15],"(DL)":[16],"workloads":[17,192],"on":[18,61,146,193],"a":[19],"GPU":[20,45,141],"with":[21,88,184],"limited":[22],"capacity.":[24],"Due":[25],"to":[26,106,123,189],"page":[28,162],"fault":[29,163],"handling":[30],"overhead":[31],"UVM,":[33],"DL":[34,54,66,103,121,191],"frameworks":[35],"have":[36],"not":[37,98],"officially":[38],"supported":[39],"UVM.":[40],"However,":[41,95],"given":[42],"rising":[44],"price,":[46],"global":[48],"semiconductor":[49],"shortages,":[50],"and":[51,74,84,128,152],"ever-increasing":[53],"model":[55],"size,":[56],"we":[57,113],"cannot":[58],"optimistically":[59],"rely":[60],"adding":[62],"more":[63],"GPUs":[64],"for":[65,120],"processing.":[67],"Various":[68],"other":[69],"solutions":[70],"such":[71,93],"as":[72],"compression":[73],"quantization":[75],"reduce":[76],"footprint":[78],"at":[79,167],"cost":[81],"accuracy.":[85],"Using":[86],"UVM":[87,96,119,135,183],"oversubscription":[90,150,169],"could":[91],"tackle":[92],"concerns.":[94],"has":[97],"been":[99],"actively":[100],"leveraged":[101],"in":[102],"computing":[104,122],"due":[105],"its":[107,143],"overhead.":[109],"In":[110],"this":[111],"study,":[112],"investigate":[114],"better":[124],"understand":[125],"benefits":[127],"limitations.":[129],"Our":[130],"results":[131],"show":[132],"that":[133,158],"enables":[136],"training":[137],"models":[139],"beyond":[140],"capacity,":[142],"effectiveness":[144],"depends":[145],"interplay":[148],"between":[149],"factors":[151],"management":[154,187],"strategies.":[155],"We":[156],"find":[157],"PCA":[159],"significantly":[160],"mitigates":[161],"overhead,":[164],"improving":[165],"moderate":[168],"levels,":[170],"but":[171],"also":[172],"increases":[173],"migration":[174],"traffic.":[175],"These":[176],"findings":[177],"highlight":[178],"potential":[180],"integrating":[182],"advanced":[185],"strategies":[188],"optimize":[190],"limited-memory":[194],"GPUs.":[195]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":3}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}