{"id":"https://openalex.org/W2796440709","doi":"https://doi.org/10.1145/3184407.3184424","title":"Involving CPUs into Multi-GPU Deep Learning","display_name":"Involving CPUs into Multi-GPU Deep Learning","publication_year":2018,"publication_date":"2018-03-30","ids":{"openalex":"https://openalex.org/W2796440709","doi":"https://doi.org/10.1145/3184407.3184424","mag":"2796440709"},"language":"en","primary_location":{"id":"doi:10.1145/3184407.3184424","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3184407.3184424","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2018 ACM/SPEC International Conference on Performance Engineering","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/A5089550246","display_name":"Tung D. Le","orcid":null},"institutions":[{"id":"https://openalex.org/I4210145865","display_name":"IBM Research - Tokyo","ror":"https://ror.org/04915qk43","country_code":"JP","type":"facility","lineage":["https://openalex.org/I1341412227","https://openalex.org/I4210114115","https://openalex.org/I4210145865"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Tung D. Le","raw_affiliation_strings":["IBM Research - Tokyo, Tokyo, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research - Tokyo, Tokyo, Japan","institution_ids":["https://openalex.org/I4210145865"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5018002187","display_name":"Taro Sekiyama","orcid":"https://orcid.org/0000-0001-9286-230X"},"institutions":[{"id":"https://openalex.org/I4210145865","display_name":"IBM Research - Tokyo","ror":"https://ror.org/04915qk43","country_code":"JP","type":"facility","lineage":["https://openalex.org/I1341412227","https://openalex.org/I4210114115","https://openalex.org/I4210145865"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Taro Sekiyama","raw_affiliation_strings":["IBM Research - Tokyo, Tokyo, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research - Tokyo, Tokyo, Japan","institution_ids":["https://openalex.org/I4210145865"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090981513","display_name":"Yasushi Negishi","orcid":null},"institutions":[{"id":"https://openalex.org/I4210145865","display_name":"IBM Research - Tokyo","ror":"https://ror.org/04915qk43","country_code":"JP","type":"facility","lineage":["https://openalex.org/I1341412227","https://openalex.org/I4210114115","https://openalex.org/I4210145865"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Yasushi Negishi","raw_affiliation_strings":["IBM Research - Tokyo, Tokyo, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research - Tokyo, Tokyo, Japan","institution_ids":["https://openalex.org/I4210145865"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102286597","display_name":"Haruki Imai","orcid":null},"institutions":[{"id":"https://openalex.org/I4210145865","display_name":"IBM Research - Tokyo","ror":"https://ror.org/04915qk43","country_code":"JP","type":"facility","lineage":["https://openalex.org/I1341412227","https://openalex.org/I4210114115","https://openalex.org/I4210145865"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Haruki Imai","raw_affiliation_strings":["IBM Research - Tokyo, Tokyo, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research - Tokyo, Tokyo, Japan","institution_ids":["https://openalex.org/I4210145865"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5063074861","display_name":"Kiyokuni Kawachiya","orcid":null},"institutions":[{"id":"https://openalex.org/I4210145865","display_name":"IBM Research - Tokyo","ror":"https://ror.org/04915qk43","country_code":"JP","type":"facility","lineage":["https://openalex.org/I1341412227","https://openalex.org/I4210114115","https://openalex.org/I4210145865"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Kiyokuni Kawachiya","raw_affiliation_strings":["IBM Research - Tokyo, Tokyo, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research - Tokyo, Tokyo, Japan","institution_ids":["https://openalex.org/I4210145865"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.742,"has_fulltext":false,"cited_by_count":10,"citation_normalized_percentile":{"value":0.7636153,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"56","last_page":"67"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":1.0,"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":1.0,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9990000128746033,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.9972000122070312,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.9015799164772034},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.7677264213562012},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.6760545372962952},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.6702257990837097},{"id":"https://openalex.org/keywords/data-parallelism","display_name":"Data parallelism","score":0.6194348335266113},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5646408200263977},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.5123842358589172},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.508311927318573},{"id":"https://openalex.org/keywords/central-processing-unit","display_name":"Central processing unit","score":0.4902384877204895},{"id":"https://openalex.org/keywords/deep-neural-networks","display_name":"Deep neural networks","score":0.48330190777778625},{"id":"https://openalex.org/keywords/gpu-cluster","display_name":"GPU cluster","score":0.47917094826698303},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.43799182772636414},{"id":"https://openalex.org/keywords/parallelism","display_name":"Parallelism (grammar)","score":0.3146522045135498},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.1639275848865509}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.9015799164772034},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.7677264213562012},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.6760545372962952},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.6702257990837097},{"id":"https://openalex.org/C61483411","wikidata":"https://www.wikidata.org/wiki/Q3124522","display_name":"Data parallelism","level":3,"score":0.6194348335266113},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5646408200263977},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5123842358589172},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.508311927318573},{"id":"https://openalex.org/C49154492","wikidata":"https://www.wikidata.org/wiki/Q5300","display_name":"Central processing unit","level":2,"score":0.4902384877204895},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.48330190777778625},{"id":"https://openalex.org/C2781335571","wikidata":"https://www.wikidata.org/wiki/Q2633544","display_name":"GPU cluster","level":3,"score":0.47917094826698303},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43799182772636414},{"id":"https://openalex.org/C2781172179","wikidata":"https://www.wikidata.org/wiki/Q853109","display_name":"Parallelism (grammar)","level":2,"score":0.3146522045135498},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.1639275848865509},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3184407.3184424","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3184407.3184424","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2018 ACM/SPEC International Conference on Performance Engineering","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":23,"referenced_works":["https://openalex.org/W753012316","https://openalex.org/W1598866093","https://openalex.org/W1686810756","https://openalex.org/W2097117768","https://openalex.org/W2117539524","https://openalex.org/W2132737349","https://openalex.org/W2155893237","https://openalex.org/W2162390675","https://openalex.org/W2168231600","https://openalex.org/W2181607856","https://openalex.org/W2186615578","https://openalex.org/W2204075824","https://openalex.org/W2265775419","https://openalex.org/W2302255633","https://openalex.org/W2557283755","https://openalex.org/W2618530766","https://openalex.org/W2949608135","https://openalex.org/W2949650786","https://openalex.org/W2949888546","https://openalex.org/W2950789693","https://openalex.org/W2952926545","https://openalex.org/W2962950660","https://openalex.org/W2963266252"],"related_works":["https://openalex.org/W2950520577","https://openalex.org/W1554644772","https://openalex.org/W2056717482","https://openalex.org/W2003935582","https://openalex.org/W2494130044","https://openalex.org/W3170887803","https://openalex.org/W74409296","https://openalex.org/W3209384898","https://openalex.org/W4400951174","https://openalex.org/W1595834484"],"abstract_inverted_index":{"The":[0,177],"most":[1],"important":[2],"part":[3],"of":[4,15,19,56,114,161,201],"deep":[5,34,183,192],"learning,":[6],"training":[7,33,79,85,98,120,157],"the":[8,13,54,66,92,97,100,112,116,122,128,140,145,175,188],"neural":[9,35,184,210],"network,":[10],"often":[11],"requires":[12],"processing":[14],"a":[16,41,73,105,165,171],"large":[17],"amount":[18],"data":[20,57,77,82,118,124,155],"and":[21,62,110,121,143,163,205,215],"can":[22],"takes":[23],"days":[24],"to":[25,45,76,94],"complete.":[26],"Data":[27],"parallelism":[28],"is":[29,51,137,179],"widely":[30],"used":[31],"for":[32,60,108,174,181,207,226],"networks":[36,228],"on":[37,91],"multiple":[38],"GPUs":[39],"in":[40,99,191],"single":[42],"machine":[43],"thanks":[44],"its":[46,49],"simplicity.":[47],"However,":[48],"scalability":[50],"bound":[52],"by":[53,158],"number":[55],"transfers,":[58],"mainly":[59],"exchanging":[61],"accumulating":[63],"gradients":[64],"among":[65],"GPUs.":[67,101,231],"In":[68],"this":[69],"paper,":[70],"we":[71,131,149,152,198],"present":[72,104,164],"novel":[74],"approach":[75,136],"parallel":[78,83,119,125,156],"called":[80],"CPU-GPU":[81,123,154],"(CGDP)":[84],"that":[86,168,197],"utilizes":[87],"free":[88],"CPU":[89],"time":[90],"host":[93],"speed":[95],"up":[96],"We":[102],"also":[103],"cost":[106,129],"model":[107],"analyzing":[109],"comparing":[111],"performances":[113],"both":[115],"typical":[117,141],"training.":[126,176],"Using":[127],"model,":[130],"formally":[132],"show":[133],"why":[134],"our":[135],"better":[138],"than":[139,222],"one":[142],"clarify":[144],"remaining":[146],"issues.":[147],"Finally,":[148],"explain":[150],"how":[151],"optimized":[153],"introducing":[159],"chunks":[160],"layers":[162],"runtime":[166],"algorithm":[167,178],"automatically":[169],"finds":[170],"good":[172],"configuration":[173],"effective":[180],"very":[182],"networks,":[185],"which":[186],"are":[187],"current":[189],"trend":[190],"learning.":[193],"Experimental":[194],"results":[195],"showed":[196],"achieved":[199,225],"speedups":[200],"$1.21$,":[202],"$1.04$,":[203],"$1.21$":[204],"$1.07$":[206],"four":[208,230],"state-of-the-art":[209],"networks:":[211],"AlexNet,":[212],"GoogLeNet-v1,":[213],"VGGNet-16,":[214],"ResNet-152,":[216],"respectively.":[217],"Weak":[218],"scaling":[219],"efficiency":[220],"greater":[221],"$90$":[223],"was":[224],"all":[227],"across":[229]},"counts_by_year":[{"year":2023,"cited_by_count":1},{"year":2022,"cited_by_count":2},{"year":2021,"cited_by_count":4},{"year":2020,"cited_by_count":2},{"year":2019,"cited_by_count":1}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}