{"id":"https://openalex.org/W2753100145","doi":"https://doi.org/10.1147/jrd.2017.2716598","title":"Optimizing the efficiency of deep learning through accelerator virtualization","display_name":"Optimizing the efficiency of deep learning through accelerator virtualization","publication_year":2017,"publication_date":"2017-07-01","ids":{"openalex":"https://openalex.org/W2753100145","doi":"https://doi.org/10.1147/jrd.2017.2716598","mag":"2753100145"},"language":"en","primary_location":{"id":"doi:10.1147/jrd.2017.2716598","is_oa":false,"landing_page_url":"https://doi.org/10.1147/jrd.2017.2716598","pdf_url":null,"source":{"id":"https://openalex.org/S4210219925","display_name":"IBM Journal of Research and Development","issn_l":"0018-8646","issn":["0018-8646","2151-8556"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320652","host_organization_name":"IBM","host_organization_lineage":["https://openalex.org/P4310320652"],"host_organization_lineage_names":["IBM"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IBM Journal of Research and Development","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/A5049413685","display_name":"Michael Gschwind","orcid":"https://orcid.org/0009-0001-4963-4915"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"M. Gschwind","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058607529","display_name":"Tim Kaldewey","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"T. Kaldewey","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5071385812","display_name":"David Tam","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"D. K. Tam","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5049413685"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.4621,"has_fulltext":false,"cited_by_count":11,"citation_normalized_percentile":{"value":0.7325217,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":99},"biblio":{"volume":"61","issue":"4/5","first_page":"12:1","last_page":"12:11"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998000264167786,"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":0.9998000264167786,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9994000196456909,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9984999895095825,"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":0.8557919859886169},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.7891348600387573},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.6769100427627563},{"id":"https://openalex.org/keywords/virtualization","display_name":"Virtualization","score":0.652883768081665},{"id":"https://openalex.org/keywords/supercomputer","display_name":"Supercomputer","score":0.6014156937599182},{"id":"https://openalex.org/keywords/graphics-processing-unit","display_name":"Graphics processing unit","score":0.5826060771942139},{"id":"https://openalex.org/keywords/graphics","display_name":"Graphics","score":0.5019807815551758},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.48005589842796326},{"id":"https://openalex.org/keywords/cloud-computing","display_name":"Cloud computing","score":0.46730944514274597},{"id":"https://openalex.org/keywords/general-purpose-computing-on-graphics-processing-units","display_name":"General-purpose computing on graphics processing units","score":0.4506247341632843},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.4237903356552124},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.41928520798683167},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.332008957862854},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.3119809627532959}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8557919859886169},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.7891348600387573},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6769100427627563},{"id":"https://openalex.org/C513985346","wikidata":"https://www.wikidata.org/wiki/Q270471","display_name":"Virtualization","level":3,"score":0.652883768081665},{"id":"https://openalex.org/C83283714","wikidata":"https://www.wikidata.org/wiki/Q121117","display_name":"Supercomputer","level":2,"score":0.6014156937599182},{"id":"https://openalex.org/C2779851693","wikidata":"https://www.wikidata.org/wiki/Q183484","display_name":"Graphics processing unit","level":2,"score":0.5826060771942139},{"id":"https://openalex.org/C21442007","wikidata":"https://www.wikidata.org/wiki/Q1027879","display_name":"Graphics","level":2,"score":0.5019807815551758},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.48005589842796326},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.46730944514274597},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.4506247341632843},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.4237903356552124},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.41928520798683167},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.332008957862854},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.3119809627532959}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1147/jrd.2017.2716598","is_oa":false,"landing_page_url":"https://doi.org/10.1147/jrd.2017.2716598","pdf_url":null,"source":{"id":"https://openalex.org/S4210219925","display_name":"IBM Journal of Research and Development","issn_l":"0018-8646","issn":["0018-8646","2151-8556"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320652","host_organization_name":"IBM","host_organization_lineage":["https://openalex.org/P4310320652"],"host_organization_lineage_names":["IBM"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IBM Journal of Research and Development","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9","score":0.47999998927116394}],"awards":[],"funders":[{"id":"https://openalex.org/F4320309370","display_name":"University of Pennsylvania","ror":"https://ror.org/00b30xv10"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":38,"referenced_works":["https://openalex.org/W315953870","https://openalex.org/W753012316","https://openalex.org/W1450125063","https://openalex.org/W1563686443","https://openalex.org/W1667652561","https://openalex.org/W1789336918","https://openalex.org/W1841592590","https://openalex.org/W1966838590","https://openalex.org/W1972399950","https://openalex.org/W2051598123","https://openalex.org/W2076063813","https://openalex.org/W2108591264","https://openalex.org/W2117625898","https://openalex.org/W2120920992","https://openalex.org/W2155893237","https://openalex.org/W2186615578","https://openalex.org/W2216599197","https://openalex.org/W2235191389","https://openalex.org/W2271840356","https://openalex.org/W2415044472","https://openalex.org/W2467163293","https://openalex.org/W2504282214","https://openalex.org/W2753217274","https://openalex.org/W2963374099","https://openalex.org/W4302296459","https://openalex.org/W6610993387","https://openalex.org/W6622239757","https://openalex.org/W6628516955","https://openalex.org/W6633628269","https://openalex.org/W6637151318","https://openalex.org/W6638020065","https://openalex.org/W6638783484","https://openalex.org/W6676715614","https://openalex.org/W6677533768","https://openalex.org/W6686509673","https://openalex.org/W6694517276","https://openalex.org/W6715976941","https://openalex.org/W6719565411"],"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/W3104348697"],"abstract_inverted_index":{"Training":[0],"deep":[1,69,100,117,163,200],"learning":[2,70,101,118,164,201],"models":[3],"often":[4],"occupies":[5],"entire":[6],"compute":[7,45,204],"clusters,":[8],"built":[9],"solely":[10],"for":[11,14,31,67],"this":[12],"purpose,":[13],"days":[15],"or":[16],"even":[17],"weeks":[18],"at":[19],"a":[20,24,64,96,122,138,144,150],"time.":[21],"There":[22],"exists":[23],"large":[25],"body":[26],"of":[27,47,89,99,110,124,140,178],"work":[28],"on":[29],"approaches":[30],"improving":[32],"training":[33],"performance,":[34],"ranging":[35],"from":[36,157],"novel":[37],"algorithms":[38],"to":[39,77,83,103,190,195],"full":[40],"custom":[41],"hardware":[42,209],"accelerators.":[43],"Offering":[44],"capabilities":[46],"multiple":[48],"teraflops":[49],"(trillion":[50],"floating":[51],"point":[52],"operations":[53],"per":[54],"second),":[55],"graphics":[56],"processing":[57],"units":[58],"(GPUs)":[59],"have":[60],"established":[61],"themselves":[62],"as":[63],"de-facto":[65],"standard":[66],"accelerating":[68],"network":[71],"training.":[72],"As":[73],"systems":[74],"with":[75],"up":[76,82],"16":[78],"GPUs\u2014each":[79],"GPU":[80,111,125,132,186],"consuming":[81],"300":[84],"W\u2014become":[85],"available,":[86],"efficient":[87],"usage":[88],"these":[90],"resources":[91,133],"becomes":[92],"imperative.":[93],"We":[94,113,181],"conduct":[95],"detailed":[97],"analysis":[98],"workloads":[102,119,156],"characterize":[104],"their":[105],"efficiency":[106],"in":[107,199],"making":[108,207],"use":[109],"acceleration.":[112],"found":[114],"that":[115,161],"many":[116],"consume":[120],"only":[121,189],"fraction":[123],"resources,":[126],"and":[127,170,176],"we":[128],"demonstrate":[129],"how":[130],"sharing":[131],"can":[134],"improve":[135],"throughput":[136],"by":[137,202],"factor":[139],"3,":[141],"effectively":[142],"turning":[143],"4-GPU":[145],"commodity":[146],"cloud":[147],"system":[148],"into":[149],"high-end":[151],"12-GPU":[152],"supercomputer.":[153],"Using":[154],"Watson":[155],"three":[158],"major":[159],"areas":[160],"incorporate":[162],"technology\u2014i.e.,":[165],"language":[166],"classification,":[167],"visual":[168],"recognition,":[169],"speech":[171],"recognition\u2014we":[172],"document":[173],"the":[174],"effectiveness":[175],"scalability":[177],"our":[179],"approach.":[180],"are":[182],"working":[183],"toward":[184],"enabling":[185],"virtualization":[187],"not":[188],"reduce":[191],"cost,":[192],"but":[193],"also":[194],"accelerate":[196],"new":[197],"breakthroughs":[198],"increasing":[203],"capacity":[205],"without":[206],"further":[208],"investments.":[210]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2023,"cited_by_count":1},{"year":2021,"cited_by_count":3},{"year":2020,"cited_by_count":2},{"year":2018,"cited_by_count":3}],"updated_date":"2026-04-10T15:06:20.359241","created_date":"2025-10-10T00:00:00"}