{"id":"https://openalex.org/W4286233789","doi":"https://doi.org/10.48550/arxiv.2207.07886","title":"An Experimental Evaluation of Machine Learning Training on a Real Processing-in-Memory System","display_name":"An Experimental Evaluation of Machine Learning Training on a Real Processing-in-Memory System","publication_year":2022,"publication_date":"2022-07-16","ids":{"openalex":"https://openalex.org/W4286233789","doi":"https://doi.org/10.48550/arxiv.2207.07886"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2207.07886","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2207.07886","pdf_url":"https://arxiv.org/pdf/2207.07886","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"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":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2207.07886","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5044416322","display_name":"Juan G\u00f3mez-Luna","orcid":"https://orcid.org/0000-0002-6514-1571","observed_orcids":["https://orcid.org/0000-0002-6514-1571"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"G\u00f3mez-Luna, Juan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101501427","display_name":"Yuxin Guo","orcid":"https://orcid.org/0009-0007-0169-3864","observed_orcids":["https://orcid.org/0009-0007-0169-3864"]},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Guo, Yuxin","raw_affiliation_strings":["ETH Z\u00fcrich"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Z\u00fcrich","institution_ids":["https://openalex.org/I35440088"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023891765","display_name":"Sylvan Brocard","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Brocard, Sylvan","raw_affiliation_strings":["UPMEM"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UPMEM","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5091278491","display_name":"Julien Legriel","orcid":"https://orcid.org/0009-0003-6183-6004","observed_orcids":["https://orcid.org/0009-0003-6183-6004"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Legriel, Julien","raw_affiliation_strings":["UPMEM"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UPMEM","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014065776","display_name":"Remy Cimadomo","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cimadomo, Remy","raw_affiliation_strings":["UPMEM"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UPMEM","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5091016237","display_name":"Geraldo Francisco de Oliveira Junior","orcid":"https://orcid.org/0000-0003-1557-4819","observed_orcids":["https://orcid.org/0000-0003-1557-4819"]},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Oliveira, Geraldo F.","raw_affiliation_strings":["ETH Z\u00fcrich"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Z\u00fcrich","institution_ids":["https://openalex.org/I35440088"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014657381","display_name":"Gagandeep Singh","orcid":"https://orcid.org/0000-0002-3502-7401","observed_orcids":["https://orcid.org/0000-0002-3502-7401"]},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Singh, Gagandeep","raw_affiliation_strings":["ETH Z\u00fcrich"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Z\u00fcrich","institution_ids":["https://openalex.org/I35440088"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5050695684","display_name":"Onur Mutlu","orcid":"https://orcid.org/0000-0002-0075-2312","observed_orcids":["https://orcid.org/0000-0002-0075-2312"]},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Mutlu, Onur","raw_affiliation_strings":["ETH Z\u00fcrich"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Z\u00fcrich","institution_ids":["https://openalex.org/I35440088"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":12,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9634000062942505,"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":0.9634000062942505,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.025200000032782555,"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/T12326","display_name":"Network Packet Processing and Optimization","score":0.0020000000949949026,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/in-memory-computing","display_name":"in-memory computing","score":0.9879999756813049},{"id":"https://openalex.org/keywords/machine-learning-model-training","display_name":"machine learning training","score":0.8180000185966492},{"id":"https://openalex.org/keywords/memory-centric-computing","display_name":"memory-centric computing","score":0.9900000095367432},{"id":"https://openalex.org/keywords/data-movement-bottleneck","display_name":"data movement bottleneck","score":0.7860000133514404},{"id":"https://openalex.org/keywords/memory-bound-workloads","display_name":"memory-bound workloads","score":0.7450000047683716},{"id":"https://openalex.org/keywords/decision-tree","display_name":"decision tree","score":0.7179999947547913},{"id":"https://openalex.org/keywords/k-means-clustering","display_name":"K-means clustering","score":0.9449999928474426},{"id":"https://openalex.org/keywords/linear-regression","display_name":"linear regression","score":0.9319999814033508},{"id":"https://openalex.org/keywords/logistic-regression","display_name":"logistic regression","score":0.9980000257492065},{"id":"https://openalex.org/keywords/cpu-vs-gpu-comparison","display_name":"CPU vs GPU comparison","score":0.6759999990463257}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8623872399330139},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.6153099536895752},{"id":"https://openalex.org/C145108525","wikidata":"https://www.wikidata.org/wiki/Q656154","display_name":"Xeon","level":2,"score":0.5860151052474976},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.583410382270813},{"id":"https://openalex.org/C73555534","wikidata":"https://www.wikidata.org/wiki/Q622825","display_name":"Cluster analysis","level":2,"score":0.4949187636375427},{"id":"https://openalex.org/C49154492","wikidata":"https://www.wikidata.org/wiki/Q5300","display_name":"Central processing unit","level":2,"score":0.4898342788219452},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3260272741317749},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.29481735825538635},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.2518250346183777},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.25104689598083496}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2207.07886","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2207.07886","pdf_url":"https://arxiv.org/pdf/2207.07886","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"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":"text"},{"id":"doi:10.48550/arxiv.2207.07886","doi":"https://doi.org/10.48550/arxiv.2207.07886","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2207.07886","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2207.07886","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2207.07886","pdf_url":"https://arxiv.org/pdf/2207.07886","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"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":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G2342797219","display_name":null,"funder_award_id":"101047160","funder_id":"https://openalex.org/F4320320300","funder_display_name":"European Commission"}],"funders":[{"id":"https://openalex.org/F4320306087","display_name":"Semiconductor Research Corporation","ror":"https://ror.org/047z4n946"},{"id":"https://openalex.org/F4320316785","display_name":"VMware","ror":"https://ror.org/01mwmh458"},{"id":"https://openalex.org/F4320320300","display_name":"European Commission","ror":"https://ror.org/00k4n6c32"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4286233789.pdf","grobid_xml":"https://content.openalex.org/works/W4286233789.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4381894570","https://openalex.org/W4312438982","https://openalex.org/W4403334538","https://openalex.org/W4416303630","https://openalex.org/W3178435800","https://openalex.org/W3202917172","https://openalex.org/W4406276690","https://openalex.org/W7126335040","https://openalex.org/W2965339519","https://openalex.org/W6926488320"],"abstract_inverted_index":{"Training":[0],"machine":[1],"learning":[2],"(ML)":[3],"algorithms":[4,89],"is":[5,11,65,174,204,220],"a":[6,22,100,132,178,191,229],"computationally":[7],"intensive":[8],"process,":[9],"which":[10,40],"frequently":[12],"memory-bound":[13,151],"due":[14],"to":[15,66,75,121,224],"repeatedly":[16],"accessing":[17],"large":[18,42],"training":[19,227],"datasets.":[20],"As":[21],"result,":[23],"processor-centric":[24],"systems":[25],"(e.g.,":[26],"CPU,":[27],"GPU)":[28],"suffer":[29],"from":[30],"costly":[31],"data":[32,60],"movement":[33,61],"between":[34],"memory":[35],"units":[36],"and":[37,46,108,116,118,127,158,187,206,211,239,252],"processing":[38],"units,":[39],"consumes":[41],"amounts":[43],"of":[44,70,113,171,245,249,257],"energy":[45],"execution":[47],"cycles.":[48],"Memory-centric":[49],"computing":[50,135,260],"systems,":[51],"i.e.,":[52],"with":[53,137,235],"processing-in-memory":[54],"(PIM)":[55],"capabilities,":[56],"can":[57,148,242],"alleviate":[58],"this":[59],"bottleneck.":[62],"Our":[63,129,199],"goal":[64],"understand":[67],"the":[68,155,221],"potential":[69],"modern":[71],"general-purpose":[72,102,145],"PIM":[73,103,141,146,164,169,203,231,250],"architectures":[74,147],"accelerate":[76,150],"ML":[77,88,152,226,246],"training.":[78],"To":[79,215],"do":[80],"so,":[81],"we":[82],"(1)":[83],"implement":[84],"several":[85],"representative":[86],"classic":[87],"(namely,":[90],"linear":[91],"regression,":[92,94],"logistic":[93],"decision":[95,172],"tree,":[96],"K-Means":[97,200],"clustering)":[98],"on":[99,125,131,182,195,202,228],"real-world":[101,230],"architecture,":[104],"(2)":[105],"rigorously":[106],"evaluate":[107,225],"characterize":[109],"them":[110],"in":[111],"terms":[112],"accuracy,":[114],"performance":[115],"scaling,":[117],"(3)":[119],"compare":[120],"their":[122],"counterpart":[123],"implementations":[124],"CPU":[126,180,210],"GPU.":[128],"evaluation":[130],"real":[133],"memory-centric":[134,259],"system":[136],"more":[138],"than":[139,177,190,208],"2500":[140],"cores":[142],"shows":[143],"that":[144,241],"greatly":[149],"workloads,":[153,247],"when":[154],"necessary":[156],"operations":[157],"datatypes":[159],"are":[160],"natively":[161],"supported":[162],"by":[163],"hardware.":[165],"For":[166],"example,":[167],"our":[168,216,218],"implementation":[170],"tree":[173],"$27\\times$":[175],"faster":[176,189],"state-of-the-art":[179,192,209],"version":[181,194],"an":[183,196],"8-core":[184],"Intel":[185],"Xeon,":[186],"$1.34\\times$":[188],"GPU":[193,212],"NVIDIA":[197],"A100.":[198],"clustering":[201],"$2.8\\times$":[205],"$3.2\\times$":[207],"versions,":[213],"respectively.":[214],"knowledge,":[217],"work":[219],"first":[222],"one":[223],"architecture.":[232],"We":[233],"conclude":[234],"key":[236],"observations,":[237],"takeaways,":[238],"recommendations":[240],"inspire":[243],"users":[244],"programmers":[248],"architectures,":[251],"hardware":[253],"designers":[254],"&":[255],"architects":[256],"future":[258],"systems.":[261]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":2},{"year":2024,"cited_by_count":6},{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":1}],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2022-07-21T00:00:00"}