{"id":"https://openalex.org/W4387302734","doi":"https://doi.org/10.1145/3600006.3613142","title":"Bagpipe: Accelerating Deep Recommendation Model Training","display_name":"Bagpipe: Accelerating Deep Recommendation Model Training","publication_year":2023,"publication_date":"2023-10-03","ids":{"openalex":"https://openalex.org/W4387302734","doi":"https://doi.org/10.1145/3600006.3613142"},"language":"en","primary_location":{"id":"doi:10.1145/3600006.3613142","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3600006.3613142","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 29th Symposium on Operating Systems Principles","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/A5101739283","display_name":"Saurabh Agarwal","orcid":"https://orcid.org/0009-0009-4880-0289"},"institutions":[{"id":"https://openalex.org/I135310074","display_name":"University of Wisconsin\u2013Madison","ror":"https://ror.org/01y2jtd41","country_code":"US","type":"education","lineage":["https://openalex.org/I135310074"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Saurabh Agarwal","raw_affiliation_strings":["Department of Computer Science, University of Wisconsin-Madison, Madison, WI, United States of America"],"raw_orcid":"https://orcid.org/0009-0009-4880-0289","affiliations":[{"raw_affiliation_string":"Department of Computer Science, University of Wisconsin-Madison, Madison, WI, United States of America","institution_ids":["https://openalex.org/I135310074"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102932966","display_name":"Chengpo Yan","orcid":"https://orcid.org/0009-0009-2618-2387"},"institutions":[{"id":"https://openalex.org/I135310074","display_name":"University of Wisconsin\u2013Madison","ror":"https://ror.org/01y2jtd41","country_code":"US","type":"education","lineage":["https://openalex.org/I135310074"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chengpo Yan","raw_affiliation_strings":["Department of Computer Science, University of Wisconsin-Madison, Madison, WI, USA"],"raw_orcid":"https://orcid.org/0009-0009-2618-2387","affiliations":[{"raw_affiliation_string":"Department of Computer Science, University of Wisconsin-Madison, Madison, WI, USA","institution_ids":["https://openalex.org/I135310074"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027735922","display_name":"Ziyi Zhang","orcid":"https://orcid.org/0009-0004-4336-6341"},"institutions":[{"id":"https://openalex.org/I40347166","display_name":"University of Chicago","ror":"https://ror.org/024mw5h28","country_code":"US","type":"education","lineage":["https://openalex.org/I40347166"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ziyi Zhang","raw_affiliation_strings":["Department of Computer Science, University of Chicago, Chicago, IL, USA"],"raw_orcid":"https://orcid.org/0009-0004-4336-6341","affiliations":[{"raw_affiliation_string":"Department of Computer Science, University of Chicago, Chicago, IL, USA","institution_ids":["https://openalex.org/I40347166"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5082710354","display_name":"Shivaram Venkataraman","orcid":"https://orcid.org/0000-0001-9575-7935"},"institutions":[{"id":"https://openalex.org/I135310074","display_name":"University of Wisconsin\u2013Madison","ror":"https://ror.org/01y2jtd41","country_code":"US","type":"education","lineage":["https://openalex.org/I135310074"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shivaram Venkataraman","raw_affiliation_strings":["Department of Computer Science, University of Wisconsin-Madison, Madison, WI, United States of America"],"raw_orcid":"https://orcid.org/0000-0001-9575-7935","affiliations":[{"raw_affiliation_string":"Department of Computer Science, University of Wisconsin-Madison, Madison, WI, United States of America","institution_ids":["https://openalex.org/I135310074"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":19,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"348","last_page":"363"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11478","display_name":"Caching and Content Delivery","score":0.9998999834060669,"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"}},"topics":[{"id":"https://openalex.org/T11478","display_name":"Caching and Content Delivery","score":0.9998999834060669,"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"}},{"id":"https://openalex.org/T10203","display_name":"Recommender Systems and Techniques","score":0.9994000196456909,"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/T11273","display_name":"Advanced Graph Neural Networks","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":0.8794605731964111},{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.753995954990387},{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.5132310390472412},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.5111215710639954},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5055184364318848},{"id":"https://openalex.org/keywords/oracle","display_name":"Oracle","score":0.4957519471645355},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.4889303743839264},{"id":"https://openalex.org/keywords/synchronization","display_name":"Synchronization (alternating current)","score":0.47318506240844727},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.47132399678230286},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3430188298225403},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.27911102771759033},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.15249288082122803},{"id":"https://openalex.org/keywords/software-engineering","display_name":"Software engineering","score":0.08983269333839417},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.08689787983894348}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8794605731964111},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.753995954990387},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.5132310390472412},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.5111215710639954},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5055184364318848},{"id":"https://openalex.org/C55166926","wikidata":"https://www.wikidata.org/wiki/Q2892946","display_name":"Oracle","level":2,"score":0.4957519471645355},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.4889303743839264},{"id":"https://openalex.org/C2778562939","wikidata":"https://www.wikidata.org/wiki/Q1298791","display_name":"Synchronization (alternating current)","level":3,"score":0.47318506240844727},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.47132399678230286},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3430188298225403},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.27911102771759033},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.15249288082122803},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.08983269333839417},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.08689787983894348},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3600006.3613142","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3600006.3613142","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 29th Symposium on Operating Systems Principles","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":52,"referenced_works":["https://openalex.org/W95608104","https://openalex.org/W1987225815","https://openalex.org/W2037959956","https://openalex.org/W2048657872","https://openalex.org/W2060393849","https://openalex.org/W2131613942","https://openalex.org/W2137028279","https://openalex.org/W2146591355","https://openalex.org/W2157323434","https://openalex.org/W2343644403","https://openalex.org/W2424347275","https://openalex.org/W2430301697","https://openalex.org/W2469952266","https://openalex.org/W2475334473","https://openalex.org/W2512971201","https://openalex.org/W2513020047","https://openalex.org/W2593507512","https://openalex.org/W2604438604","https://openalex.org/W2605350416","https://openalex.org/W2742306177","https://openalex.org/W2745534091","https://openalex.org/W2785570910","https://openalex.org/W2808859004","https://openalex.org/W2809112621","https://openalex.org/W2884700152","https://openalex.org/W2901299405","https://openalex.org/W2914209329","https://openalex.org/W2964182926","https://openalex.org/W2969388332","https://openalex.org/W2979245724","https://openalex.org/W3016842236","https://openalex.org/W3037847693","https://openalex.org/W3043433718","https://openalex.org/W3044211235","https://openalex.org/W3094431459","https://openalex.org/W3096484587","https://openalex.org/W3121395088","https://openalex.org/W3146147323","https://openalex.org/W3158146252","https://openalex.org/W3158296418","https://openalex.org/W3164760399","https://openalex.org/W3166914430","https://openalex.org/W3167625290","https://openalex.org/W3169198323","https://openalex.org/W3200211247","https://openalex.org/W3214762859","https://openalex.org/W4226328099","https://openalex.org/W6600297362","https://openalex.org/W6600351811","https://openalex.org/W6601566401","https://openalex.org/W6769062451","https://openalex.org/W6796909790"],"related_works":["https://openalex.org/W2058965144","https://openalex.org/W2164382479","https://openalex.org/W98480971","https://openalex.org/W2150291671","https://openalex.org/W2027972911","https://openalex.org/W2146343568","https://openalex.org/W2013643406","https://openalex.org/W2157978810","https://openalex.org/W1966837078","https://openalex.org/W2081416538"],"abstract_inverted_index":{"Deep":[0],"learning":[1],"based":[2],"recommendation":[3,16,139],"models":[4,17,140,226],"(DLRM)":[5],"are":[6,85],"widely":[7],"used":[8,76],"in":[9,62,124,198],"several":[10],"business":[11],"critical":[12],"applications.":[13],"Training":[14],"such":[15],"efficiently":[18],"is":[19,51,65],"challenging":[20],"because":[21],"they":[22],"contain":[23],"billions":[24],"of":[25,47,91,97,234,241],"embedding-based":[26],"parameters,":[27],"leading":[28],"to":[29,77,114,146,167,236,239],"significant":[30],"overheads":[31,197],"from":[32],"embedding":[33,54,67,83,149],"access.":[34],"By":[35],"profiling":[36],"existing":[37],"systems":[38],"for":[39,136],"DLRM":[40],"training,":[41],"we":[42,101,108,131,203],"observe":[43,81,103],"that":[44,66,82,104,141,162,191,211,228],"around":[45,89],"75%":[46],"the":[48,125,152,242,247],"iteration":[49,123],"time":[50],"spent":[52],"on":[53,128],"access":[55,68],"and":[56,144,189,209,224,250],"model":[57],"synchronization.":[58],"Our":[59,219],"key":[60],"insight":[61],"this":[63],"paper":[64],"has":[69],"a":[70,134,159,164,183,199,205,231],"specific":[71],"structure":[72],"which":[73,116],"can":[74,109,194,214],"be":[75,119],"accelerate":[78],"training.":[79,255],"We":[80,154,180],"accesses":[84,150],"heavily":[86],"skewed,":[87],"with":[88,151],"1%":[90],"embeddings":[92,117],"representing":[93],"more":[94],"than":[95],"92%":[96],"total":[98],"accesses.":[99],"Further,":[100],"also":[102,181],"during":[105],"offline":[106],"training":[107,137],"lookahead":[110,165],"at":[111,121],"future":[112],"batches":[113],"determine":[115],"will":[118],"needed":[120],"what":[122],"future.":[126],"Based":[127],"these":[129],"insights,":[130],"develop":[132],"Bagpipe,":[133],"system":[135,207],"deep":[138],"uses":[142,163],"caching":[143],"prefetching":[145],"overlap":[147],"remote":[148],"computation.":[153],"design":[155,182,193,213],"an":[156],"Oracle":[157],"Cacher,":[158],"new":[160],"component":[161],"algorithm":[166],"generate":[168],"optimal":[169],"cache":[170,188],"update":[171],"decisions":[172],"while":[173,245],"providing":[174,246],"strong":[175],"consistency":[176],"guarantees":[177,252],"against":[178],"staleness.":[179],"logically":[184],"replicated,":[185],"physically":[186],"partitioned":[187],"show":[190,210,227],"our":[192,212],"reduce":[195],"synchronization":[196],"distributed":[200],"setting.":[201],"Finally,":[202],"propose":[204],"disaggregated":[206],"architecture":[208],"enable":[215],"low-overhead":[216],"fault":[217],"tolerance.":[218],"experiments":[220],"using":[221],"three":[222],"datasets":[223],"four":[225],"Bagpipe":[229],"provides":[230],"speed":[232],"up":[233,235],"5.6x":[237],"compared":[238],"state":[240],"art":[243],"baselines,":[244],"same":[248],"convergence":[249],"reproducibility":[251],"as":[253],"synchronous":[254]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":9},{"year":2023,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}