{"id":"https://openalex.org/W7114917578","doi":"https://doi.org/10.1109/tc.2025.3643434","title":"Performance Analysis and CXL Memory Optimization in Cluster-Based RAG Systems","display_name":"Performance Analysis and CXL Memory Optimization in Cluster-Based RAG Systems","publication_year":2025,"publication_date":"2025-12-12","ids":{"openalex":"https://openalex.org/W7114917578","doi":"https://doi.org/10.1109/tc.2025.3643434"},"language":null,"primary_location":{"id":"doi:10.1109/tc.2025.3643434","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tc.2025.3643434","pdf_url":null,"source":{"id":"https://openalex.org/S157670870","display_name":"IEEE Transactions on Computers","issn_l":"0018-9340","issn":["0018-9340","0016-9340","1557-9956","2326-3814"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computers","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":null,"display_name":"Jesin Kim","orcid":"https://orcid.org/0009-0008-7432-0150"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jesin Kim","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0008-7432-0150","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Hochul Shin","orcid":"https://orcid.org/0009-0000-1592-8803"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hochul Shin","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0000-1592-8803","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Byeonghun Hwang","orcid":"https://orcid.org/0009-0008-3432-4747"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Byeonghun Hwang","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0008-3432-4747","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Seungpyo Cho","orcid":"https://orcid.org/0009-0006-8649-7020"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Seungpyo Cho","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0006-8649-7020","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Kyumin Park","orcid":"https://orcid.org/0009-0001-9541-5756"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kyumin Park","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0001-9541-5756","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Young Geon Yoo","orcid":"https://orcid.org/0009-0004-0502-8544"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Young Geon Yoo","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0004-0502-8544","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Byungyo Lee","orcid":"https://orcid.org/0009-0005-4790-436X"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Byungyo Lee","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0005-4790-436X","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"last","author":{"id":null,"display_name":"Hojun Shim","orcid":"https://orcid.org/0009-0002-6797-7994"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hojun Shim","raw_affiliation_strings":["Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea"],"raw_orcid":"https://orcid.org/0009-0002-6797-7994","affiliations":[{"raw_affiliation_string":"Memory Business, Device Solutions, Samsung Electronics, Seoul, Korea","institution_ids":["https://openalex.org/I2250650973"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I2250650973"],"apc_list":null,"apc_paid":null,"fwci":1.377,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.85667216,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":"75","issue":"4","first_page":"1323","last_page":"1334"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.2312999963760376,"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":0.2312999963760376,"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/T11181","display_name":"Advanced Data Storage Technologies","score":0.15330000221729279,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.07339999824762344,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.6069999933242798},{"id":"https://openalex.org/keywords/interleaved-memory","display_name":"Interleaved memory","score":0.5412999987602234},{"id":"https://openalex.org/keywords/interleaving","display_name":"Interleaving","score":0.5370000004768372},{"id":"https://openalex.org/keywords/cas-latency","display_name":"CAS latency","score":0.527999997138977},{"id":"https://openalex.org/keywords/semiconductor-memory","display_name":"Semiconductor memory","score":0.525600016117096},{"id":"https://openalex.org/keywords/memory-management","display_name":"Memory management","score":0.4918000102043152},{"id":"https://openalex.org/keywords/flat-memory-model","display_name":"Flat memory model","score":0.47940000891685486},{"id":"https://openalex.org/keywords/dram","display_name":"Dram","score":0.46320000290870667},{"id":"https://openalex.org/keywords/context-switch","display_name":"Context switch","score":0.45829999446868896}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8424999713897705},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.6069999933242798},{"id":"https://openalex.org/C63511323","wikidata":"https://www.wikidata.org/wiki/Q908936","display_name":"Interleaved memory","level":4,"score":0.5412999987602234},{"id":"https://openalex.org/C28034677","wikidata":"https://www.wikidata.org/wiki/Q17092530","display_name":"Interleaving","level":2,"score":0.5370000004768372},{"id":"https://openalex.org/C189930140","wikidata":"https://www.wikidata.org/wiki/Q1112878","display_name":"CAS latency","level":4,"score":0.527999997138977},{"id":"https://openalex.org/C98986596","wikidata":"https://www.wikidata.org/wiki/Q1143031","display_name":"Semiconductor memory","level":2,"score":0.525600016117096},{"id":"https://openalex.org/C176649486","wikidata":"https://www.wikidata.org/wiki/Q2308807","display_name":"Memory management","level":3,"score":0.4918000102043152},{"id":"https://openalex.org/C57863822","wikidata":"https://www.wikidata.org/wiki/Q905488","display_name":"Flat memory model","level":4,"score":0.47940000891685486},{"id":"https://openalex.org/C7366592","wikidata":"https://www.wikidata.org/wiki/Q1255620","display_name":"Dram","level":2,"score":0.46320000290870667},{"id":"https://openalex.org/C53833338","wikidata":"https://www.wikidata.org/wiki/Q1061424","display_name":"Context switch","level":2,"score":0.45829999446868896},{"id":"https://openalex.org/C93446704","wikidata":"https://www.wikidata.org/wiki/Q449328","display_name":"Registered memory","level":3,"score":0.44699999690055847},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.4140999913215637},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.385699987411499},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.38510000705718994},{"id":"https://openalex.org/C188045654","wikidata":"https://www.wikidata.org/wiki/Q17148339","display_name":"Memory bandwidth","level":2,"score":0.3846000134944916},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.36010000109672546},{"id":"https://openalex.org/C2984984529","wikidata":"https://www.wikidata.org/wiki/Q7619925","display_name":"Storage management","level":2,"score":0.3555000126361847},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.3465000092983246},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3188999891281128},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3043999969959259},{"id":"https://openalex.org/C51290061","wikidata":"https://www.wikidata.org/wiki/Q1936765","display_name":"Uniform memory access","level":4,"score":0.30300000309944153},{"id":"https://openalex.org/C74426580","wikidata":"https://www.wikidata.org/wiki/Q719484","display_name":"Memory map","level":3,"score":0.2815999984741211},{"id":"https://openalex.org/C171675096","wikidata":"https://www.wikidata.org/wiki/Q1143380","display_name":"Extended memory","level":4,"score":0.2759000062942505},{"id":"https://openalex.org/C82687282","wikidata":"https://www.wikidata.org/wiki/Q66221","display_name":"Auxiliary memory","level":2,"score":0.27410000562667847},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.26809999346733093},{"id":"https://openalex.org/C133875982","wikidata":"https://www.wikidata.org/wiki/Q764810","display_name":"Shared memory","level":2,"score":0.2596000134944916},{"id":"https://openalex.org/C39528615","wikidata":"https://www.wikidata.org/wiki/Q1229610","display_name":"Distributed shared memory","level":5,"score":0.25679999589920044},{"id":"https://openalex.org/C100800780","wikidata":"https://www.wikidata.org/wiki/Q1175867","display_name":"Memory controller","level":3,"score":0.25040000677108765}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tc.2025.3643434","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tc.2025.3643434","pdf_url":null,"source":{"id":"https://openalex.org/S157670870","display_name":"IEEE Transactions on Computers","issn_l":"0018-9340","issn":["0018-9340","0016-9340","1557-9956","2326-3814"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computers","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":19,"referenced_works":["https://openalex.org/W2124509324","https://openalex.org/W2963469388","https://openalex.org/W3006878628","https://openalex.org/W3174809957","https://openalex.org/W4312285477","https://openalex.org/W4327930439","https://openalex.org/W4399677840","https://openalex.org/W4404181298","https://openalex.org/W4405755164","https://openalex.org/W4407197060","https://openalex.org/W4408901239","https://openalex.org/W4410359057","https://openalex.org/W4411553040","https://openalex.org/W4412081840","https://openalex.org/W4412610672","https://openalex.org/W4413430504","https://openalex.org/W4413968235","https://openalex.org/W4416203810","https://openalex.org/W7133184130"],"related_works":[],"abstract_inverted_index":{"Retrieval-augmented":[0],"generation":[1],"(RAG)":[2],"systems":[3],"rely":[4],"on":[5],"large-scale":[6],"embedding":[7],"search.":[8],"However,":[9],"the":[10,15,21,46,108,136],"growth":[11],"of":[12,27,50,110,140],"corpora":[13],"and":[14,24,44,53,63,73,77,94,119,126],"demand":[16],"for":[17],"real-time":[18],"queries":[19],"expose":[20],"memory":[22,37,61,66,143,149,155],"capacity":[23],"bandwidth":[25],"limitations":[26],"DRAM-centric":[28],"architectures.":[29],"To":[30],"address":[31],"this":[32],"challenge,":[33],"we":[34,114],"adopt":[35],"CXL":[36,60,78,141,148],"in":[38,48,107,122,157],"conjunction":[39],"with":[40,65,82],"cluster-scale":[41],"RAG":[42,159],"deployments":[43],"optimize":[45],"system":[47,87,112],"terms":[49],"performance,":[51],"latency":[52,139],"cost":[54,127],"efficiency.":[55],"Our":[56],"design":[57],"leverages":[58],"state-of-the-art":[59],"devices":[62],"platforms":[64],"management":[67,144],"policies":[68],"such":[69],"as":[70,152],"weighted":[71],"interleaving":[72],"tiering":[74],"between":[75],"DRAM":[76,84],"memory.":[79],"In":[80],"comparison":[81],"a":[83,153],"baseline,":[85],"our":[86],"archives":[88],"20%":[89,118],"higher":[90],"vector":[91],"search":[92],"performance":[93,104],"43%":[95],"lower":[96],"LLM":[97],"time-to-first":[98],"token":[99],"(TTFT),":[100],"demonstrating":[101],"clear":[102],"end-to-end":[103],"benefits.":[105],"Moreover,":[106],"context":[109],"overall":[111],"efficiency,":[113],"observed":[115],"up":[116],"to":[117,150],"42%":[120],"improvement":[121],"power":[123],"efficiency":[124,128],"(QPS/Watt)":[125],"(QPS/$),":[129],"respectively.":[130],"These":[131],"results":[132],"demonstrate":[133],"that,":[134],"despite":[135],"relatively":[137],"high":[138],"memory,":[142],"policy-level":[145],"optimization":[146],"enable":[147],"serve":[151],"first-class":[154],"resource":[156],"cluster-based":[158],"systems.":[160]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-12-12T00:00:00"}
