{"id":"https://openalex.org/W4414197296","doi":"https://doi.org/10.1109/dac63849.2025.11133377","title":"A Memory-Efficient LLM Accelerator with Q-K Correlation Prediction using Cluster-Based Associative Array for Selective KV Accessing","display_name":"A Memory-Efficient LLM Accelerator with Q-K Correlation Prediction using Cluster-Based Associative Array for Selective KV Accessing","publication_year":2025,"publication_date":"2025-06-22","ids":{"openalex":"https://openalex.org/W4414197296","doi":"https://doi.org/10.1109/dac63849.2025.11133377"},"language":"en","primary_location":{"id":"doi:10.1109/dac63849.2025.11133377","is_oa":false,"landing_page_url":"https://doi.org/10.1109/dac63849.2025.11133377","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 62nd ACM/IEEE Design Automation Conference (DAC)","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/A5073709403","display_name":"Zikang Zhou","orcid":"https://orcid.org/0009-0006-2530-3886"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zikang Zhou","raw_affiliation_strings":["Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022033867","display_name":"Kaiqi Chen","orcid":"https://orcid.org/0009-0006-6439-3255"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kaiqi Chen","raw_affiliation_strings":["Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089781655","display_name":"Xuyang Duan","orcid":"https://orcid.org/0000-0002-2609-846X"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xuyang Duan","raw_affiliation_strings":["Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5066674438","display_name":"Jun Han","orcid":"https://orcid.org/0000-0002-5245-0754"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jun Han","raw_affiliation_strings":["Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fudan University,State Key Laboratory of Integrated Chips and Systems,Shanghai,China","institution_ids":["https://openalex.org/I24943067"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I24943067"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"7"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10125","display_name":"Advanced Wireless Communication Techniques","score":0.9753999710083008,"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"}},"topics":[{"id":"https://openalex.org/T10125","display_name":"Advanced Wireless Communication Techniques","score":0.9753999710083008,"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/T11321","display_name":"Error Correcting Code Techniques","score":0.9732000231742859,"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/T11269","display_name":"Algorithms and Data Compression","score":0.9642999768257141,"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/speedup","display_name":"Speedup","score":0.695900022983551},{"id":"https://openalex.org/keywords/xeon","display_name":"Xeon","score":0.5859000086784363},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.45350000262260437},{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.4399999976158142},{"id":"https://openalex.org/keywords/data-access","display_name":"Data access","score":0.43650001287460327},{"id":"https://openalex.org/keywords/xeon-phi","display_name":"Xeon Phi","score":0.3698999881744385},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.36809998750686646},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.34549999237060547}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7789999842643738},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.695900022983551},{"id":"https://openalex.org/C145108525","wikidata":"https://www.wikidata.org/wiki/Q656154","display_name":"Xeon","level":2,"score":0.5859000086784363},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.49559998512268066},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.45350000262260437},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.4399999976158142},{"id":"https://openalex.org/C47487241","wikidata":"https://www.wikidata.org/wiki/Q5227230","display_name":"Data access","level":2,"score":0.43650001287460327},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.38940000534057617},{"id":"https://openalex.org/C96972482","wikidata":"https://www.wikidata.org/wiki/Q1049168","display_name":"Xeon Phi","level":2,"score":0.3698999881744385},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.36809998750686646},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.34549999237060547},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3427000045776367},{"id":"https://openalex.org/C194080101","wikidata":"https://www.wikidata.org/wiki/Q46306","display_name":"Access time","level":2,"score":0.33079999685287476},{"id":"https://openalex.org/C2994168587","wikidata":"https://www.wikidata.org/wiki/Q5295","display_name":"Random access memory","level":2,"score":0.3124000132083893},{"id":"https://openalex.org/C2779304628","wikidata":"https://www.wikidata.org/wiki/Q3503480","display_name":"Face (sociological concept)","level":2,"score":0.2989000082015991},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.29120001196861267},{"id":"https://openalex.org/C186370098","wikidata":"https://www.wikidata.org/wiki/Q442787","display_name":"Energy (signal processing)","level":2,"score":0.2883000075817108},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.28110000491142273},{"id":"https://openalex.org/C134835016","wikidata":"https://www.wikidata.org/wiki/Q690265","display_name":"Lookup table","level":2,"score":0.2741999924182892},{"id":"https://openalex.org/C53442348","wikidata":"https://www.wikidata.org/wiki/Q745101","display_name":"Content-addressable memory","level":3,"score":0.26739999651908875},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.26739999651908875},{"id":"https://openalex.org/C159423971","wikidata":"https://www.wikidata.org/wiki/Q177251","display_name":"Associative property","level":2,"score":0.2662999927997589},{"id":"https://openalex.org/C2987763966","wikidata":"https://www.wikidata.org/wiki/Q211382","display_name":"High speed train","level":2,"score":0.26010000705718994},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.2590999901294708},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.25040000677108765}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/dac63849.2025.11133377","is_oa":false,"landing_page_url":"https://doi.org/10.1109/dac63849.2025.11133377","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 62nd ACM/IEEE Design Automation Conference (DAC)","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":11,"referenced_works":["https://openalex.org/W1983394510","https://openalex.org/W2890894339","https://openalex.org/W2998617917","https://openalex.org/W3006586535","https://openalex.org/W3017024317","https://openalex.org/W3159727696","https://openalex.org/W3189877953","https://openalex.org/W3194676777","https://openalex.org/W3206453033","https://openalex.org/W4214686755","https://openalex.org/W4401211590"],"related_works":[],"abstract_inverted_index":{"Attention-based":[0],"LLMs":[1],"excel":[2],"in":[3,10],"text":[4],"generation":[5],"but":[6],"face":[7],"redundant":[8],"computations":[9],"autoregressive":[11],"token":[12],"generation.":[13],"While":[14],"KV":[15,44,108],"cache":[16,45],"mitigates":[17],"this,":[18],"it":[19],"introduces":[20],"increased":[21],"memory":[22,49,94],"access":[23,46,50,95],"overhead":[24],"as":[25],"sequences":[26],"grow.":[27],"We":[28],"propose":[29],"Sella,":[30],"a":[31,55,59],"hardware-software":[32],"co-design":[33],"using":[34],"cluster-based":[35],"associative":[36],"arrays":[37],"to":[38,62,98],"predict":[39],"Q-K":[40],"correlations,":[41],"enabling":[42],"selective":[43],"and":[47,65,78,88],"reducing":[48,92],"without":[51],"retraining.":[52],"Sella":[53,70],"includes":[54],"specialized":[56],"accelerator":[57],"featuring":[58],"prediction":[60],"engine":[61],"improve":[63],"performance":[64],"energy":[66],"efficiency.":[67],"Experiments":[68],"show":[69],"achieves":[71],"$2.1":[72],"\\times$,":[73,75,77],"$93.8":[74],"$31.4":[76],"$53.5":[79],"\\times$":[80],"speedup":[81],"over":[82],"SpAtten,":[83],"Sanger,":[84],"TITAN":[85],"RTX":[86],"GPU,":[87],"Xeon":[89],"CPU,":[90],"respectively,":[91],"off-chip":[93],"by":[96],"up":[97],"$66":[99],"\\%$":[100],"with":[101],"negligible":[102],"accuracy":[103],"loss.":[104],"-Large":[105],"Language":[106],"Models,":[107],"Cache,":[109],"Accelerator":[110]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}