{"id":"https://openalex.org/W4410584355","doi":"https://doi.org/10.23919/date64628.2025.10993224","title":"Leveraging Compute-in-Memory for Efficient Generative Model Inference in TPUs","display_name":"Leveraging Compute-in-Memory for Efficient Generative Model Inference in TPUs","publication_year":2025,"publication_date":"2025-03-31","ids":{"openalex":"https://openalex.org/W4410584355","doi":"https://doi.org/10.23919/date64628.2025.10993224"},"language":"en","primary_location":{"id":"doi:10.23919/date64628.2025.10993224","is_oa":false,"landing_page_url":"https://doi.org/10.23919/date64628.2025.10993224","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 Design, Automation & Test in Europe Conference (DATE)","raw_type":"proceedings-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/A5023075240","display_name":"Zhiguang Zhu","orcid":"https://orcid.org/0009-0008-7859-9734"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Zhantong Zhu","raw_affiliation_strings":["School of Integrated Circuits, Peking University,Beijing,China"],"affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, Peking University,Beijing,China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101684590","display_name":"Hong Li","orcid":"https://orcid.org/0000-0002-7614-3484"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hongou Li","raw_affiliation_strings":["School of Integrated Circuits, Peking University,Beijing,China"],"affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, Peking University,Beijing,China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100314793","display_name":"Wenjie Ren","orcid":null},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenjie Ren","raw_affiliation_strings":["School of Integrated Circuits, Peking University,Beijing,China"],"affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, Peking University,Beijing,China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Meng Wu","orcid":null},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Meng Wu","raw_affiliation_strings":["School of Integrated Circuits, Peking University,Beijing,China"],"affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, Peking University,Beijing,China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003039083","display_name":"Le Ye","orcid":"https://orcid.org/0000-0003-0599-7762"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Le Ye","raw_affiliation_strings":["School of Integrated Circuits, Peking University,Beijing,China"],"affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, Peking University,Beijing,China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062886480","display_name":"Ru Huang","orcid":"https://orcid.org/0000-0002-8146-4821"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ru Huang","raw_affiliation_strings":["School of Integrated Circuits, Peking University,Beijing,China"],"affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, Peking University,Beijing,China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5088551028","display_name":"Tianyu Jia","orcid":"https://orcid.org/0000-0002-4570-4613"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tianyu Jia","raw_affiliation_strings":["School of Integrated Circuits, Peking University,Beijing,China"],"affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, Peking University,Beijing,China","institution_ids":["https://openalex.org/I20231570"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":7,"corresponding_author_ids":["https://openalex.org/A5023075240"],"corresponding_institution_ids":["https://openalex.org/I20231570"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.05212211,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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/T10320","display_name":"Neural Networks and Applications","score":0.9315999746322632,"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"}},"topics":[{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.9315999746322632,"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":0.8008807897567749},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.7360712885856628},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.538201093673706},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.5209406614303589},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3945888876914978},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.33090078830718994},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.31596267223358154}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8008807897567749},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7360712885856628},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.538201093673706},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.5209406614303589},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3945888876914978},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.33090078830718994},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.31596267223358154}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.23919/date64628.2025.10993224","is_oa":false,"landing_page_url":"https://doi.org/10.23919/date64628.2025.10993224","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 Design, Automation & Test in Europe Conference (DATE)","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":22,"referenced_works":["https://openalex.org/W2922487710","https://openalex.org/W2940862705","https://openalex.org/W3015432327","https://openalex.org/W3015982917","https://openalex.org/W3097528158","https://openalex.org/W3130554079","https://openalex.org/W3190062760","https://openalex.org/W3194056411","https://openalex.org/W3213528054","https://openalex.org/W4221101426","https://openalex.org/W4312933868","https://openalex.org/W4323022446","https://openalex.org/W4360605703","https://openalex.org/W4360605969","https://openalex.org/W4360606939","https://openalex.org/W4380874786","https://openalex.org/W4390872297","https://openalex.org/W4392746406","https://openalex.org/W4400681289","https://openalex.org/W4401211642","https://openalex.org/W4403674993","https://openalex.org/W6778883912"],"related_works":["https://openalex.org/W4365211920","https://openalex.org/W3014948380","https://openalex.org/W4391584540","https://openalex.org/W4380551139","https://openalex.org/W4317695495","https://openalex.org/W4395044357","https://openalex.org/W4287117424","https://openalex.org/W4387506531","https://openalex.org/W2087346071","https://openalex.org/W2967848559"],"abstract_inverted_index":{"With":[0],"the":[1,86,97,142],"rapid":[2],"advent":[3],"of":[4,88],"generative":[5,92],"models,":[6],"efficiently":[7],"deploying":[8],"these":[9],"models":[10],"on":[11],"specialized":[12],"hardware":[13],"has":[14,39],"become":[15],"critical.":[16],"Tensor":[17],"Processing":[18],"Units":[19],"(TPUs)":[20],"are":[21],"designed":[22],"to":[23,63,84,111,141],"accelerate":[24],"AI":[25],"workloads,":[26],"but":[27],"their":[28],"high":[29],"power":[30],"consumption":[31,132],"neces-sitates":[32],"innovations":[33],"for":[34,90,117],"improving":[35],"efficiency.":[36,50],"Compute-in-memory":[37],"(CIM)":[38],"emerged":[40],"as":[41],"a":[42,56,77],"promising":[43],"paradigm":[44],"with":[45,136],"superior":[46],"area":[47],"and":[48,82,113,121,125],"energy":[49,131],"In":[51],"this":[52],"work,":[53],"we":[54,101],"present":[55],"TPU":[57,79,106],"architecture":[58,80],"that":[59],"integrates":[60],"digital":[61,66],"CIM":[62,89],"replace":[64],"conventional":[65],"systolic":[67],"arrays":[68],"in":[69,129],"matrix":[70],"multiply":[71],"units":[72],"(MXUs).":[73],"We":[74],"first":[75],"establish":[76],"CIM-based":[78,105],"model":[81,93,120],"simulator":[83],"evaluate":[85],"benefits":[87],"diverse":[91],"inference.":[94],"Building":[95],"upon":[96],"observed":[98],"design":[99,108,138],"insights,":[100],"further":[102],"explore":[103],"various":[104],"architectural":[107],"choices.":[109],"Up":[110],"44.2%":[112],"33.8%":[114],"performance":[115],"improvement":[116],"large":[118],"language":[119],"diffusion":[122],"transformer":[123],"inference,":[124],"27.3":[126],"\u00d7":[127],"reduction":[128],"MXU":[130],"can":[133],"be":[134],"achieved":[135],"different":[137],"choices,":[139],"compared":[140],"baseline":[143],"TPUv4i":[144],"architecture.":[145]},"counts_by_year":[],"updated_date":"2026-03-25T23:56:10.502304","created_date":"2025-10-10T00:00:00"}