{"id":"https://openalex.org/W4405934793","doi":"https://doi.org/10.1109/icm63406.2024.10815822","title":"Hardware Accelerator for Bidirectional Encoder Representations from Transformers (BERT)","display_name":"Hardware Accelerator for Bidirectional Encoder Representations from Transformers (BERT)","publication_year":2024,"publication_date":"2024-12-14","ids":{"openalex":"https://openalex.org/W4405934793","doi":"https://doi.org/10.1109/icm63406.2024.10815822"},"language":"en","primary_location":{"id":"doi:10.1109/icm63406.2024.10815822","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icm63406.2024.10815822","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 International Conference on Microelectronics (ICM)","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/A5100377935","display_name":"Yiming Wang","orcid":"https://orcid.org/0000-0002-5932-4371"},"institutions":[{"id":"https://openalex.org/I4210111595","display_name":"Prodrive Technologies (Netherlands)","ror":"https://ror.org/01zhpc307","country_code":"NL","type":"company","lineage":["https://openalex.org/I4210111595"]}],"countries":["NL"],"is_corresponding":true,"raw_author_name":"Yiming Wang","raw_affiliation_strings":["Prodrive Technologies,Dept. of Embedded Computing Systems,Eindhoven,Netherlands"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Prodrive Technologies,Dept. of Embedded Computing Systems,Eindhoven,Netherlands","institution_ids":["https://openalex.org/I4210111595"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5100377935"],"corresponding_institution_ids":["https://openalex.org/I4210111595"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.8133000135421753,"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":0.8133000135421753,"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/encoder","display_name":"Encoder","score":0.7128740549087524},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6429749131202698},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5911881327629089},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.4448881149291992},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.42215076088905334},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.1923905313014984},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.14972344040870667},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.11739304661750793}],"concepts":[{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.7128740549087524},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6429749131202698},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5911881327629089},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.4448881149291992},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.42215076088905334},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1923905313014984},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.14972344040870667},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.11739304661750793}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icm63406.2024.10815822","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icm63406.2024.10815822","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 International Conference on Microelectronics (ICM)","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":12,"referenced_works":["https://openalex.org/W2606722458","https://openalex.org/W2970454332","https://openalex.org/W3007007518","https://openalex.org/W3012178976","https://openalex.org/W3196923642","https://openalex.org/W6755207826","https://openalex.org/W6768021236","https://openalex.org/W6780226713","https://openalex.org/W6847478871","https://openalex.org/W6852962002","https://openalex.org/W6856696905","https://openalex.org/W6861098061"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2363440576","https://openalex.org/W2864363823"],"abstract_inverted_index":{"Recently":[0],"Bidirectional":[1],"Encoder":[2],"Representations":[3],"from":[4],"Transformers":[5],"(BERT)":[6],"model":[7,35,62,78,99],"has":[8],"gained":[9],"lots":[10],"of":[11,14,41,91,134,178,217],"attention":[12],"because":[13,90],"its":[15],"state-of-the-art":[16],"performance":[17,94,216],"in":[18,207],"multiple":[19],"natu-ral":[20],"language":[21],"processing":[22,115],"(NLP)":[23],"tasks.":[24],"However,":[25],"just":[26],"like":[27],"many":[28],"other":[29],"deep":[30],"learning":[31],"based":[32,171],"tasks,":[33],"large":[34],"size":[36,100],"and":[37,46,50,67,80,101,123,132,146,185,201,210,223],"intensive":[38],"computation":[39],"load":[40],"BERT":[42,61],"make":[43],"it":[44],"difficult":[45],"expensive":[47],"to":[48,119,154,162,193],"run":[49,197],"implement":[51],"on":[52,172,198],"general":[53],"purpose":[54],"processors.":[55],"The":[56,175,214],"proposed":[57],"hardware":[58,81,106,141,166,180],"accelerator":[59],"for":[60,144],"realizes":[63],"faster":[64],"inference":[65],"speed":[66,177,203],"higher":[68],"energy":[69],"efficiency.":[70],"Design":[71],"procedure":[72],"is":[73,84,112,142,152,169,182,189,220,229],"elaborated":[74],"with":[75,159],"two":[76],"stages:":[77],"compression":[79,88],"architecture.":[82],"Quantization":[83],"chosen":[85],"as":[86,95,97,114],"the":[87,105,128,164,186,194,224],"technique":[89],"good":[92],"speed-up":[93],"well":[96],"small":[98],"less":[102],"complexity.":[103],"In":[104],"design,":[107],"systolic":[108],"tensor":[109],"array":[110,118],"(STA)":[111],"applied":[113],"elements":[116],"(PE)":[117],"achieve":[120],"lower":[121],"area":[122],"power":[124],"consumption":[125],"by":[126],"reducing":[127],"ratio":[129],"between":[130],"registers":[131],"number":[133],"Floating-point":[135],"operations":[136,161],"per":[137],"second":[138],"(FLOPS).":[139],"Dedicated":[140],"designed":[143],"Softmax":[145],"layer":[147],"normalization":[148],"operations.":[149],"Mathematical":[150],"transformation":[151],"used":[153],"replace":[155],"complicate":[156],"nonlinear":[157],"functions":[158],"simple":[160],"reduce":[163],"required":[165,226],"resources.":[167],"Performance":[168],"evaluated":[170],"transformer-base":[173],"model.":[174],"maximum":[176,225],"overall":[179],"design":[181,219],"125":[183],"MHz":[184],"total":[187],"latency":[188],"165.9":[190],"us.":[191],"Compared":[192],"same":[195],"task":[196],"GPU,":[199],"22.4x":[200],"7.5x":[202],"up":[204],"are":[205],"achieved":[206],"multi-head-attention":[208],"(MHA)":[209],"feed-forward":[211],"networks(FFN)":[212],"separately.":[213],"peak":[215],"this":[218],"4.1":[221],"TOPs/s":[222],"memory":[227],"bandwidth":[228],"80":[230],"GB/s.":[231]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}