{"id":"https://openalex.org/W4211105506","doi":"https://doi.org/10.1145/3490422.3502355","title":"An FPGA-based RNN-T Inference Accelerator with PIM-HBM","display_name":"An FPGA-based RNN-T Inference Accelerator with PIM-HBM","publication_year":2022,"publication_date":"2022-02-11","ids":{"openalex":"https://openalex.org/W4211105506","doi":"https://doi.org/10.1145/3490422.3502355"},"language":"en","primary_location":{"id":"doi:10.1145/3490422.3502355","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3490422.3502355","pdf_url":null,"source":null,"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2022 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays","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/A5084932060","display_name":"Shin-haeng Kang","orcid":null},"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":"Shinhaeng Kang","raw_affiliation_strings":["Samsung Electronics, Hwaseong, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Electronics, Hwaseong, South Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103020232","display_name":"Sukhan Lee","orcid":"https://orcid.org/0000-0002-1281-6889"},"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":"Sukhan Lee","raw_affiliation_strings":["Samsung Electronics, Hwaseong, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Electronics, Hwaseong, South Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004581533","display_name":"Byeongho Kim","orcid":"https://orcid.org/0000-0002-3227-2436"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Byeongho Kim","raw_affiliation_strings":["Seoul National University, Seoul, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060526894","display_name":"Hweesoo Kim","orcid":"https://orcid.org/0000-0002-5147-0972"},"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":"Hweesoo Kim","raw_affiliation_strings":["Samsung Electronics, Hwaseong, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Electronics, Hwaseong, South Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050861404","display_name":"Kyomin Sohn","orcid":"https://orcid.org/0000-0002-8094-9843"},"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":"Kyomin Sohn","raw_affiliation_strings":["Samsung Electronics, Hwaseong, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Electronics, Hwaseong, South Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5037648751","display_name":"Nam Sung Kim","orcid":"https://orcid.org/0000-0002-0442-5634"},"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":"Nam Sung Kim","raw_affiliation_strings":["Samsung Electronics, Hwaseong, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Electronics, Hwaseong, South Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5035539518","display_name":"Eojin Lee","orcid":"https://orcid.org/0000-0003-2739-2924"},"institutions":[{"id":"https://openalex.org/I191879574","display_name":"Inha University","ror":"https://ror.org/01easw929","country_code":"KR","type":"education","lineage":["https://openalex.org/I191879574"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Eojin Lee","raw_affiliation_strings":["Inha University, Incheon, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Inha University, Incheon, South Korea","institution_ids":["https://openalex.org/I191879574"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":13,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"146","last_page":"152"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9969000220298767,"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":0.9969000220298767,"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/T12326","display_name":"Network Packet Processing and Optimization","score":0.9968000054359436,"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/T11693","display_name":"Cryptography and Residue Arithmetic","score":0.9919999837875366,"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/computer-science","display_name":"Computer science","score":0.7973436117172241},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.7837951183319092},{"id":"https://openalex.org/keywords/hardware-acceleration","display_name":"Hardware acceleration","score":0.5602754354476929},{"id":"https://openalex.org/keywords/lookup-table","display_name":"Lookup table","score":0.5226452350616455},{"id":"https://openalex.org/keywords/reconfigurability","display_name":"Reconfigurability","score":0.4898674190044403},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.4598456621170044},{"id":"https://openalex.org/keywords/recurrent-neural-network","display_name":"Recurrent neural network","score":0.4259444773197174},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.42448073625564575},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.1622571051120758},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.08840325474739075}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7973436117172241},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.7837951183319092},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.5602754354476929},{"id":"https://openalex.org/C134835016","wikidata":"https://www.wikidata.org/wiki/Q690265","display_name":"Lookup table","level":2,"score":0.5226452350616455},{"id":"https://openalex.org/C2780149590","wikidata":"https://www.wikidata.org/wiki/Q7302742","display_name":"Reconfigurability","level":2,"score":0.4898674190044403},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.4598456621170044},{"id":"https://openalex.org/C147168706","wikidata":"https://www.wikidata.org/wiki/Q1457734","display_name":"Recurrent neural network","level":3,"score":0.4259444773197174},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.42448073625564575},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.1622571051120758},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.08840325474739075},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3490422.3502355","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3490422.3502355","pdf_url":null,"source":null,"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2022 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.8999999761581421,"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":11,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1606129576","https://openalex.org/W2024122052","https://openalex.org/W2194775991","https://openalex.org/W2401249903","https://openalex.org/W2741967249","https://openalex.org/W2912281309","https://openalex.org/W2962760690","https://openalex.org/W3043571714","https://openalex.org/W3134274954","https://openalex.org/W6687424872"],"related_works":["https://openalex.org/W2159103767","https://openalex.org/W2117300767","https://openalex.org/W2024574431","https://openalex.org/W2131696304","https://openalex.org/W2374017528","https://openalex.org/W4239932082","https://openalex.org/W4285503609","https://openalex.org/W2126248441","https://openalex.org/W2518118925","https://openalex.org/W4285429137"],"abstract_inverted_index":{"In":[0,99],"this":[1],"paper,":[2],"we":[3,74,101,133],"implemented":[4],"a":[5,60,76],"world-first":[6],"RNN-T":[7,34,127,165],"inference":[8],"accelerator":[9,24,157],"using":[10],"FPGA":[11],"with":[12,158,171],"PIM-HBM":[13,38,159],"that":[14,53],"can":[15],"multiply":[16],"the":[17,21,40,54,67,86,92,97,103,144,161,186],"internal":[18,49],"bandwidth":[19],"of":[20,30,43,66,85,96,164],"memory.":[22],"The":[23],"offloads":[25],"matrix-vector":[26],"multiplication":[27],"(GEMV)":[28],"operations":[29,128],"LSTM":[31],"layers":[32],"in":[33,59,71],"into":[35],"PIM-HBM,":[36,73],"and":[37,82,94,115,120,138,142,148,176],"reduces":[39,160],"execution":[41,162],"time":[42,163],"GEMV":[44],"significantly":[45],"by":[46,90,166],"exploiting":[47,72],"HBM":[48],"bandwidth.":[50],"To":[51],"ensure":[52],"memory":[55,78,88],"commands":[56],"are":[57],"issued":[58],"pre-defined":[61],"order,":[62],"which":[63],"is":[64],"one":[65],"most":[68],"important":[69],"constraints":[70],"implement":[75],"direct":[77],"access":[79],"(DMA)":[80],"module":[81],"change":[83],"configuration":[84],"on-chip":[87],"controller":[89],"utilizing":[91],"flexibility":[93],"reconfigurability":[95],"FPGA.":[98,130],"addition,":[100],"design":[102],"other":[104],"hardware":[105],"modules":[106],"for":[107],"acceleration":[108],"such":[109],"as":[110],"non-linear":[111],"functions":[112],"(i.e.,":[113],"sigmoid":[114],"hyperbolic":[116],"tangent),":[117],"element-wise":[118],"operation,":[119],"ReLU":[121],"module,":[122],"to":[123,181,185],"operate":[124],"these":[125],"compute-bound":[126],"on":[129,169],"For":[131],"this,":[132],"prepare":[134],"FP16":[135],"quantized":[136],"weight":[137],"MLPerf":[139],"input":[140],"datasets,":[141],"modify":[143],"PCIe":[145],"device":[146],"driver":[147],"C++":[149],"based":[150],"control":[151],"codes.":[152],"On":[153],"our":[154,156],"evaluation,":[155],"2.5":[167],"\u00d7":[168,183],"average":[170],"11.09%":[172],"reduced":[173],"LUT":[174],"size":[175],"improves":[177],"energy":[178],"efficiency":[179],"up":[180],"2.6":[182],"compared":[184],"baseline.":[187]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":4},{"year":2024,"cited_by_count":4},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}