{"id":"https://openalex.org/W4312659766","doi":"https://doi.org/10.1109/asap54787.2022.00022","title":"Answer Fast: Accelerating BERT on the Tensor Streaming Processor","display_name":"Answer Fast: Accelerating BERT on the Tensor Streaming Processor","publication_year":2022,"publication_date":"2022-07-01","ids":{"openalex":"https://openalex.org/W4312659766","doi":"https://doi.org/10.1109/asap54787.2022.00022"},"language":"en","primary_location":{"id":"doi:10.1109/asap54787.2022.00022","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asap54787.2022.00022","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2022 IEEE 33rd International Conference on Application-specific Systems, Architectures and Processors (ASAP)","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/A5059900984","display_name":"Ibrahim Ahmed","orcid":null},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":true,"raw_author_name":"Ibrahim Ahmed","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5017713318","display_name":"Sahil Parmar","orcid":null},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Sahil Parmar","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5013209064","display_name":"Matthew Boyd","orcid":"https://orcid.org/0000-0003-2997-5090"},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Matthew Boyd","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027331427","display_name":"Michael Beidler","orcid":null},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Michael Beidler","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075070812","display_name":"Kris Kang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Kris Kang","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028312498","display_name":"Bill Liu","orcid":"https://orcid.org/0000-0002-3073-1598"},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Bill Liu","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5018000797","display_name":"Kyle Roach","orcid":null},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Kyle Roach","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100698124","display_name":"John Kim","orcid":"https://orcid.org/0000-0003-3958-3891"},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"John Kim","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5015385537","display_name":"Dennis Abts","orcid":"https://orcid.org/0009-0007-7108-9013"},"institutions":[{"id":"https://openalex.org/I4210132883","display_name":"IQ Samh\u00e4llsbyggnad","ror":"https://ror.org/03knfbb08","country_code":"SE","type":"other","lineage":["https://openalex.org/I4210132883"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Dennis Abts","raw_affiliation_strings":["Groq Inc"],"affiliations":[{"raw_affiliation_string":"Groq Inc","institution_ids":["https://openalex.org/I4210132883"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":9,"corresponding_author_ids":["https://openalex.org/A5059900984"],"corresponding_institution_ids":["https://openalex.org/I4210132883"],"apc_list":null,"apc_paid":null,"fwci":2.5547,"has_fulltext":false,"cited_by_count":12,"citation_normalized_percentile":{"value":0.89828054,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"80","last_page":"87"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9995999932289124,"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":0.9995999932289124,"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/T12303","display_name":"Tensor decomposition and applications","score":0.9966999888420105,"subfield":{"id":"https://openalex.org/subfields/2605","display_name":"Computational Mathematics"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T13650","display_name":"Computational Physics and Python Applications","score":0.9876999855041504,"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":0.8090071678161621},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6272834539413452},{"id":"https://openalex.org/keywords/machine-translation","display_name":"Machine translation","score":0.6041184067726135},{"id":"https://openalex.org/keywords/matrix-multiplication","display_name":"Matrix multiplication","score":0.5973397493362427},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.5366105437278748},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.43846607208251953},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.3715912699699402},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3489427864551544},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.3342466354370117},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.33017945289611816},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.2510221600532532},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.23920926451683044}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8090071678161621},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6272834539413452},{"id":"https://openalex.org/C203005215","wikidata":"https://www.wikidata.org/wiki/Q79798","display_name":"Machine translation","level":2,"score":0.6041184067726135},{"id":"https://openalex.org/C17349429","wikidata":"https://www.wikidata.org/wiki/Q1049914","display_name":"Matrix multiplication","level":3,"score":0.5973397493362427},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.5366105437278748},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.43846607208251953},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3715912699699402},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3489427864551544},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3342466354370117},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.33017945289611816},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2510221600532532},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.23920926451683044},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C84114770","wikidata":"https://www.wikidata.org/wiki/Q46344","display_name":"Quantum","level":2,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/asap54787.2022.00022","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asap54787.2022.00022","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2022 IEEE 33rd International Conference on Application-specific Systems, Architectures and Processors (ASAP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.699999988079071}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":24,"referenced_works":["https://openalex.org/W1583837637","https://openalex.org/W2037450062","https://openalex.org/W2612690371","https://openalex.org/W2896457183","https://openalex.org/W2899663614","https://openalex.org/W2963323070","https://openalex.org/W3005724337","https://openalex.org/W3017024317","https://openalex.org/W3017746288","https://openalex.org/W3035038672","https://openalex.org/W3042416028","https://openalex.org/W3130716829","https://openalex.org/W3159727696","https://openalex.org/W3177265267","https://openalex.org/W4288088457","https://openalex.org/W4309067679","https://openalex.org/W6739901393","https://openalex.org/W6755207826","https://openalex.org/W6769806307","https://openalex.org/W6774131485","https://openalex.org/W6776320331","https://openalex.org/W6779436764","https://openalex.org/W6780226713","https://openalex.org/W6810549907"],"related_works":["https://openalex.org/W2595172197","https://openalex.org/W2084856301","https://openalex.org/W2127970246","https://openalex.org/W2885125400","https://openalex.org/W1989889224","https://openalex.org/W4382618745","https://openalex.org/W1973775000","https://openalex.org/W2748922771","https://openalex.org/W1987128138","https://openalex.org/W2743976221"],"abstract_inverted_index":{"Transformers":[0],"have":[1],"become":[2,83],"a":[3,129,137],"predominant":[4],"machine":[5,45],"learning":[6],"workload,":[7],"they":[8,21],"are":[9,22,40,117],"not":[10],"only":[11],"the":[12,37,64,67,84,94,100,108,112,122,147],"de-facto":[13],"standard":[14],"for":[15,136],"natural":[16],"language":[17],"processing":[18],"tasks,":[19],"but":[20],"also":[23,75],"being":[24],"deployed":[25],"in":[26,128],"other":[27],"domains":[28],"such":[29,43],"as":[30,44],"vision":[31],"and":[32,47],"speech":[33],"recognition.":[34],"Many":[35],"of":[36,66,96,133],"transformer-based":[38],"applications":[39],"real-time":[41],"systems":[42,53],"translation":[46],"web":[48],"search.":[49],"These":[50],"real":[51],"time":[52],"often":[54],"come":[55],"with":[56,111],"strict":[57],"end-to-end":[58],"inference":[59,95,139],"latency":[60,132],"requirements.":[61],"Unfortunately,":[62],"while":[63],"majority":[65],"transformer":[68],"computation":[69],"comes":[70],"from":[71],"matrix":[72,113,124],"multiplications,":[73],"transformers":[74],"include":[76],"several":[77],"non-linear":[78],"components":[79,110],"that":[80],"tend":[81],"to":[82,119],"bottleneck":[85],"during":[86],"an":[87],"inference.":[88],"In":[89],"this":[90],"work,":[91],"we":[92,116],"accelerate":[93],"BERT":[97],"models":[98],"on":[99],"tensor":[101],"streaming":[102],"processor.":[103],"By":[104],"carefully":[105],"fusing":[106],"all":[107],"nonlinear":[109],"multiplication":[114,125],"components,":[115],"able":[118],"efficiently":[120],"utilize":[121],"on-chip":[123],"units":[126],"resulting":[127],"deterministic":[130],"tail":[131],"130":[134],"\u03bcs":[135],"batch-1":[138],"through":[140],"BERT-base,":[141],"which":[142],"is":[143],"6\u00d7":[144],"faster":[145],"than":[146],"current":[148],"state-of-the-art.":[149]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":5},{"year":2023,"cited_by_count":4},{"year":2022,"cited_by_count":1}],"updated_date":"2026-03-26T15:22:09.906841","created_date":"2025-10-10T00:00:00"}