{"id":"https://openalex.org/W4407953524","doi":"https://doi.org/10.1145/3706628.3708849","title":"BAQET: BRAM-aware Quantization for Efficient Transformer Inference via Stream-based Architecture on an FPGA","display_name":"BAQET: BRAM-aware Quantization for Efficient Transformer Inference via Stream-based Architecture on an FPGA","publication_year":2025,"publication_date":"2025-02-26","ids":{"openalex":"https://openalex.org/W4407953524","doi":"https://doi.org/10.1145/3706628.3708849"},"language":"en","primary_location":{"id":"doi:10.1145/3706628.3708849","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3706628.3708849","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 ACM/SIGDA International Symposium on Field Programmable Gate Arrays","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/A5113115855","display_name":"Ling-Chi Yang","orcid":"https://orcid.org/0009-0008-3064-6108"},"institutions":[{"id":"https://openalex.org/I148366613","display_name":"National Yang Ming Chiao Tung University","ror":"https://ror.org/00se2k293","country_code":"TW","type":"education","lineage":["https://openalex.org/I148366613"]}],"countries":["TW"],"is_corresponding":true,"raw_author_name":"Ling-Chi Yang","raw_affiliation_strings":["Institute of Electronics National Yang Ming Chiao Tung University, Hsinchu, Taiwan"],"affiliations":[{"raw_affiliation_string":"Institute of Electronics National Yang Ming Chiao Tung University, Hsinchu, Taiwan","institution_ids":["https://openalex.org/I148366613"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112375572","display_name":"C. H. Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I148366613","display_name":"National Yang Ming Chiao Tung University","ror":"https://ror.org/00se2k293","country_code":"TW","type":"education","lineage":["https://openalex.org/I148366613"]}],"countries":["TW"],"is_corresponding":false,"raw_author_name":"Chi-Jui Chen","raw_affiliation_strings":["Graduate Degree Program of College of Electrical and Computer Engineering in National Yang Ming Chiao Tung University, Hsinchu, Taiwan"],"affiliations":[{"raw_affiliation_string":"Graduate Degree Program of College of Electrical and Computer Engineering in National Yang Ming Chiao Tung University, Hsinchu, Taiwan","institution_ids":["https://openalex.org/I148366613"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5068000289","display_name":"Trung Le","orcid":"https://orcid.org/0000-0002-7047-8451"},"institutions":[{"id":"https://openalex.org/I201448701","display_name":"University of Washington","ror":"https://ror.org/00cvxb145","country_code":"US","type":"education","lineage":["https://openalex.org/I201448701"]},{"id":"https://openalex.org/I58610484","display_name":"Seattle University","ror":"https://ror.org/02jqc0m91","country_code":"US","type":"education","lineage":["https://openalex.org/I58610484"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Trung Le","raw_affiliation_strings":["Electrical and Computer Engineering in University of Washington, Seattle, USA"],"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering in University of Washington, Seattle, USA","institution_ids":["https://openalex.org/I201448701","https://openalex.org/I58610484"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5013913342","display_name":"Bo\u2010Cheng Lai","orcid":"https://orcid.org/0000-0002-9729-5196"},"institutions":[{"id":"https://openalex.org/I148366613","display_name":"National Yang Ming Chiao Tung University","ror":"https://ror.org/00se2k293","country_code":"TW","type":"education","lineage":["https://openalex.org/I148366613"]}],"countries":["TW"],"is_corresponding":false,"raw_author_name":"Bo-Cheng Lai","raw_affiliation_strings":["Institute of Electronics National Yang Ming Chiao Tung University, Hsinchu, Taiwan"],"affiliations":[{"raw_affiliation_string":"Institute of Electronics National Yang Ming Chiao Tung University, Hsinchu, Taiwan","institution_ids":["https://openalex.org/I148366613"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003894282","display_name":"Scott Hauck","orcid":"https://orcid.org/0000-0001-9516-0311"},"institutions":[{"id":"https://openalex.org/I201448701","display_name":"University of Washington","ror":"https://ror.org/00cvxb145","country_code":"US","type":"education","lineage":["https://openalex.org/I201448701"]},{"id":"https://openalex.org/I58610484","display_name":"Seattle University","ror":"https://ror.org/02jqc0m91","country_code":"US","type":"education","lineage":["https://openalex.org/I58610484"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Scott Hauck","raw_affiliation_strings":["Electrical and Computer Engineering University of Washington, Seattle, USA"],"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering University of Washington, Seattle, USA","institution_ids":["https://openalex.org/I201448701","https://openalex.org/I58610484"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5105676474","display_name":"S.\u2010C. Hsu","orcid":"https://orcid.org/0000-0001-6214-8500"},"institutions":[{"id":"https://openalex.org/I4210138199","display_name":"University of Washington Applied Physics Laboratory","ror":"https://ror.org/03d17d270","country_code":"US","type":"facility","lineage":["https://openalex.org/I201448701","https://openalex.org/I4210138199"]},{"id":"https://openalex.org/I58610484","display_name":"Seattle University","ror":"https://ror.org/02jqc0m91","country_code":"US","type":"education","lineage":["https://openalex.org/I58610484"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shih-Chieh Hsu","raw_affiliation_strings":["Physics University of Washington, Seattle, USA"],"affiliations":[{"raw_affiliation_string":"Physics University of Washington, Seattle, USA","institution_ids":["https://openalex.org/I4210138199","https://openalex.org/I58610484"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5113115855"],"corresponding_institution_ids":["https://openalex.org/I148366613"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.02334932,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"51","last_page":"51"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10688","display_name":"Image and Signal Denoising Methods","score":0.974399983882904,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T10688","display_name":"Image and Signal Denoising Methods","score":0.974399983882904,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11034","display_name":"Digital Filter Design and Implementation","score":0.9695000052452087,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T11019","display_name":"Image Enhancement Techniques","score":0.9598000049591064,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.661096453666687},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6405848860740662},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.6052402257919312},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6035829782485962},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.5958421230316162},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5459564924240112},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.42556607723236084},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.2973918318748474},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.27935001254081726},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.16521751880645752},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.1577664017677307},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.13221824169158936}],"concepts":[{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.661096453666687},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6405848860740662},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.6052402257919312},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6035829782485962},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.5958421230316162},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5459564924240112},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.42556607723236084},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2973918318748474},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.27935001254081726},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.16521751880645752},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.1577664017677307},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.13221824169158936},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3706628.3708849","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3706628.3708849","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 ACM/SIGDA International Symposium on Field Programmable Gate Arrays","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7","score":0.5400000214576721}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":2,"referenced_works":["https://openalex.org/W4384648639","https://openalex.org/W4409361965"],"related_works":["https://openalex.org/W2111241003","https://openalex.org/W2355315220","https://openalex.org/W4200391368","https://openalex.org/W4206178588","https://openalex.org/W3094491777","https://openalex.org/W3214715529","https://openalex.org/W4287635093","https://openalex.org/W1967938402","https://openalex.org/W2386041993","https://openalex.org/W1608572506"],"abstract_inverted_index":{"FPGAs":[0,141],"are":[1],"a":[2,69,88,112],"compelling":[3],"substrate":[4],"for":[5,22,38,102],"supporting":[6],"machine":[7,104],"learning":[8,105],"inference.":[9],"Tools":[10],"such":[11],"as":[12],"High-Level":[13],"Synthesis":[14],"and":[15,129],"hls4ml":[16],"can":[17,29,55,123],"shorten":[18],"the":[19,33,45],"development":[20],"cycle":[21],"deploying":[23],"ML":[24],"algorithms":[25],"on":[26,140],"FPGAs,":[27],"but":[28],"struggle":[30],"to":[31,73,86,136],"handle":[32],"large":[34],"on-chip":[35,95],"storage":[36],"needed":[37],"many":[39,51],"of":[40,52],"these":[41,53],"models.":[42],"In":[43,63],"particular":[44],"high":[46],"BRAM":[47,84,125],"usage":[48,126],"found":[49],"in":[50,111],"flows":[54],"cause":[56],"Place":[57],"&":[58],"Route":[59],"failures":[60],"during":[61],"synthesis.":[62],"this":[64,100],"paper":[65],"we":[66],"propose":[67],"using":[68],"Simulated-Annealing":[70],"based":[71],"flow":[72,101],"perform":[74],"BRAM-aware":[75],"quantization.":[76],"This":[77],"approach":[78],"trades":[79],"off":[80],"inference":[81,90],"accuracy":[82,120],"with":[83],"usage,":[85],"provide":[87],"high-quality":[89],"engine":[91],"that":[92],"still":[93],"meets":[94],"resource":[96],"constraints.":[97],"We":[98],"demonstrate":[99],"Transformer-based":[103,138],"algorithms,":[106],"which":[107],"include":[108],"Flash":[109],"Attention":[110],"Stream-based":[113],"Dataflow":[114],"architecture.":[115],"Our":[116],"system":[117],"imposes":[118],"minimal":[119],"drops,":[121],"yet":[122],"reduce":[124],"by":[127,133],"20%-50%,":[128],"improve":[130],"power":[131],"efficiency":[132],"264%-812%":[134],"compared":[135],"existing":[137],"accelerators":[139]},"counts_by_year":[],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-10-10T00:00:00"}