{"id":"https://openalex.org/W7163523980","doi":"https://doi.org/10.23919/date69613.2026.11539329","title":"AO-BFP: An Adaptive Mixed-Precision and Outlier-Aware Block Floating-Point Accelerator for Large Language Model Inference","display_name":"AO-BFP: An Adaptive Mixed-Precision and Outlier-Aware Block Floating-Point Accelerator for Large Language Model Inference","publication_year":2026,"publication_date":"2026-04-20","ids":{"openalex":"https://openalex.org/W7163523980","doi":"https://doi.org/10.23919/date69613.2026.11539329"},"language":null,"primary_location":{"id":"doi:10.23919/date69613.2026.11539329","is_oa":false,"landing_page_url":"https://doi.org/10.23919/date69613.2026.11539329","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 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/A5137853857","display_name":"Yifan Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yifan Wang","raw_affiliation_strings":["University of Science and Technology of China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023414459","display_name":"Zetao Guo","orcid":"https://orcid.org/0000-0001-9264-7677"},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zetao Guo","raw_affiliation_strings":["University of Science and Technology of China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101669107","display_name":"Wendi Sun","orcid":"https://orcid.org/0009-0002-3198-2345"},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wendi Sun","raw_affiliation_strings":["University of Science and Technology of China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5106636992","display_name":"Wenhao Sun","orcid":null},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenhao Sun","raw_affiliation_strings":["University of Science and Technology of China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137904802","display_name":"Qiyan Fang","orcid":null},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qiyan Fang","raw_affiliation_strings":["University of Science and Technology of China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100423614","display_name":"Song Chen","orcid":"https://orcid.org/0000-0003-0341-3428"},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Song Chen","raw_affiliation_strings":["University of Science and Technology of China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5137913228","display_name":"Yi Kang","orcid":null},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yi Kang","raw_affiliation_strings":["University of Science and Technology of China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China","institution_ids":["https://openalex.org/I126520041"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":7,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.91335209,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"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/T10028","display_name":"Topic Modeling","score":0.20640000700950623,"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/T10028","display_name":"Topic Modeling","score":0.20640000700950623,"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"}},{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.12600000202655792,"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"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.066600002348423,"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/block","display_name":"Block (permutation group theory)","score":0.5196999907493591},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.392300009727478},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.35120001435279846},{"id":"https://openalex.org/keywords/scheme","display_name":"Scheme (mathematics)","score":0.271699994802475},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.2655999958515167}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6687999963760376},{"id":"https://openalex.org/C2777210771","wikidata":"https://www.wikidata.org/wiki/Q4927124","display_name":"Block (permutation group theory)","level":2,"score":0.5196999907493591},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.45820000767707825},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3955000042915344},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.392300009727478},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.35120001435279846},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3156000077724457},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.27619999647140503},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.271699994802475},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2655999958515167},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2648000121116638},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.25999999046325684}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.23919/date69613.2026.11539329","is_oa":false,"landing_page_url":"https://doi.org/10.23919/date69613.2026.11539329","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 Design, Automation & Test in Europe Conference (DATE)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320321133","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35"},{"id":"https://openalex.org/F4320325599","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":15,"referenced_works":["https://openalex.org/W2123736966","https://openalex.org/W2725159389","https://openalex.org/W2982041622","https://openalex.org/W3006586535","https://openalex.org/W4221149185","https://openalex.org/W4308083739","https://openalex.org/W4366341968","https://openalex.org/W4386764384","https://openalex.org/W4389519100","https://openalex.org/W4402670692","https://openalex.org/W4404783683","https://openalex.org/W4408399636","https://openalex.org/W4409248709","https://openalex.org/W4410582506","https://openalex.org/W4410583578"],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1,10],"Models":[2],"(LLMs)":[3],"have":[4],"achieved":[5],"remarkable":[6],"success":[7],"in":[8,85],"Natural":[9],"Processing":[11],"(NLP)":[12],"tasks,":[13],"but":[14],"their":[15],"deployment":[16],"is":[17],"severely":[18],"constrained":[19],"by":[20,117],"intensive":[21],"computation":[22],"and":[23,41,71,74,150,165],"memory":[24,40],"costs.":[25],"Block":[26],"Floating-Point":[27],"(BFP)":[28],"extends":[29],"the":[30,75,101,122],"dynamic":[31],"range":[32],"beyond":[33],"INT":[34],"with":[35,113,131,142],"shared":[36],"exponents,":[37],"while":[38,156],"reducing":[39],"alignment":[42],"overhead":[43],"compared":[44],"to":[45,57,81],"floating-point":[46],"formats.":[47],"However,":[48],"when":[49],"bit-widths":[50],"are":[51],"further":[52],"reduced,":[53],"BFP":[54,61,95],"becomes":[55],"sensitive":[56],"outliers;":[58],"existing":[59],"mixed-precision":[60,114],"methods":[62],"largely":[63],"rely":[64],"on":[65],"heuristic":[66],"settings":[67],"of":[68,162],"mantissa":[69],"width":[70],"block":[72],"size,":[73],"induced":[76],"bit-level":[77],"sparsity":[78],"has":[79],"yet":[80],"be":[82],"systematically":[83],"leveraged":[84],"hardware.":[86],"In":[87],"this":[88],"paper,":[89],"we":[90,104,125],"propose":[91,105],"AO-BFP,":[92],"an":[93,106],"adaptive":[94,107],"framework":[96],"for":[97],"LLM":[98,144],"inference.":[99],"At":[100,121],"algorithm":[102],"level,":[103,124],"outlier":[108],"exponent":[109],"mapping":[110],"mechanism":[111],"combined":[112],"exploration":[115],"driven":[116],"layer-wise":[118],"sensitivity":[119],"analysis.":[120],"hardware":[123],"design":[126],"a":[127,132],"reconfigurable":[128],"bit-serial":[129],"accelerator":[130],"unified":[133],"datapath":[134],"that":[135],"efficiently":[136],"leverages":[137],"BFP-induced":[138],"bit":[139],"sparsity.":[140],"Compared":[141],"prior":[143],"accelerators":[145],"such":[146],"as":[147],"ANT,":[148],"OliVe,":[149],"BitMoD,":[151],"AO-BFP":[152],"achieves":[153],"superior":[154],"performance":[155],"preserving":[157],"model":[158],"accuracy,":[159],"delivering":[160],"speedups":[161],"1.61\u00d7,":[163],"1.39\u00d7,":[164],"1.11\u00d7,":[166],"respectively.":[167]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-06-05T00:00:00"}