{"id":"https://openalex.org/W4404133735","doi":"https://doi.org/10.1145/3649329.3656497","title":"FNM-Trans: Efficient FPGA-based Transformer Architecture with Full N:M Sparsity","display_name":"FNM-Trans: Efficient FPGA-based Transformer Architecture with Full N:M Sparsity","publication_year":2024,"publication_date":"2024-06-23","ids":{"openalex":"https://openalex.org/W4404133735","doi":"https://doi.org/10.1145/3649329.3656497"},"language":"en","primary_location":{"id":"doi:10.1145/3649329.3656497","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649329.3656497","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 61st ACM/IEEE Design Automation Conference","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/A5113426367","display_name":"Manting Zhang","orcid":"https://orcid.org/0009-0007-7437-6587"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Manting Zhang","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0007-7437-6587","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5072580709","display_name":"Jialin Cao","orcid":"https://orcid.org/0009-0001-9054-5500"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jialin Cao","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0001-9054-5500","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102312689","display_name":"Kejia Shi","orcid":"https://orcid.org/0009-0001-8471-0995"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kejia Shi","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0001-8471-0995","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027433777","display_name":"Keqing Zhao","orcid":"https://orcid.org/0009-0001-4951-4059"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Keqing Zhao","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0001-4951-4059","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049836780","display_name":"Genhao Zhang","orcid":"https://orcid.org/0000-0002-9834-6477"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Genhao Zhang","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0000-3863-981X","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103098516","display_name":"Jun Yu","orcid":"https://orcid.org/0000-0003-4286-9292"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jun Yu","raw_affiliation_strings":["Fudan Universiy, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0003-4286-9292","affiliations":[{"raw_affiliation_string":"Fudan Universiy, Shanghai, Shanghai, China","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5015932760","display_name":"Kun Wang","orcid":"https://orcid.org/0000-0002-7288-1789"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kun Wang","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0002-7288-1789","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":10,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"6"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10468","display_name":"Photovoltaic System Optimization Techniques","score":0.9951000213623047,"subfield":{"id":"https://openalex.org/subfields/2105","display_name":"Renewable Energy, Sustainability and the Environment"},"field":{"id":"https://openalex.org/fields/21","display_name":"Energy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10468","display_name":"Photovoltaic System Optimization Techniques","score":0.9951000213623047,"subfield":{"id":"https://openalex.org/subfields/2105","display_name":"Renewable Energy, Sustainability and the Environment"},"field":{"id":"https://openalex.org/fields/21","display_name":"Energy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10363","display_name":"Low-power high-performance VLSI design","score":0.9940000176429749,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10323","display_name":"Analog and Mixed-Signal Circuit Design","score":0.988099992275238,"subfield":{"id":"https://openalex.org/subfields/2204","display_name":"Biomedical Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"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":0.7120055556297302},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.6453120112419128},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.634009063243866},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6119994521141052},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.4407428801059723},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.37393832206726074},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.36459794640541077},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.17080223560333252},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.1635763943195343},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.0624108612537384}],"concepts":[{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.7120055556297302},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.6453120112419128},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.634009063243866},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6119994521141052},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4407428801059723},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.37393832206726074},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.36459794640541077},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.17080223560333252},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1635763943195343},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0624108612537384},{"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}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3649329.3656497","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649329.3656497","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 61st ACM/IEEE Design Automation Conference","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.6600000262260437,"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7"}],"awards":[{"id":"https://openalex.org/G397489182","display_name":null,"funder_award_id":"2021YFA100-3602","funder_id":"https://openalex.org/F4320335777","funder_display_name":"National Key Research and Development Program of China"}],"funders":[{"id":"https://openalex.org/F4320335777","display_name":"National Key Research and Development Program of China","ror":null}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":13,"referenced_works":["https://openalex.org/W2970971581","https://openalex.org/W3162542754","https://openalex.org/W3196923642","https://openalex.org/W3213174544","https://openalex.org/W4291653336","https://openalex.org/W4293025165","https://openalex.org/W4295312788","https://openalex.org/W4312569226","https://openalex.org/W4321466385","https://openalex.org/W4360606782","https://openalex.org/W4385245566","https://openalex.org/W4388214757","https://openalex.org/W4389166707"],"related_works":["https://openalex.org/W2111241003","https://openalex.org/W2355315220","https://openalex.org/W4200391368","https://openalex.org/W2210979487","https://openalex.org/W2316202402","https://openalex.org/W2074043759","https://openalex.org/W1967938402","https://openalex.org/W2386041993","https://openalex.org/W1608572506","https://openalex.org/W2038503502"],"abstract_inverted_index":{"Transformer":[0,58,85],"models":[1],"have":[2],"become":[3],"popular":[4],"in":[5,37],"various":[6],"AI":[7],"applications":[8],"due":[9],"to":[10,47,52,134,162,189],"their":[11,15],"exceptional":[12],"performance.":[13],"However,":[14,43],"impressive":[16],"performance":[17],"comes":[18],"with":[19,105],"significant":[20],"computing":[21,127],"and":[22,40,63,73,129,157,171,176,182],"memory":[23],"costs,":[24],"hindering":[25],"efficient":[26,74],"deployment":[27],"of":[28,57,83,102,138,169],"Transformer-based":[29],"applications.":[30],"Many":[31],"solutions":[32],"focus":[33],"on":[34],"leveraging":[35],"sparsity":[36,156],"weight":[38,159],"matrix":[39],"attention":[41,61,155],"computation.":[42],"previous":[44],"studies":[45],"fail":[46],"exploit":[48],"unified":[49],"sparse":[50],"pattern":[51],"accelerate":[53],"all":[54,80],"three":[55,81],"modules":[56,82],"(QKV":[59],"generation,":[60],"computation":[62],"FFN).":[64],"In":[65],"this":[66],"paper,":[67],"we":[68,97,118],"propose":[69],"FNM-Trans,":[70],"an":[71],"adaptable":[72],"algorithm-hardware":[75],"co-design":[76],"aimed":[77],"at":[78],"optimizing":[79],"the":[84,94,100,115],"by":[86,151,187],"fully":[87,98],"harnessing":[88],"N":[89,110,139],":":[90,111,140],"M":[91,112,141],"sparsity.":[92,113,142],"At":[93,114],"algorithm":[95,148],"level,":[96,117],"explore":[99],"interplay":[101],"dynamic":[103],"pruning":[104,107],"static":[106],"under":[108,153],"high":[109],"hardware":[116,122],"develop":[119],"a":[120,125,130],"dedicated":[121],"architecture":[123],"featuring":[124],"custom":[126],"engine":[128],"softmax":[131],"module,":[132],"tailored":[133],"support":[135],"varying":[136],"levels":[137],"Experiment":[143],"results":[144],"show":[145],"that,":[146],"our":[147],"optimizes":[149],"accuracy":[150],"11.03%":[152],"2:16":[154],"4:16":[158],"sparsity,":[160],"compared":[161],"other":[163],"methods.":[164],"Additionally,":[165],"FNM-Trans":[166],"achieves":[167],"speedups":[168],"27.13\u00d7":[170],"21.24\u00d7":[172],"over":[173],"Intel":[174],"i9-9900X":[175],"NVIDIA":[177],"RTX":[178],"2080":[179],"Ti,":[180],"respectively,":[181],"outpaces":[183],"current":[184],"FPGA-based":[185],"Transformers":[186],"1.88\u00d7":[188],"36.51\u00d7.":[190]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":9}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}