{"id":"https://openalex.org/W4404134154","doi":"https://doi.org/10.1145/3649329.3656507","title":"FLAME: Fully Leveraging MoE Sparsity for Transformer on FPGA","display_name":"FLAME: Fully Leveraging MoE Sparsity for Transformer on FPGA","publication_year":2024,"publication_date":"2024-06-23","ids":{"openalex":"https://openalex.org/W4404134154","doi":"https://doi.org/10.1145/3649329.3656507"},"language":"en","primary_location":{"id":"doi:10.1145/3649329.3656507","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649329.3656507","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/A5113058728","display_name":"Xuanda Lin","orcid":"https://orcid.org/0009-0009-6355-2439"},"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":"Xuanda Lin","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0009-6355-2439","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022672817","display_name":"Huinan Tian","orcid":"https://orcid.org/0009-0002-3887-6852"},"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":"Huinan Tian","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0002-3887-6852","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5108103373","display_name":"Wu Xue","orcid":"https://orcid.org/0009-0001-1394-9149"},"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":"Wenxiao Xue","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0001-1394-9149","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113808220","display_name":"L. Ma","orcid":"https://orcid.org/0009-0002-9697-4442"},"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":"Lanqi Ma","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0002-9697-4442","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/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/A5103098516","display_name":"Jun Yu","orcid":"https://orcid.org/0000-0003-4286-9292"},"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":"Jun Yu","raw_affiliation_strings":["Fudan University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0003-4286-9292","affiliations":[{"raw_affiliation_string":"Fudan University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"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":["https://openalex.org/I24943067"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":8,"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/T11343","display_name":"Power Transformer Diagnostics and Insulation","score":0.9936000108718872,"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"}},"topics":[{"id":"https://openalex.org/T11343","display_name":"Power Transformer Diagnostics and Insulation","score":0.9936000108718872,"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/T10363","display_name":"Low-power high-performance VLSI design","score":0.9876999855041504,"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/T10320","display_name":"Neural Networks and Applications","score":0.9861000180244446,"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/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.7160024642944336},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6373021602630615},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5025300979614258},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.3426561951637268},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.3355245292186737},{"id":"https://openalex.org/keywords/reliability-engineering","display_name":"Reliability engineering","score":0.33500105142593384},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.1858132779598236},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.17135295271873474}],"concepts":[{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.7160024642944336},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6373021602630615},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5025300979614258},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3426561951637268},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.3355245292186737},{"id":"https://openalex.org/C200601418","wikidata":"https://www.wikidata.org/wiki/Q2193887","display_name":"Reliability engineering","level":1,"score":0.33500105142593384},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.1858132779598236},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.17135295271873474}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3649329.3656507","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649329.3656507","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":[],"awards":[{"id":"https://openalex.org/G6507269736","display_name":null,"funder_award_id":"2021YFA1003602","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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":6,"referenced_works":["https://openalex.org/W2150884987","https://openalex.org/W2989743967","https://openalex.org/W3100560913","https://openalex.org/W3162542754","https://openalex.org/W4291653336","https://openalex.org/W4388214757"],"related_works":["https://openalex.org/W2748952813","https://openalex.org/W2111241003","https://openalex.org/W2355315220","https://openalex.org/W4200391368","https://openalex.org/W2210979487","https://openalex.org/W2096844293","https://openalex.org/W2363944576","https://openalex.org/W2351041855","https://openalex.org/W2570254841","https://openalex.org/W2864363823"],"abstract_inverted_index":{"MoE":[0,27,51,56,87,113,120,187],"(Mixture-of-Experts)":[1],"mechanism":[2,28],"has":[3],"been":[4],"widely":[5],"adopted":[6],"in":[7,89],"transformer-based":[8,90],"models":[9],"to":[10,40,73,95,117,129,153,173],"facilitate":[11],"further":[12],"expansion":[13],"of":[14,26,144,192,208],"model":[15,150],"parameter":[16],"size":[17],"and":[18,44,99,221,226],"enhance":[19],"generalization":[20],"capabilities.":[21],"However,":[22],"the":[23,50,55,109,142,177,190],"practical":[24],"deployment":[25,125],"for":[29,86,122,141],"transformer":[30,124],"on":[31,126],"resource-constrained":[32],"platforms,":[33],"such":[34],"as":[35],"FPGA,":[36],"remains":[37],"challenging":[38],"due":[39],"heavy":[41,67],"memory":[42],"footprints":[43],"impractical":[45],"runtime":[46],"costs":[47],"introduced":[48],"by":[49,102],"mechanism.":[52,199],"Diving":[53],"into":[54],"mechanism,":[57],"we":[58,106,134,158,184],"raise":[59],"two":[60,104,213],"key":[61],"observations:":[62],"(1)":[63],"Expert":[64],"weights":[65,146,169],"are":[66],"but":[68],"cold,":[69],"making":[70,92],"it":[71,93],"ideal":[72],"leverage":[74,119,130],"expert":[75,83,97,131,145,155,162,168,179,196,209,214],"weight":[76,132],"sparsity.":[77],"(2)":[78],"There":[79],"exists":[80],"highly":[81],"skewed":[82],"activation":[84,156],"paths":[85],"layers":[88],"models,":[91],"feasible":[94],"conduct":[96],"prediction":[98,163,210],"prefetching.":[100],"Motivated":[101],"these":[103],"observations,":[105],"propose":[107,159],"FLAME,":[108],"first":[110],"algorithm-hardware":[111],"co-optimized":[112],"accelerating":[114],"framework":[115],"designed":[116],"fully":[118],"sparsity":[121,188],"efficient":[123,194],"FPGA.":[127],"First,":[128],"sparsity,":[133,157],"integrate":[135],"an":[136,193],"N:M":[137],"pruning":[138,143],"algorithm,":[139],"allowing":[140],"without":[147],"significantly":[148],"compromising":[149],"accuracy.":[151],"Second,":[152],"settle":[154],"a":[160],"circular":[161],"(CEPR)":[164],"strategy.":[165],"CEPR":[166],"prefetches":[167],"from":[170],"external":[171],"storage":[172],"on-chip":[174],"cache":[175],"before":[176],"activated":[178],"index":[180],"is":[181],"determined.":[182],"Last,":[183],"co-optimize":[185],"both":[186],"through":[189],"introduction":[191],"pruning-aware":[195],"buffering":[197],"(PA-BUF)":[198],"Experimental":[200],"results":[201],"demonstrate":[202],"that":[203],"FLAME":[204,223],"achieves":[205,224],"84.4%":[206],"accuracy":[207],"with":[211,219],"merely":[212],"caches":[215],"on-chip.":[216],"In":[217],"comparison":[218],"CPU":[220],"GPU,":[222],"4.12\u00d7":[225],"1.49\u00d7":[227],"speedup,":[228],"respectively.":[229]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":7}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}