{"id":"https://openalex.org/W4393579178","doi":"https://doi.org/10.1145/3626202.3637585","title":"E4SA: An Ultra-Efficient Systolic Array Architecture for 4-Bit Convolutional Neural Networks","display_name":"E4SA: An Ultra-Efficient Systolic Array Architecture for 4-Bit Convolutional Neural Networks","publication_year":2024,"publication_date":"2024-04-01","ids":{"openalex":"https://openalex.org/W4393579178","doi":"https://doi.org/10.1145/3626202.3637585"},"language":"en","primary_location":{"id":"doi:10.1145/3626202.3637585","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3626202.3637585","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2024 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/A5100611279","display_name":"Geng Yang","orcid":"https://orcid.org/0000-0001-6921-1007"},"institutions":[{"id":"https://openalex.org/I149594827","display_name":"Xidian University","ror":"https://ror.org/05s92vm98","country_code":"CN","type":"education","lineage":["https://openalex.org/I149594827"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Geng Yang","raw_affiliation_strings":["Xidian University, Xi'an, Shaanxi, China"],"raw_orcid":"https://orcid.org/0000-0001-6921-1007","affiliations":[{"raw_affiliation_string":"Xidian University, Xi'an, Shaanxi, China","institution_ids":["https://openalex.org/I149594827"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5007285444","display_name":"Jie Lei","orcid":"https://orcid.org/0000-0003-0851-6565"},"institutions":[{"id":"https://openalex.org/I149594827","display_name":"Xidian University","ror":"https://ror.org/05s92vm98","country_code":"CN","type":"education","lineage":["https://openalex.org/I149594827"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jie Lei","raw_affiliation_strings":["Xidian University, Xi'an, China"],"raw_orcid":"https://orcid.org/0000-0003-0851-6565","affiliations":[{"raw_affiliation_string":"Xidian University, Xi'an, China","institution_ids":["https://openalex.org/I149594827"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065889904","display_name":"Zhenman Fang","orcid":"https://orcid.org/0000-0003-0603-9697"},"institutions":[{"id":"https://openalex.org/I18014758","display_name":"Simon Fraser University","ror":"https://ror.org/0213rcc28","country_code":"CA","type":"education","lineage":["https://openalex.org/I18014758"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Zhenman Fang","raw_affiliation_strings":["Simon Fraser University, Burnaby, British Columbia, Canada"],"raw_orcid":"https://orcid.org/0000-0003-0603-9697","affiliations":[{"raw_affiliation_string":"Simon Fraser University, Burnaby, British Columbia, Canada","institution_ids":["https://openalex.org/I18014758"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090736129","display_name":"Jiaqing Zhang","orcid":"https://orcid.org/0009-0007-3111-1116"},"institutions":[{"id":"https://openalex.org/I149594827","display_name":"Xidian University","ror":"https://ror.org/05s92vm98","country_code":"CN","type":"education","lineage":["https://openalex.org/I149594827"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiaqing Zhang","raw_affiliation_strings":["Xidian University, Xi'an, Shaanxi, China"],"raw_orcid":"https://orcid.org/0009-0007-3111-1116","affiliations":[{"raw_affiliation_string":"Xidian University, Xi'an, Shaanxi, China","institution_ids":["https://openalex.org/I149594827"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022541011","display_name":"J. C. Zhang","orcid":"https://orcid.org/0009-0001-6828-9758"},"institutions":[{"id":"https://openalex.org/I149594827","display_name":"Xidian University","ror":"https://ror.org/05s92vm98","country_code":"CN","type":"education","lineage":["https://openalex.org/I149594827"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Junrong Zhang","raw_affiliation_strings":["Xidian University, Xi'an, Shaanxi, China"],"raw_orcid":"https://orcid.org/0009-0001-6828-9758","affiliations":[{"raw_affiliation_string":"Xidian University, Xi'an, Shaanxi, China","institution_ids":["https://openalex.org/I149594827"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052163069","display_name":"Weiying Xie","orcid":"https://orcid.org/0000-0001-8310-024X"},"institutions":[{"id":"https://openalex.org/I149594827","display_name":"Xidian University","ror":"https://ror.org/05s92vm98","country_code":"CN","type":"education","lineage":["https://openalex.org/I149594827"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Weiying Xie","raw_affiliation_strings":["Xidian University, Xi'an, Shaanxi, China"],"raw_orcid":"https://orcid.org/0000-0001-8310-024X","affiliations":[{"raw_affiliation_string":"Xidian University, Xi'an, Shaanxi, China","institution_ids":["https://openalex.org/I149594827"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5067798266","display_name":"Yunsong Li","orcid":"https://orcid.org/0000-0002-0234-6270"},"institutions":[{"id":"https://openalex.org/I149594827","display_name":"Xidian University","ror":"https://ror.org/05s92vm98","country_code":"CN","type":"education","lineage":["https://openalex.org/I149594827"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yunsong Li","raw_affiliation_strings":["Xidian University, Xi'an, Shaanxi, China"],"raw_orcid":"https://orcid.org/0000-0002-0234-6270","affiliations":[{"raw_affiliation_string":"Xidian University, Xi'an, Shaanxi, China","institution_ids":["https://openalex.org/I149594827"]}]}],"institutions":[],"countries_distinct_count":2,"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":0.03195096,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"183","last_page":"183"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.9599999785423279,"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/T10320","display_name":"Neural Networks and Applications","score":0.9599999785423279,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9484000205993652,"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/computer-science","display_name":"Computer science","score":0.7750558853149414},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.5956019163131714},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.5805671215057373},{"id":"https://openalex.org/keywords/digital-signal-processing","display_name":"Digital signal processing","score":0.5363949537277222},{"id":"https://openalex.org/keywords/systolic-array","display_name":"Systolic array","score":0.523993194103241},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.4699808657169342},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.46976611018180847},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.4499826431274414},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.43635204434394836},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.43557652831077576},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.41825366020202637},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.35172736644744873},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.27460265159606934},{"id":"https://openalex.org/keywords/very-large-scale-integration","display_name":"Very-large-scale integration","score":0.2678975462913513},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.1029089093208313}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7750558853149414},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.5956019163131714},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5805671215057373},{"id":"https://openalex.org/C84462506","wikidata":"https://www.wikidata.org/wiki/Q173142","display_name":"Digital signal processing","level":2,"score":0.5363949537277222},{"id":"https://openalex.org/C150741067","wikidata":"https://www.wikidata.org/wiki/Q2377218","display_name":"Systolic array","level":3,"score":0.523993194103241},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.4699808657169342},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.46976611018180847},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.4499826431274414},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.43635204434394836},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.43557652831077576},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.41825366020202637},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.35172736644744873},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.27460265159606934},{"id":"https://openalex.org/C14580979","wikidata":"https://www.wikidata.org/wiki/Q876049","display_name":"Very-large-scale integration","level":2,"score":0.2678975462913513},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.1029089093208313},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.0},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3626202.3637585","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3626202.3637585","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2024 ACM/SIGDA International Symposium on Field Programmable Gate Arrays","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/8","display_name":"Decent work and economic growth","score":0.5699999928474426}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4240320454","https://openalex.org/W2070314832","https://openalex.org/W2395557210","https://openalex.org/W4319431385","https://openalex.org/W3204400881","https://openalex.org/W3214410901","https://openalex.org/W3204296682","https://openalex.org/W3183118997","https://openalex.org/W2917767146","https://openalex.org/W4283703276"],"abstract_inverted_index":{"Many":[0],"studies":[1],"have":[2],"demonstrated":[3],"that":[4,32,112,177,224],"4-bit":[5,24,53,67,99,119,243],"precision":[6],"quantization":[7],"can":[8],"achieve":[9],"comparable":[10],"accuracy":[11],"to":[12,83,141,154,198,228,242],"floating-point":[13],"DNNs,":[14,22],"sparking":[15],"significant":[16],"interest":[17],"in":[18,65,194],"efficiently":[19],"accelerating":[20],"compressed":[21],"especially":[23,80],"convolutions,":[25],"on":[26,126],"edge":[27],"devices.":[28],"However,":[29],"we":[30,104,128,171],"observe":[31],"conventional":[33],"systolic":[34],"array":[35],"(SA)":[36],"architectures":[37,60],"designed":[38],"for":[39,98],"DNNs":[40],"cannot":[41],"fully":[42],"exploit":[43],"the":[44,115,179,195,208],"advantages":[45],"of":[46,118,136,160],"high":[47],"DSP":[48,54,68,120],"computational":[49,143],"density":[50],"offered":[51],"by":[52],"packing.":[55],"Although":[56],"state-of-the-art":[57],"FPGA-based":[58],"SA":[59,95,132,176,182],"(e.g.,":[61],"AutoSA)":[62],"exhibit":[63],"flexibility":[64],"accommodating":[66],"packing,":[69],"they":[70],"suffer":[71],"from":[72],"resource":[73],"consumption":[74],"and":[75,93,150,202,205,236],"data":[76,157,200],"supply":[77],"latency":[78,215],"issues,":[79],"when":[81],"adapting":[82],"various":[84,217],"convolution":[85],"spatial":[86,219],"sizes.":[87,220],"This":[88,145],"work":[89],"introduces":[90],"a":[91,106,130,173],"customizable":[92],"ultra-efficient":[94],"architectural":[96],"template":[97],"convolution,":[100],"called":[101],"E4SA.":[102],"First,":[103],"propose":[105],"fine-grained":[107],"row-temporal":[108],"weight":[109],"stationary":[110],"dataflow":[111],"aligns":[113],"with":[114,231],"specific":[116],"requirements":[117],"full":[121],"packing":[122],"(4bF":[123],"packing).":[124],"Based":[125],"this,":[127],"design":[129],"cost-effective":[131,167],"unit":[133],"(SAU)":[134],"composed":[135],"4bF-packing-based":[137],"processing":[138],"elements":[139],"(PEs)":[140],"enhance":[142],"efficiency.":[144],"includes":[146],"column-shared":[147],"packed-data":[148],"splitters":[149],"shift-register-based":[151],"feature-map/weight":[152],"fetchers":[153],"ensure":[155],"continuous":[156],"supply,":[158],"all":[159],"which":[161,188],"are":[162],"locally":[163],"interconnected":[164],"via":[165],"more":[166],"registers.":[168],"In":[169],"addition,":[170],"develop":[172],"two-level":[174],"hierarchy":[175],"decomposes":[178],"original":[180],"large":[181],"into":[183],"parallel":[184],"4\u00d74":[185],"SAU":[186],"sets,":[187],"not":[189],"only":[190],"allows":[191],"multiple":[192],"PEs":[193],"same":[196],"column":[197],"share":[199],"splitting":[201],"reorganization":[203],"logic":[204],"thus":[206],"reducing":[207],"LUT":[209],"overhead,":[210],"but":[211],"also":[212],"maintains":[213],"near-theoretical":[214],"across":[216],"convolutional":[218],"Experimental":[221],"results":[222],"demonstrate":[223],"E4SA":[225],"achieves":[226],"up":[227],"576.6":[229],"GOPS":[230],"13.8\u00d7":[232],"higher":[233,238],"GOPS/DSP":[234],"efficiency":[235,240],"51.6\u00d7":[237],"GOPS/kLUTs":[239],"compared":[241],"AutoSA-based":[244],"design.":[245]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}