{"id":"https://openalex.org/W3205192296","doi":"https://doi.org/10.1109/tcad.2022.3170848","title":"Energon: Toward Efficient Acceleration of Transformers Using Dynamic Sparse Attention","display_name":"Energon: Toward Efficient Acceleration of Transformers Using Dynamic Sparse Attention","publication_year":2022,"publication_date":"2022-04-27","ids":{"openalex":"https://openalex.org/W3205192296","doi":"https://doi.org/10.1109/tcad.2022.3170848","mag":"3205192296"},"language":"en","primary_location":{"id":"doi:10.1109/tcad.2022.3170848","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2022.3170848","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-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/A5001078878","display_name":"Zhe Zhou","orcid":"https://orcid.org/0000-0001-7929-8054"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhe Zhou","raw_affiliation_strings":["Center for Energy-Efficient Computing and Applications, Peking University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-7929-8054","affiliations":[{"raw_affiliation_string":"Center for Energy-Efficient Computing and Applications, Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101880137","display_name":"Junlin Liu","orcid":"https://orcid.org/0000-0001-5917-2251"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Junlin Liu","raw_affiliation_strings":["Center for Energy-Efficient Computing and Applications, Peking University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-5917-2251","affiliations":[{"raw_affiliation_string":"Center for Energy-Efficient Computing and Applications, Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005095014","display_name":"Zhenyu Gu","orcid":"https://orcid.org/0000-0003-3921-5837"},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhenyu Gu","raw_affiliation_strings":["DAMO Academy, Alibaba Inc., Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"DAMO Academy, Alibaba Inc., Beijing, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101850376","display_name":"Guangyu Sun","orcid":"https://orcid.org/0000-0001-6436-0820"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guangyu Sun","raw_affiliation_strings":["Center for Energy-Efficient Computing and Applications, Peking University, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Center for Energy-Efficient Computing and Applications, Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":4.6004,"has_fulltext":false,"cited_by_count":58,"citation_normalized_percentile":{"value":0.95974991,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":100},"biblio":{"volume":"42","issue":"1","first_page":"136","last_page":"149"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9987000226974487,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9987000226974487,"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/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.9984999895095825,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9983000159263611,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.6097109317779541},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5661836862564087},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5657111406326294},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.19490313529968262},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.15528005361557007},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.15009057521820068},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.05727449059486389}],"concepts":[{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.6097109317779541},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5661836862564087},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5657111406326294},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.19490313529968262},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.15528005361557007},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.15009057521820068},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.05727449059486389},{"id":"https://openalex.org/C74650414","wikidata":"https://www.wikidata.org/wiki/Q11397","display_name":"Classical mechanics","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tcad.2022.3170848","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2022.3170848","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Affordable and clean energy","score":0.8999999761581421,"id":"https://metadata.un.org/sdg/7"}],"awards":[{"id":"https://openalex.org/G4460768019","display_name":null,"funder_award_id":"62032001","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G4602124056","display_name":null,"funder_award_id":"92064006","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6847781164","display_name":null,"funder_award_id":"B18001","funder_id":"https://openalex.org/F4320327912","funder_display_name":"Higher Education Discipline Innovation Project"},{"id":"https://openalex.org/G7584325168","display_name":null,"funder_award_id":"61832020","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320327912","display_name":"Higher Education Discipline Innovation Project","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":101,"referenced_works":["https://openalex.org/W179875071","https://openalex.org/W1983394510","https://openalex.org/W2034861439","https://openalex.org/W2042519306","https://openalex.org/W2048266589","https://openalex.org/W2067523571","https://openalex.org/W2108598243","https://openalex.org/W2119144962","https://openalex.org/W2152839228","https://openalex.org/W2194775991","https://openalex.org/W2285660444","https://openalex.org/W2462831000","https://openalex.org/W2525332836","https://openalex.org/W2565851976","https://openalex.org/W2606722458","https://openalex.org/W2625457103","https://openalex.org/W2799054028","https://openalex.org/W2896457183","https://openalex.org/W2899663614","https://openalex.org/W2904902077","https://openalex.org/W2906043559","https://openalex.org/W2911491685","https://openalex.org/W2923014074","https://openalex.org/W2931316642","https://openalex.org/W2940744433","https://openalex.org/W2950656546","https://openalex.org/W2953212265","https://openalex.org/W2962835968","https://openalex.org/W2963341956","https://openalex.org/W2963563276","https://openalex.org/W2963748441","https://openalex.org/W2964299589","https://openalex.org/W2965373594","https://openalex.org/W2970971581","https://openalex.org/W2973061659","https://openalex.org/W2979314664","https://openalex.org/W2981851019","https://openalex.org/W2996428491","https://openalex.org/W2998108143","https://openalex.org/W2998183051","https://openalex.org/W3006983028","https://openalex.org/W3012027465","https://openalex.org/W3012126539","https://openalex.org/W3015468748","https://openalex.org/W3016939927","https://openalex.org/W3017024317","https://openalex.org/W3034999214","https://openalex.org/W3035251378","https://openalex.org/W3043504674","https://openalex.org/W3085139254","https://openalex.org/W3092462694","https://openalex.org/W3094502228","https://openalex.org/W3096609285","https://openalex.org/W3098576111","https://openalex.org/W3099513244","https://openalex.org/W3100985894","https://openalex.org/W3104393472","https://openalex.org/W3106298483","https://openalex.org/W3111747337","https://openalex.org/W3118608800","https://openalex.org/W3122239467","https://openalex.org/W3131922516","https://openalex.org/W3159727696","https://openalex.org/W3170874841","https://openalex.org/W3172509117","https://openalex.org/W3177265267","https://openalex.org/W3177318507","https://openalex.org/W3207338936","https://openalex.org/W3214431650","https://openalex.org/W4200559829","https://openalex.org/W4240168186","https://openalex.org/W4247470470","https://openalex.org/W4288089799","https://openalex.org/W4295312788","https://openalex.org/W4323654151","https://openalex.org/W6634868918","https://openalex.org/W6637373629","https://openalex.org/W6677580257","https://openalex.org/W6679915538","https://openalex.org/W6684191040","https://openalex.org/W6727099177","https://openalex.org/W6739901393","https://openalex.org/W6755207826","https://openalex.org/W6755977528","https://openalex.org/W6758823024","https://openalex.org/W6761628794","https://openalex.org/W6763701032","https://openalex.org/W6766673545","https://openalex.org/W6766978945","https://openalex.org/W6768021236","https://openalex.org/W6769627184","https://openalex.org/W6771626834","https://openalex.org/W6772372951","https://openalex.org/W6772853553","https://openalex.org/W6774418012","https://openalex.org/W6776048684","https://openalex.org/W6784094891","https://openalex.org/W6784333009","https://openalex.org/W6786756943","https://openalex.org/W6787972765","https://openalex.org/W6788135285"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"In":[0],"recent":[1],"years,":[2],"transformer":[3],"models":[4],"have":[5],"revolutionized":[6],"natural":[7],"language":[8],"processing":[9],"(NLP)":[10],"and":[11,36,104,138,151,158,167,181,191,201,216,227,245,254,262],"shown":[12],"promising":[13],"performance":[14,157],"on":[15,44,75,164],"computer":[16],"vision":[17],"(CV)":[18],"tasks.":[19],"Despite":[20],"their":[21],"effectiveness,":[22],"transformers\u2019":[23],"attention":[24,71,111,224],"operations":[25],"are":[26],"hard":[27],"to":[28,31,89,113,193,222],"accelerate":[29],"due":[30],"the":[32,41,68,110,123,156],"complicated":[33],"data":[34],"movement":[35],"quadratic":[37],"computational":[38,124],"complexity,":[39],"prohibiting":[40],"real-time":[42],"inference":[43],"resource-constrained":[45],"edge-computing":[46],"platforms.":[47],"To":[48,130],"tackle":[49],"this":[50,118],"challenge,":[51],"we":[52,81,120,142],"propose":[53,82,144],"Energon,":[54],"an":[55,133,145],"algorithm-architecture":[56],"co-design":[57],"approach":[58],"that":[59,70,171],"accelerates":[60],"various":[61],"transformers":[62],"using":[63],"dynamic":[64],"sparse":[65],"attention.":[66],"With":[67],"observation":[69],"results":[72],"only":[73,105],"depend":[74],"a":[76,83],"few":[77],"important":[78],"query-key":[79],"pairs,":[80],"mix-precision":[84],"multiround":[85],"filtering":[86,102],"(MP-MRF)":[87],"algorithm":[88,134],"dynamically":[90],"identify":[91],"such":[92,132],"pairs":[93],"at":[94],"runtime.":[95],"We":[96],"adopt":[97],"low":[98],"bitwidth":[99],"in":[100,109],"each":[101],"round":[103],"use":[106],"high-precision":[107],"tensors":[108],"stage":[112],"reduce":[114,159],"overall":[115],"complexity.":[116],"By":[117],"means,":[119],"significantly":[121],"mitigate":[122],"cost":[125],"with":[126,135],"negligible":[127],"accuracy":[128],"loss.":[129],"enable":[131],"lower":[136],"latency":[137],"better":[139],"energy":[140,209,271],"efficiency,":[141],"also":[143,236],"Energon":[146,172,235],"co-processor":[147],"architecture.":[148],"Elaborated":[149],"pipelines":[150],"specialized":[152],"optimizations":[153],"jointly":[154],"boost":[155],"power":[160],"consumption.":[161],"Extensive":[162],"experiments":[163],"both":[165],"NLP":[166],"CV":[168],"benchmarks":[169],"demonstrate":[170],"achieves":[173,237],"<inline-formula":[174,182,194,202,228,238,246,255,263],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[175,183,195,203,229,239,247,256,264],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[176,184,196,204,230,240,248,257,265],"<tex-math":[177,185,197,205,231,241,249,258,266],"notation=\"LaTeX\">$168\\times":[178],"$":[179,187,199,207,243,251,260,268],"</tex-math></inline-formula>":[180,188,200,208,233,244,252,261,269],"notation=\"LaTeX\">$8.7\\times":[186],"geo-mean":[189],"speedup":[190],"up":[192],"notation=\"LaTeX\">$10^{4}\\times":[198],"notation=\"LaTeX\">$10^{3}\\times":[206],"reduction":[210],"over":[211],"Intel":[212],"Xeon":[213],"5220":[214],"CPU":[215],"NVIDIA":[217],"V100":[218],"GPU,":[219],"respectively,":[220],"Compared":[221],"state-of-the-art":[223],"accelerators":[225],"SpAtten":[226],"notation=\"LaTeX\">$A^{3}$":[232],",":[234],"notation=\"LaTeX\">$1.7\\times":[242],"notation=\"LaTeX\">$1.25\\times":[250],"speedup,":[253],"notation=\"LaTeX\">$1.6\\times":[259],"notation=\"LaTeX\">$1.5\\times":[267],"higher":[270],"efficiency.":[272]},"counts_by_year":[{"year":2026,"cited_by_count":11},{"year":2025,"cited_by_count":27},{"year":2024,"cited_by_count":13},{"year":2023,"cited_by_count":7}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}