{"id":"https://openalex.org/W3199934250","doi":"https://doi.org/10.1145/3477002","title":"Algorithm-hardware Co-design of Attention Mechanism on FPGA Devices","display_name":"Algorithm-hardware Co-design of Attention Mechanism on FPGA Devices","publication_year":2021,"publication_date":"2021-09-17","ids":{"openalex":"https://openalex.org/W3199934250","doi":"https://doi.org/10.1145/3477002","mag":"3199934250"},"language":"en","primary_location":{"id":"doi:10.1145/3477002","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3477002","pdf_url":null,"source":{"id":"https://openalex.org/S136160450","display_name":"ACM Transactions on Embedded Computing Systems","issn_l":"1539-9087","issn":["1539-9087","1558-3465"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Embedded Computing 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/A5016612088","display_name":"Xinyi Zhang","orcid":"https://orcid.org/0000-0002-6577-7705"},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xinyi Zhang","raw_affiliation_strings":["University of Pittsburgh, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Pittsburgh, USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019020335","display_name":"Yawen Wu","orcid":"https://orcid.org/0000-0001-6840-267X"},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yawen Wu","raw_affiliation_strings":["University of Pittsburgh, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Pittsburgh, USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063866156","display_name":"Peipei Zhou","orcid":"https://orcid.org/0000-0002-0493-1844"},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Peipei Zhou","raw_affiliation_strings":["University of Pittsburgh, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Pittsburgh, USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087859795","display_name":"Xulong Tang","orcid":"https://orcid.org/0000-0002-3385-2053"},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xulong Tang","raw_affiliation_strings":["University of Pittsburgh, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Pittsburgh, USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5066534595","display_name":"Jingtong Hu","orcid":"https://orcid.org/0000-0003-4029-4034"},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jingtong Hu","raw_affiliation_strings":["University of Pittsburgh, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Pittsburgh, USA","institution_ids":["https://openalex.org/I170201317"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I170201317"],"apc_list":null,"apc_paid":null,"fwci":3.7687,"has_fulltext":false,"cited_by_count":51,"citation_normalized_percentile":{"value":0.94769961,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":97,"max":100},"biblio":{"volume":"20","issue":"5s","first_page":"1","last_page":"24"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998000264167786,"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":0.9998000264167786,"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/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.9955999851226807,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9937000274658203,"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/computer-science","display_name":"Computer science","score":0.8787665963172913},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.7358518838882446},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.5340413451194763},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.49178531765937805},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.49033141136169434},{"id":"https://openalex.org/keywords/rectangle","display_name":"Rectangle","score":0.4261625409126282},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.3795708417892456},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.3379838764667511},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.3343052268028259}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8787665963172913},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.7358518838882446},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.5340413451194763},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.49178531765937805},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.49033141136169434},{"id":"https://openalex.org/C2781302577","wikidata":"https://www.wikidata.org/wiki/Q209","display_name":"Rectangle","level":2,"score":0.4261625409126282},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3795708417892456},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.3379838764667511},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3343052268028259},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.0},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","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}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3477002","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3477002","pdf_url":null,"source":{"id":"https://openalex.org/S136160450","display_name":"ACM Transactions on Embedded Computing Systems","issn_l":"1539-9087","issn":["1539-9087","1558-3465"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Embedded Computing Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.8999999761581421}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":14,"referenced_works":["https://openalex.org/W2094756095","https://openalex.org/W2276892413","https://openalex.org/W2442974303","https://openalex.org/W2911884654","https://openalex.org/W2962953210","https://openalex.org/W2979569264","https://openalex.org/W2997828269","https://openalex.org/W3016142271","https://openalex.org/W3047848469","https://openalex.org/W3092581393","https://openalex.org/W3096609285","https://openalex.org/W3114397406","https://openalex.org/W3121413886","https://openalex.org/W3135841641"],"related_works":["https://openalex.org/W2242098268","https://openalex.org/W4244945894","https://openalex.org/W2760515006","https://openalex.org/W2802042052","https://openalex.org/W1987609185","https://openalex.org/W2003462717","https://openalex.org/W4224881918","https://openalex.org/W174828583","https://openalex.org/W2059098648","https://openalex.org/W4210772589"],"abstract_inverted_index":{"Multi-head":[0],"self-attention":[1],"(attention":[2],"mechanism)":[3],"has":[4],"been":[5],"employed":[6],"in":[7,25,91,117,133,193,226,326],"a":[8,36,67,73,231],"variety":[9],"of":[10,39,200,224,323],"fields":[11],"such":[12],"as":[13,149],"machine":[14],"translation,":[15],"language":[16],"modeling,":[17],"and":[18,28,41,88,136,152,172,191,209,234,258,279],"image":[19],"processing":[20,313],"due":[21],"to":[22,60,111,180,197,213,220],"its":[23],"superiority":[24],"feature":[26],"extraction":[27],"sequential":[29],"data":[30,169],"analysis.":[31],"This":[32],"is":[33,81,243,277,285],"benefited":[34],"from":[35],"large":[37,222],"number":[38,223],"parameters":[40,93],"sophisticated":[42],"model":[43,63,70,80,92,98,103,119],"architecture":[44,104],"behind":[45],"the":[46,62,85,102,118,127,168,205,221,227,252,259,264,280,310,314,319],"attention":[47,52,140,228,265,298,316],"mechanism.":[48],"To":[49],"efficiently":[50,109,144],"deploy":[51],"mechanism":[53,141,266,299,317],"on":[54,146,241,255,283],"resource-constrained":[55],"devices,":[56],"existing":[57],"works":[58],"propose":[59],"reduce":[61],"size":[64,99],"by":[65,302],"building":[66,230],"customized":[68,78],"smaller":[69,79,190],"or":[71],"compressing":[72],"big":[74],"standard":[75],"model.":[76],"A":[77,268],"usually":[82,121],"optimized":[83],"for":[84,162,237],"specific":[86],"task":[87],"needs":[89],"effort":[90],"exploration.":[94],"Model":[95],"compression":[96,253],"reduces":[97],"without":[100],"hurting":[101],"robustness,":[105],"which":[106,195],"can":[107,142,155,294,307],"be":[108,143,156],"applied":[110],"different":[112,192,206],"tasks.":[113],"The":[114,287,304],"compressed":[115,139,239],"weights":[116,187,225],"are":[120,175,188],"regularly":[122],"shaped":[123],"(e.g.":[124,131],"rectangle)":[125],"but":[126],"dimension":[128],"sizes":[129],"vary":[130],"differs":[132],"rectangle":[134],"height":[135],"width).":[137],"Such":[138],"deployed":[145],"CPU/GPU":[147],"platforms":[148],"their":[150],"memory":[151,274],"computing":[153,173,216,235,261],"resources":[154],"flexibly":[157],"assigned":[158],"with":[159,273,318],"demand.":[160],"However,":[161],"Field":[163],"Programmable":[164],"Gate":[165],"Arrays":[166],"(FPGAs),":[167],"buffer":[170,233,256],"allocation":[171,257],"kernel":[174,217,236,262],"fixed":[176],"at":[177],"run":[178],"time":[179],"achieve":[181],"maximum":[182],"energy":[183],"efficiency.":[184],"After":[185],"compression,":[186],"much":[189],"size,":[194],"leads":[196],"inefficient":[198,214],"utilization":[199],"FPGA":[201,215,242,284,311],"on-chip":[202],"buffer.":[203],"Moreover,":[204],"weight":[207,240],"heights":[208],"widths":[210],"may":[211],"lead":[212],"execution.":[218],"Due":[219],"mechanism,":[229],"unique":[232],"each":[238],"not":[244],"feasible.":[245],"In":[246],"this":[247],"work,":[248],"we":[249],"jointly":[250],"consider":[251],"impact":[254],"required":[260],"during":[263],"compressing.":[267],"novel":[269],"structural":[270],"pruning":[271],"method":[272],"footprint":[275],"awareness":[276],"proposed":[278],"associated":[281],"accelerator":[282,306],"designed.":[286],"experimental":[288],"results":[289],"show":[290],"that":[291],"our":[292],"work":[293],"compress":[295],"Transformer":[296],"(an":[297],"based":[300],"model)":[301],"95x.":[303],"developed":[305],"fully":[308],"utilize":[309],"resource,":[312],"sparse":[315],"run-time":[320],"throughput":[321],"performance":[322],"1.87":[324],"Tops":[325],"ZCU102":[327],"FPGA.":[328]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":9},{"year":2024,"cited_by_count":20},{"year":2023,"cited_by_count":12},{"year":2022,"cited_by_count":8}],"updated_date":"2026-05-21T06:26:12.895304","created_date":"2025-10-10T00:00:00"}