{"id":"https://openalex.org/W3137649401","doi":"https://doi.org/10.1109/dac18074.2021.9586134","title":"Softermax: Hardware/Software Co-Design of an Efficient Softmax for Transformers","display_name":"Softermax: Hardware/Software Co-Design of an Efficient Softmax for Transformers","publication_year":2021,"publication_date":"2021-11-08","ids":{"openalex":"https://openalex.org/W3137649401","doi":"https://doi.org/10.1109/dac18074.2021.9586134","mag":"3137649401"},"language":"en","primary_location":{"id":"doi:10.1109/dac18074.2021.9586134","is_oa":false,"landing_page_url":"https://doi.org/10.1109/dac18074.2021.9586134","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 58th ACM/IEEE Design Automation Conference (DAC)","raw_type":"proceedings-article"},"type":"preprint","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2103.09301","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5044634700","display_name":"Jacob R. Stevens","orcid":"https://orcid.org/0000-0002-7626-2846"},"institutions":[{"id":"https://openalex.org/I219193219","display_name":"Purdue University West Lafayette","ror":"https://ror.org/02dqehb95","country_code":"US","type":"education","lineage":["https://openalex.org/I219193219"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Jacob R. Stevens","raw_affiliation_strings":["Purdue University, West Lafayette","Purdue Univ., West Lafayette#TAB#"],"affiliations":[{"raw_affiliation_string":"Purdue University, West Lafayette","institution_ids":["https://openalex.org/I219193219"]},{"raw_affiliation_string":"Purdue Univ., West Lafayette#TAB#","institution_ids":["https://openalex.org/I219193219"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5045219356","display_name":"Rangharajan Venkatesan","orcid":null},"institutions":[{"id":"https://openalex.org/I1304085615","display_name":"Nvidia (United Kingdom)","ror":"https://ror.org/02kr42612","country_code":"GB","type":"company","lineage":["https://openalex.org/I1304085615","https://openalex.org/I4210127875"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Rangharajan Venkatesan","raw_affiliation_strings":["NVIDIA","nVidia"],"affiliations":[{"raw_affiliation_string":"NVIDIA","institution_ids":[]},{"raw_affiliation_string":"nVidia","institution_ids":["https://openalex.org/I1304085615"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030894829","display_name":"Steve Dai","orcid":"https://orcid.org/0000-0002-5045-1964"},"institutions":[{"id":"https://openalex.org/I1304085615","display_name":"Nvidia (United Kingdom)","ror":"https://ror.org/02kr42612","country_code":"GB","type":"company","lineage":["https://openalex.org/I1304085615","https://openalex.org/I4210127875"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Steve Dai","raw_affiliation_strings":["NVIDIA","nVidia"],"affiliations":[{"raw_affiliation_string":"NVIDIA","institution_ids":[]},{"raw_affiliation_string":"nVidia","institution_ids":["https://openalex.org/I1304085615"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010156116","display_name":"Brucek Khailany","orcid":"https://orcid.org/0000-0002-7584-3489"},"institutions":[{"id":"https://openalex.org/I1304085615","display_name":"Nvidia (United Kingdom)","ror":"https://ror.org/02kr42612","country_code":"GB","type":"company","lineage":["https://openalex.org/I1304085615","https://openalex.org/I4210127875"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Brucek Khailany","raw_affiliation_strings":["NVIDIA","nVidia"],"affiliations":[{"raw_affiliation_string":"NVIDIA","institution_ids":[]},{"raw_affiliation_string":"nVidia","institution_ids":["https://openalex.org/I1304085615"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5065766721","display_name":"Anand Raghunathan","orcid":"https://orcid.org/0000-0002-4624-564X"},"institutions":[{"id":"https://openalex.org/I219193219","display_name":"Purdue University West Lafayette","ror":"https://ror.org/02dqehb95","country_code":"US","type":"education","lineage":["https://openalex.org/I219193219"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Anand Raghunathan","raw_affiliation_strings":["Purdue University, West Lafayette","Purdue Univ., West Lafayette#TAB#"],"affiliations":[{"raw_affiliation_string":"Purdue University, West Lafayette","institution_ids":["https://openalex.org/I219193219"]},{"raw_affiliation_string":"Purdue Univ., West Lafayette#TAB#","institution_ids":["https://openalex.org/I219193219"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5044634700"],"corresponding_institution_ids":["https://openalex.org/I219193219"],"apc_list":null,"apc_paid":null,"fwci":0.2024,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.48865346,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":93,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"469","last_page":"474"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9987000226974487,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9987000226974487,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9983999729156494,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9977999925613403,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/softmax-function","display_name":"Softmax function","score":0.973034143447876},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.7431572675704956},{"id":"https://openalex.org/keywords/normalization","display_name":"Normalization (sociology)","score":0.6778908371925354},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.6526008248329163},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5556652545928955},{"id":"https://openalex.org/keywords/software","display_name":"Software","score":0.5325288772583008},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.4131249785423279},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.29910922050476074},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.2817903161048889},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.2805393934249878},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.23410865664482117},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.1920088231563568}],"concepts":[{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.973034143447876},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.7431572675704956},{"id":"https://openalex.org/C136886441","wikidata":"https://www.wikidata.org/wiki/Q926129","display_name":"Normalization (sociology)","level":2,"score":0.6778908371925354},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.6526008248329163},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5556652545928955},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.5325288772583008},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.4131249785423279},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.29910922050476074},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2817903161048889},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2805393934249878},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.23410865664482117},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.1920088231563568},{"id":"https://openalex.org/C19165224","wikidata":"https://www.wikidata.org/wiki/Q23404","display_name":"Anthropology","level":1,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0},{"id":"https://openalex.org/C144024400","wikidata":"https://www.wikidata.org/wiki/Q21201","display_name":"Sociology","level":0,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.1109/dac18074.2021.9586134","is_oa":false,"landing_page_url":"https://doi.org/10.1109/dac18074.2021.9586134","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 58th ACM/IEEE Design Automation Conference (DAC)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2103.09301","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2103.09301","pdf_url":"https://arxiv.org/pdf/2103.09301","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"mag:3137649401","is_oa":true,"landing_page_url":"http://export.arxiv.org/pdf/2103.09301","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"arXiv (Cornell University)","raw_type":null},{"id":"doi:10.48550/arxiv.2103.09301","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2103.09301","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2103.09301","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2103.09301","pdf_url":"https://arxiv.org/pdf/2103.09301","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[{"score":0.8999999761581421,"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":false},"content_urls":{"pdf":"https://content.openalex.org/works/W3137649401.pdf"},"referenced_works_count":32,"referenced_works":["https://openalex.org/W2606722458","https://openalex.org/W2801655600","https://openalex.org/W2915716523","https://openalex.org/W2946480111","https://openalex.org/W2953212265","https://openalex.org/W2963341956","https://openalex.org/W2963403868","https://openalex.org/W2973727699","https://openalex.org/W2978017171","https://openalex.org/W2979314664","https://openalex.org/W2980282514","https://openalex.org/W2997929983","https://openalex.org/W3017024317","https://openalex.org/W3017746288","https://openalex.org/W3030163527","https://openalex.org/W3035083896","https://openalex.org/W3036061420","https://openalex.org/W3090555870","https://openalex.org/W3091783050","https://openalex.org/W3094502228","https://openalex.org/W3098873988","https://openalex.org/W6739901393","https://openalex.org/W6751068846","https://openalex.org/W6755207826","https://openalex.org/W6767997687","https://openalex.org/W6768807518","https://openalex.org/W6768851824","https://openalex.org/W6775160533","https://openalex.org/W6776320331","https://openalex.org/W6778883912","https://openalex.org/W6783744057","https://openalex.org/W6784333009"],"related_works":["https://openalex.org/W3211525823","https://openalex.org/W2959286300","https://openalex.org/W3203065286","https://openalex.org/W3000320208","https://openalex.org/W2909711603","https://openalex.org/W2927803337","https://openalex.org/W3116241656","https://openalex.org/W2996690709","https://openalex.org/W3207577041","https://openalex.org/W3113114492","https://openalex.org/W2908696560","https://openalex.org/W3046842446","https://openalex.org/W2910518169","https://openalex.org/W2625483679","https://openalex.org/W2402098947","https://openalex.org/W2891230170","https://openalex.org/W3177366646","https://openalex.org/W2952833359","https://openalex.org/W2973313988","https://openalex.org/W2973506077"],"abstract_inverted_index":{"Transformers":[0],"have":[1],"transformed":[2],"the":[3,16,41,50,84,89],"field":[4],"of":[5,18,23,26,49,53,67,91],"natural":[6],"language":[7],"processing.":[8],"Their":[9],"superior":[10],"performance":[11],"is":[12],"largely":[13],"attributed":[14],"to":[15],"use":[17],"stacked":[19],"\u201cself-attention\u201d":[20],"layers,":[21],"each":[22],"which":[24],"consists":[25,66],"matrix":[27],"multiplies":[28],"as":[29,31],"well":[30],"softmax":[32,42,63,71],"operations.":[33],"As":[34],"a":[35,46,61,92],"result,":[36],"unlike":[37],"other":[38],"neural":[39],"networks,":[40],"operation":[43],"accounts":[44],"for":[45],"significant":[47],"fraction":[48],"total":[51],"run-time":[52],"Transformers.":[54],"To":[55],"address":[56],"this,":[57],"we":[58],"propose":[59],"Softermax,":[60],"hardware-friendly":[62],"design.":[64],"Softermax":[65,80],"base":[68],"replacement,":[69],"low-precision":[70],"computations,":[72],"and":[73],"an":[74],"online":[75],"normalization":[76],"calculation.":[77],"We":[78],"show":[79],"results":[81],"in":[82],"2.35x":[83],"energy":[85],"efficiency":[86],"at":[87],"0.90x":[88],"size":[90],"comparable":[93],"baseline,":[94],"with":[95],"negligible":[96],"impact":[97],"on":[98],"network":[99],"accuracy.":[100]},"counts_by_year":[{"year":2021,"cited_by_count":2}],"updated_date":"2026-03-20T23:20:44.827607","created_date":"2025-10-10T00:00:00"}