{"id":"https://openalex.org/W4389166736","doi":"https://doi.org/10.1109/iccad57390.2023.10323725","title":"SOLE: Hardware-Software Co-design of Softmax and LayerNorm for Efficient Transformer Inference","display_name":"SOLE: Hardware-Software Co-design of Softmax and LayerNorm for Efficient Transformer Inference","publication_year":2023,"publication_date":"2023-10-28","ids":{"openalex":"https://openalex.org/W4389166736","doi":"https://doi.org/10.1109/iccad57390.2023.10323725"},"language":"en","primary_location":{"id":"doi:10.1109/iccad57390.2023.10323725","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccad57390.2023.10323725","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/ACM International Conference on Computer Aided Design (ICCAD)","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/A5023029393","display_name":"Wenxun Wang","orcid":"https://orcid.org/0009-0007-1999-6441"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenxun Wang","raw_affiliation_strings":["Tsinghua University,Department of Electronic Engineering,Beijing,China","Department of Electronic Engineering, Tsinghua University, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University,Department of Electronic Engineering,Beijing,China","institution_ids":["https://openalex.org/I99065089"]},{"raw_affiliation_string":"Department of Electronic Engineering, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008993224","display_name":"Shuchang Zhou","orcid":"https://orcid.org/0000-0002-7294-5869"},"institutions":[{"id":"https://openalex.org/I4401726805","display_name":"Megvii (China)","ror":"https://ror.org/040b32p69","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726805"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shuchang Zhou","raw_affiliation_strings":["MEGVII Technology,Beijing,China","MEGVII Technology, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MEGVII Technology,Beijing,China","institution_ids":["https://openalex.org/I4401726805"]},{"raw_affiliation_string":"MEGVII Technology, Beijing, China","institution_ids":["https://openalex.org/I4401726805"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085072777","display_name":"Wenyu Sun","orcid":"https://orcid.org/0000-0002-4793-0972"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenyu Sun","raw_affiliation_strings":["Tsinghua University,Department of Electronic Engineering,Beijing,China","Department of Electronic Engineering, Tsinghua University, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University,Department of Electronic Engineering,Beijing,China","institution_ids":["https://openalex.org/I99065089"]},{"raw_affiliation_string":"Department of Electronic Engineering, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5025439316","display_name":"Peiqin Sun","orcid":"https://orcid.org/0000-0003-3644-2372"},"institutions":[{"id":"https://openalex.org/I4401726805","display_name":"Megvii (China)","ror":"https://ror.org/040b32p69","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726805"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Peiqin Sun","raw_affiliation_strings":["MEGVII Technology,Beijing,China","MEGVII Technology, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MEGVII Technology,Beijing,China","institution_ids":["https://openalex.org/I4401726805"]},{"raw_affiliation_string":"MEGVII Technology, Beijing, China","institution_ids":["https://openalex.org/I4401726805"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5045721867","display_name":"Yongpan Liu","orcid":"https://orcid.org/0000-0002-4892-2309"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yongpan Liu","raw_affiliation_strings":["Tsinghua University,Department of Electronic Engineering,Beijing,China","Department of Electronic Engineering, Tsinghua University, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University,Department of Electronic Engineering,Beijing,China","institution_ids":["https://openalex.org/I99065089"]},{"raw_affiliation_string":"Department of Electronic Engineering, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":46,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"9"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9984999895095825,"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":0.9984999895095825,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9879999756813049,"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/T10320","display_name":"Neural Networks and Applications","score":0.9824000000953674,"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/softmax-function","display_name":"Softmax function","score":0.9728029370307922},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6898128986358643},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6171895861625671},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.6130474209785461},{"id":"https://openalex.org/keywords/normalization","display_name":"Normalization (sociology)","score":0.5740554928779602},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5448246002197266},{"id":"https://openalex.org/keywords/efficient-energy-use","display_name":"Efficient energy use","score":0.49860429763793945},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4863227307796478},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.46817129850387573},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.4616246819496155},{"id":"https://openalex.org/keywords/software","display_name":"Software","score":0.42799103260040283},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.4054439067840576},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3402828574180603},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3284570574760437},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.11965984106063843},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.09633085131645203},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.08731335401535034}],"concepts":[{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.9728029370307922},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6898128986358643},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6171895861625671},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.6130474209785461},{"id":"https://openalex.org/C136886441","wikidata":"https://www.wikidata.org/wiki/Q926129","display_name":"Normalization (sociology)","level":2,"score":0.5740554928779602},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5448246002197266},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.49860429763793945},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4863227307796478},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.46817129850387573},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.4616246819496155},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.42799103260040283},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4054439067840576},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3402828574180603},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3284570574760437},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.11965984106063843},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.09633085131645203},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.08731335401535034},{"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/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.0},{"id":"https://openalex.org/C19165224","wikidata":"https://www.wikidata.org/wiki/Q23404","display_name":"Anthropology","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":1,"locations":[{"id":"doi:10.1109/iccad57390.2023.10323725","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccad57390.2023.10323725","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/ACM International Conference on Computer Aided Design (ICCAD)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.7300000190734863,"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":47,"referenced_works":["https://openalex.org/W1498436455","https://openalex.org/W1686810756","https://openalex.org/W2108598243","https://openalex.org/W2111013824","https://openalex.org/W2469490737","https://openalex.org/W2511929605","https://openalex.org/W2801655600","https://openalex.org/W2896457183","https://openalex.org/W2903972532","https://openalex.org/W2910396952","https://openalex.org/W2923014074","https://openalex.org/W2951008357","https://openalex.org/W2963748441","https://openalex.org/W2998183051","https://openalex.org/W3017024317","https://openalex.org/W3017746288","https://openalex.org/W3122239467","https://openalex.org/W3138516171","https://openalex.org/W3159727696","https://openalex.org/W3170841864","https://openalex.org/W3170874841","https://openalex.org/W3189138869","https://openalex.org/W3196923642","https://openalex.org/W3206453033","https://openalex.org/W3211525823","https://openalex.org/W3211787299","https://openalex.org/W4214686755","https://openalex.org/W4285601701","https://openalex.org/W4292779060","https://openalex.org/W4293025109","https://openalex.org/W4320722432","https://openalex.org/W4322718253","https://openalex.org/W4323572061","https://openalex.org/W4385245566","https://openalex.org/W6638318767","https://openalex.org/W6720242923","https://openalex.org/W6751068846","https://openalex.org/W6776320331","https://openalex.org/W6778883912","https://openalex.org/W6780226713","https://openalex.org/W6784094891","https://openalex.org/W6788001715","https://openalex.org/W6790978476","https://openalex.org/W6797854001","https://openalex.org/W6849928207","https://openalex.org/W6850162387","https://openalex.org/W6850503672"],"related_works":["https://openalex.org/W3107204728","https://openalex.org/W4287591324","https://openalex.org/W3108503355","https://openalex.org/W3090555870","https://openalex.org/W4226420367","https://openalex.org/W3204400881","https://openalex.org/W3214410901","https://openalex.org/W3204296682","https://openalex.org/W3183118997","https://openalex.org/W2917767146"],"abstract_inverted_index":{"Transformers":[0],"have":[1],"shown":[2],"remarkable":[3],"performance":[4],"in":[5,29],"both":[6,118],"natural":[7],"language":[8],"processing":[9],"(NLP)":[10],"and":[11,21,31,71,84,91,100,121,127,144,154,166],"computer":[12],"vision":[13],"(CV)":[14],"tasks.":[15],"However,":[16],"their":[17],"real-time":[18],"inference":[19,134],"speed":[20],"efficiency":[22],"are":[23],"limited":[24],"due":[25],"to":[26,62,103],"the":[27],"inefficiency":[28],"Softmax":[30,83,105,126,165],"Layer":[32],"Normalization":[33],"(LayerNorm).":[34],"Previous":[35],"works":[36],"based":[37],"on":[38,49,60,125],"function":[39,99],"approximation":[40,65],"suffer":[41],"from":[42],"inefficient":[43],"implementation":[44],"as":[45],"they":[46],"place":[47],"emphasis":[48],"computation":[50],"while":[51,106,138],"disregarding":[52],"memory":[53],"overhead":[54],"concerns.":[55],"Moreover,":[56],"such":[57],"methods":[58],"rely":[59],"retraining":[61,137],"compensate":[63],"for":[64,82,164],"error":[66],"which":[67,86],"can":[68],"be":[69],"costly":[70],"inconvenient.":[72],"In":[73],"this":[74],"paper,":[75],"we":[76,116],"present":[77],"SOLE,":[78],"a":[79],"hardware-software":[80],"co-design":[81],"LayerNorm":[85],"is":[87],"composed":[88],"of":[89,97,141],"E2Softmax":[90,93],"AILayerNorm.":[92],"utilizes":[94],"log2":[95],"quantization":[96],"exponent":[98],"log-based":[101],"division":[102],"approximate":[104],"AILayerNorm":[107],"adopts":[108],"low-precision":[109,119],"statistic":[110],"calculation.":[111],"Compared":[112],"with":[113],"state-of-the-art":[114,161],"designs,":[115],"achieve":[117],"calculation":[120],"low":[122],"bit-width":[123],"storage":[124],"LayerNorm.":[128],"Experiments":[129],"show":[130],"that":[131],"SOLE":[132],"maintains":[133],"accuracy":[135],"without":[136],"offering":[139],"orders":[140],"magnitude":[142],"speedup":[143],"energy":[145],"savings":[146],"over":[147,159],"GPU,":[148],"achieving":[149],"3.04\u00d7,":[150],"3.86\u00d7":[151],"energy-efficiency":[152],"improvements":[153,158],"2.82\u00d7,":[155],"3.32\u00d7":[156],"area-efficiency":[157],"prior":[160],"custom":[162],"hardware":[163],"LayerNorm,":[167],"respectively.":[168]},"counts_by_year":[{"year":2026,"cited_by_count":6},{"year":2025,"cited_by_count":36},{"year":2024,"cited_by_count":4}],"updated_date":"2026-07-22T07:51:19.307946","created_date":"2025-10-10T00:00:00"}