{"id":"https://openalex.org/W2793471971","doi":"https://doi.org/10.1145/3174243.3174999","title":"Exploration of Low Numeric Precision Deep Learning Inference Using Intel\u00ae FPGAs","display_name":"Exploration of Low Numeric Precision Deep Learning Inference Using Intel\u00ae FPGAs","publication_year":2018,"publication_date":"2018-02-15","ids":{"openalex":"https://openalex.org/W2793471971","doi":"https://doi.org/10.1145/3174243.3174999","mag":"2793471971"},"language":"en","primary_location":{"id":"doi:10.1145/3174243.3174999","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3174243.3174999","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2018 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/A5055662480","display_name":"Philip Colangelo","orcid":null},"institutions":[{"id":"https://openalex.org/I1343180700","display_name":"Intel (United States)","ror":"https://ror.org/01ek73717","country_code":"US","type":"company","lineage":["https://openalex.org/I1343180700"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Philip Colangelo","raw_affiliation_strings":["Intel, San Jose, CA, USA"],"affiliations":[{"raw_affiliation_string":"Intel, San Jose, CA, USA","institution_ids":["https://openalex.org/I1343180700"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022011722","display_name":"Nasibeh Nasiri","orcid":null},"institutions":[{"id":"https://openalex.org/I1343180700","display_name":"Intel (United States)","ror":"https://ror.org/01ek73717","country_code":"US","type":"company","lineage":["https://openalex.org/I1343180700"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Nasibeh Nasiri","raw_affiliation_strings":["Intel, San Jose, CA, USA"],"affiliations":[{"raw_affiliation_string":"Intel, San Jose, CA, USA","institution_ids":["https://openalex.org/I1343180700"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5084078152","display_name":"Eriko Nurvitadhi","orcid":"https://orcid.org/0000-0002-2347-9590"},"institutions":[{"id":"https://openalex.org/I1343180700","display_name":"Intel (United States)","ror":"https://ror.org/01ek73717","country_code":"US","type":"company","lineage":["https://openalex.org/I1343180700"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Eriko Nurvitadhi","raw_affiliation_strings":["Intel, Hillsboro, OR, USA"],"affiliations":[{"raw_affiliation_string":"Intel, Hillsboro, OR, USA","institution_ids":["https://openalex.org/I1343180700"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101912070","display_name":"Asit Mishra","orcid":"https://orcid.org/0000-0001-6489-6895"},"institutions":[{"id":"https://openalex.org/I1343180700","display_name":"Intel (United States)","ror":"https://ror.org/01ek73717","country_code":"US","type":"company","lineage":["https://openalex.org/I1343180700"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Asit Mishra","raw_affiliation_strings":["Intel, Hillsboro, OR, USA"],"affiliations":[{"raw_affiliation_string":"Intel, Hillsboro, OR, USA","institution_ids":["https://openalex.org/I1343180700"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062664944","display_name":"Martin Margala","orcid":"https://orcid.org/0000-0002-0034-0369"},"institutions":[{"id":"https://openalex.org/I133738476","display_name":"University of Massachusetts Lowell","ror":"https://ror.org/03hamhx47","country_code":"US","type":"education","lineage":["https://openalex.org/I133738476"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Martin Margala","raw_affiliation_strings":["University of Massachusetts Lowell, Lowell, MA, USA"],"affiliations":[{"raw_affiliation_string":"University of Massachusetts Lowell, Lowell, MA, USA","institution_ids":["https://openalex.org/I133738476"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5013923723","display_name":"Kevin Nealis","orcid":null},"institutions":[{"id":"https://openalex.org/I1343180700","display_name":"Intel (United States)","ror":"https://ror.org/01ek73717","country_code":"US","type":"company","lineage":["https://openalex.org/I1343180700"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kevin Nealis","raw_affiliation_strings":["Intel, San Jose, CA, USA"],"affiliations":[{"raw_affiliation_string":"Intel, San Jose, CA, USA","institution_ids":["https://openalex.org/I1343180700"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5055662480"],"corresponding_institution_ids":["https://openalex.org/I1343180700"],"apc_list":null,"apc_paid":null,"fwci":1.0446,"has_fulltext":false,"cited_by_count":12,"citation_normalized_percentile":{"value":0.81820506,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"294","last_page":"294"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9959999918937683,"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":0.9959999918937683,"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":0.9959999918937683,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9851999878883362,"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/stratix","display_name":"Stratix","score":0.9384294748306274},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.823430061340332},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.7580416202545166},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.6553532481193542},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.6002098321914673},{"id":"https://openalex.org/keywords/single-precision-floating-point-format","display_name":"Single-precision floating-point format","score":0.538602352142334},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.506890594959259},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.4670478403568268},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.45150458812713623},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.4361041486263275},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.41869133710861206},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.38957351446151733},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.3728013038635254},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3106161653995514}],"concepts":[{"id":"https://openalex.org/C2776277307","wikidata":"https://www.wikidata.org/wiki/Q22074755","display_name":"Stratix","level":3,"score":0.9384294748306274},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.823430061340332},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.7580416202545166},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.6553532481193542},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.6002098321914673},{"id":"https://openalex.org/C133095886","wikidata":"https://www.wikidata.org/wiki/Q1307173","display_name":"Single-precision floating-point format","level":3,"score":0.538602352142334},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.506890594959259},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.4670478403568268},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.45150458812713623},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.4361041486263275},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.41869133710861206},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.38957351446151733},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.3728013038635254},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3106161653995514},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.0},{"id":"https://openalex.org/C555944384","wikidata":"https://www.wikidata.org/wiki/Q249","display_name":"Wireless","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3174243.3174999","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3174243.3174999","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2018 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9","score":0.44999998807907104}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W1509155667","https://openalex.org/W2518118925","https://openalex.org/W3208151864","https://openalex.org/W1564576805","https://openalex.org/W4254372399","https://openalex.org/W3129734321","https://openalex.org/W2793471971","https://openalex.org/W4220984504","https://openalex.org/W2131812496","https://openalex.org/W2889596074"],"abstract_inverted_index":{"Convolutional":[0],"neural":[1],"networks":[2,42,63,104,154],"have":[3],"been":[4],"shown":[5],"to":[6,14,18,88,159],"maintain":[7],"reasonable":[8],"classification":[9,27],"accuracy":[10,28,38,56,151,211,244],"when":[11],"quantized":[12],"down":[13],"8-bits,":[15],"however,":[16],"quantizing":[17],"sub":[19,40],"8-bit":[20,41],"activations":[21],"and":[22,55,57,68,101,113,134,150,155,179,191],"weights":[23],"can":[24,58,170],"result":[25],"in":[26,50,196],"falling":[29],"below":[30],"an":[31,106],"acceptable":[32],"threshold.":[33],"Techniques":[34],"exist":[35],"for":[36,61,78,111,125,152,176,219,230],"increasing":[37,47],"of":[39,46,66,109,130,137,212],"typically":[43],"by":[44,95],"means":[45],"computation":[48,82,135,169],"resulting":[49],"a":[51,122,188,209,215,231,238],"trade-off":[52],"between":[53,148],"throughput":[54,149],"be":[59,171],"tailored":[60],"different":[62],"through":[64,83],"combinations":[65],"activation":[67,190],"weight":[69],"precisions.":[70],"Customizable":[71],"hardware":[72,123,197,216],"architectures":[73],"like":[74],"FPGAs":[75,126,175],"provide":[76,143],"opportunity":[77],"data":[79,115],"width":[80],"specific":[81],"unique":[84],"logic":[85],"configurations":[86],"leading":[87],"highly":[89],"optimized":[90],"processing":[91],"that":[92,127,198],"is":[93],"unattainable":[94],"full":[96],"precision":[97,140,168,223],"networks.":[98],"Specifically,":[99],"ternary":[100,178,192],"binary":[102,180],"weighted":[103,193],"offer":[105],"efficient":[107],"method":[108],"inference":[110],"2-bit":[112,189],"1-bit":[114],"respectively.":[116],"In":[117],"this":[118],"paper,":[119],"we":[120,163,186,225],"present":[121],"design":[124],"takes":[128],"advantage":[129],"the":[131,146,205],"bandwidth,":[132],"memory,":[133],"savings":[136],"limited":[138,166],"numerical":[139],"data.":[141],"We":[142],"insights":[144],"into":[145],"trade-offs":[147],"various":[153],"how":[156,165],"they":[157],"map":[158],"our":[160,220],"framework.":[161],"Further,":[162],"show":[164,187],"numeric":[167,222],"efficiently":[172],"mapped":[173],"onto":[174],"both":[177],"cases.":[181],"Starting":[182],"with":[183,208,241,247],"Arria":[184],"10,":[185],"AlexNet":[194],"running":[195,237],"achieves":[199],"3,700":[200],"images":[201],"per":[202],"second":[203],"on":[204],"ImageNet":[206],"dataset":[207],"top-1":[210],"0.49.":[213],"Using":[214],"modeler":[217],"designed":[218],"low":[221],"framework":[224],"project":[226],"performance":[227],"most":[228],"notably":[229],"55.5":[232],"TOPS":[233],"Stratix":[234],"10":[235],"device":[236],"modified":[239],"ResNet-34":[240],"only":[242],"3.7%":[243],"degradation":[245],"compared":[246],"single":[248],"precision.":[249]},"counts_by_year":[{"year":2024,"cited_by_count":1},{"year":2022,"cited_by_count":1},{"year":2021,"cited_by_count":5},{"year":2020,"cited_by_count":1},{"year":2019,"cited_by_count":4}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}