{"id":"https://openalex.org/W3212505503","doi":"https://doi.org/10.1145/3474597","title":"Low-precision Floating-point Arithmetic for High-performance FPGA-based CNN Acceleration","display_name":"Low-precision Floating-point Arithmetic for High-performance FPGA-based CNN Acceleration","publication_year":2021,"publication_date":"2021-11-09","ids":{"openalex":"https://openalex.org/W3212505503","doi":"https://doi.org/10.1145/3474597","mag":"3212505503"},"language":"en","primary_location":{"id":"doi:10.1145/3474597","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3474597","pdf_url":null,"source":{"id":"https://openalex.org/S112809824","display_name":"ACM Transactions on Reconfigurable Technology and Systems","issn_l":"1936-7406","issn":["1936-7406","1936-7414"],"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 Reconfigurable Technology 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/A5100707380","display_name":"Chen Wu","orcid":"https://orcid.org/0000-0002-2492-0640"},"institutions":[{"id":"https://openalex.org/I2802886717","display_name":"Westwood College","ror":"https://ror.org/05kjfpp30","country_code":"US","type":"education","lineage":["https://openalex.org/I2802886717"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Chen Wu","raw_affiliation_strings":["Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA"],"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA","institution_ids":["https://openalex.org/I2802886717"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100350102","display_name":"Mingyu Wang","orcid":"https://orcid.org/0000-0001-5722-9752"},"institutions":[{"id":"https://openalex.org/I2802886717","display_name":"Westwood College","ror":"https://ror.org/05kjfpp30","country_code":"US","type":"education","lineage":["https://openalex.org/I2802886717"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Mingyu Wang","raw_affiliation_strings":["Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA"],"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA","institution_ids":["https://openalex.org/I2802886717"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066732321","display_name":"Xinyuan Chu","orcid":"https://orcid.org/0009-0002-0773-7606"},"institutions":[{"id":"https://openalex.org/I2802886717","display_name":"Westwood College","ror":"https://ror.org/05kjfpp30","country_code":"US","type":"education","lineage":["https://openalex.org/I2802886717"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xinyuan Chu","raw_affiliation_strings":["Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA"],"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA","institution_ids":["https://openalex.org/I2802886717"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100366692","display_name":"Kun Wang","orcid":"https://orcid.org/0000-0002-9099-2781"},"institutions":[{"id":"https://openalex.org/I2802886717","display_name":"Westwood College","ror":"https://ror.org/05kjfpp30","country_code":"US","type":"education","lineage":["https://openalex.org/I2802886717"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kun Wang","raw_affiliation_strings":["Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA"],"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA","institution_ids":["https://openalex.org/I2802886717"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5008695429","display_name":"Lei He","orcid":"https://orcid.org/0000-0002-5266-3805"},"institutions":[{"id":"https://openalex.org/I2802886717","display_name":"Westwood College","ror":"https://ror.org/05kjfpp30","country_code":"US","type":"education","lineage":["https://openalex.org/I2802886717"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Lei He","raw_affiliation_strings":["Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA"],"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering, University of California, Westwood, Los Angeles, CA","institution_ids":["https://openalex.org/I2802886717"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5100707380"],"corresponding_institution_ids":["https://openalex.org/I2802886717"],"apc_list":null,"apc_paid":null,"fwci":2.7129,"has_fulltext":false,"cited_by_count":45,"citation_normalized_percentile":{"value":0.92003501,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":100},"biblio":{"volume":"15","issue":"1","first_page":"1","last_page":"21"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9991999864578247,"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":0.9991999864578247,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9965999722480774,"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/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.9961000084877014,"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":0.8337618112564087},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.7092475295066833},{"id":"https://openalex.org/keywords/adder","display_name":"Adder","score":0.7041066884994507},{"id":"https://openalex.org/keywords/floating-point","display_name":"Floating point","score":0.6948333382606506},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.6456485390663147},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.6155716180801392},{"id":"https://openalex.org/keywords/single-precision-floating-point-format","display_name":"Single-precision floating-point format","score":0.5989970564842224},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.5839688181877136},{"id":"https://openalex.org/keywords/digital-signal-processing","display_name":"Digital signal processing","score":0.576267421245575},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.45320701599121094},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.39215362071990967},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3479098081588745},{"id":"https://openalex.org/keywords/arithmetic","display_name":"Arithmetic","score":0.3333718180656433},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.24092403054237366},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.10085973143577576}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8337618112564087},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.7092475295066833},{"id":"https://openalex.org/C164620267","wikidata":"https://www.wikidata.org/wiki/Q376953","display_name":"Adder","level":3,"score":0.7041066884994507},{"id":"https://openalex.org/C84211073","wikidata":"https://www.wikidata.org/wiki/Q117879","display_name":"Floating point","level":2,"score":0.6948333382606506},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.6456485390663147},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.6155716180801392},{"id":"https://openalex.org/C133095886","wikidata":"https://www.wikidata.org/wiki/Q1307173","display_name":"Single-precision floating-point format","level":3,"score":0.5989970564842224},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.5839688181877136},{"id":"https://openalex.org/C84462506","wikidata":"https://www.wikidata.org/wiki/Q173142","display_name":"Digital signal processing","level":2,"score":0.576267421245575},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.45320701599121094},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.39215362071990967},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3479098081588745},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.3333718180656433},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.24092403054237366},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.10085973143577576},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.0},{"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:10.1145/3474597","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3474597","pdf_url":null,"source":{"id":"https://openalex.org/S112809824","display_name":"ACM Transactions on Reconfigurable Technology and Systems","issn_l":"1936-7406","issn":["1936-7406","1936-7414"],"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 Reconfigurable Technology and Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","score":0.6800000071525574,"display_name":"Affordable and clean energy"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":43,"referenced_works":["https://openalex.org/W1686810756","https://openalex.org/W2117539524","https://openalex.org/W2119144962","https://openalex.org/W2186222003","https://openalex.org/W2198190323","https://openalex.org/W2207050309","https://openalex.org/W2272300165","https://openalex.org/W2285660444","https://openalex.org/W2520083297","https://openalex.org/W2565851976","https://openalex.org/W2613574453","https://openalex.org/W2616014673","https://openalex.org/W2625954420","https://openalex.org/W2627042741","https://openalex.org/W2790167166","https://openalex.org/W2793471971","https://openalex.org/W2795915628","https://openalex.org/W2803935332","https://openalex.org/W2883542588","https://openalex.org/W2883920103","https://openalex.org/W2884071170","https://openalex.org/W2899915146","https://openalex.org/W2900316824","https://openalex.org/W2904331435","https://openalex.org/W2904902077","https://openalex.org/W2911708692","https://openalex.org/W2917087921","https://openalex.org/W2946355854","https://openalex.org/W2962820060","https://openalex.org/W2972918064","https://openalex.org/W2979455536","https://openalex.org/W2982479999","https://openalex.org/W3008072732","https://openalex.org/W3008408165","https://openalex.org/W3034100311","https://openalex.org/W3123290820","https://openalex.org/W3130240120","https://openalex.org/W3155487259","https://openalex.org/W3166056295","https://openalex.org/W4234552385","https://openalex.org/W4253012315","https://openalex.org/W4288083474","https://openalex.org/W4288413318"],"related_works":["https://openalex.org/W1564887326","https://openalex.org/W3037505396","https://openalex.org/W2116803521","https://openalex.org/W3150370983","https://openalex.org/W3215589575","https://openalex.org/W2239119680","https://openalex.org/W2773283032","https://openalex.org/W3150959508","https://openalex.org/W3009327594","https://openalex.org/W2797902698"],"abstract_inverted_index":{"Low-precision":[0],"data":[1,73],"representation":[2,74],"is":[3],"important":[4],"to":[5,27,60,165],"reduce":[6],"storage":[7],"size":[8],"and":[9,33,86,106,110,162,176,178,185],"memory":[10],"access":[11],"for":[12,30,41,57,92,145,160],"convolutional":[13],"neural":[14],"networks":[15],"(CNNs).":[16],"Yet,":[17],"existing":[18,90,156],"methods":[19,91],"have":[20],"two":[21,136],"major":[22],"limitations:":[23],"(1)":[24],"requiring":[25],"re-training":[26],"maintain":[28],"accuracy":[29,78],"deep":[31,93],"CNNs":[32,144],"(2)":[34],"needing":[35],"16-bit":[36],"floating-point":[37,53],"or":[38,124],"8-bit":[39,72,99,114,137],"fixed-point":[40,138],"a":[42,50],"good":[43],"accuracy.":[44],"In":[45],"this":[46],"article,":[47],"we":[48,96,151,170],"propose":[49],"low-precision":[51],"(8-bit)":[52],"(LPFP)":[54],"quantization":[55],"method":[56],"FPGA-based":[58],"acceleration":[59],"overcome":[61],"the":[62],"above":[63],"limitations.":[64],"Without":[65],"any":[66],"re-training,":[67],"LPFP":[68,100,115],"finds":[69],"an":[70],"optimal":[71],"with":[75],"negligible":[76],"top-1/top-5":[77],"loss":[79],"(within":[80],"0.5%/0.3%":[81],"in":[82],"our":[83],"experiments,":[84],"respectively,":[85],"significantly":[87],"better":[88],"than":[89],"CNNs).":[94],"Furthermore,":[95],"implement":[97,112,134],"one":[98,103,107,118,131],"multiplication":[101],"by":[102,154,174,183],"4-bit":[104],"multiply-adder":[105],"3-bit":[108],"adder,":[109],"therefore":[111],"four":[113],"multiplications":[116],"using":[117],"DSP48E1":[119],"of":[120,126],"Xilinx":[121,127],"Kintex-7":[122],"family":[123],"DSP48E2":[125],"Ultrascale/Ultrascale+":[128],"family,":[129],"whereas":[130],"DSP":[132,182],"can":[133],"only":[135],"multiplications.":[139],"Experiments":[140],"on":[141,149],"six":[142,166],"typical":[143],"inference":[146],"show":[147],"that":[148],"average,":[150],"improve":[152,171],"throughput":[153,173,180],"over":[155],"FPGA":[157,168],"accelerators.":[158],"Particularly":[159],"VGG16":[161],"YOLO,":[163],"compared":[164],"recent":[167],"accelerators,":[169],"average":[172,179],"3.5":[175],"27.5":[177],"per":[181],"4.1":[184],"5":[186],",":[187],"respectively.":[188]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":15},{"year":2024,"cited_by_count":6},{"year":2023,"cited_by_count":14},{"year":2022,"cited_by_count":8}],"updated_date":"2026-03-27T14:29:43.386196","created_date":"2025-10-10T00:00:00"}