{"id":"https://openalex.org/W4285264221","doi":"https://doi.org/10.1109/tcsi.2022.3184115","title":"PL-NPU: An Energy-Efficient Edge-Device DNN Training Processor With Posit-Based Logarithm-Domain Computing","display_name":"PL-NPU: An Energy-Efficient Edge-Device DNN Training Processor With Posit-Based Logarithm-Domain Computing","publication_year":2022,"publication_date":"2022-06-22","ids":{"openalex":"https://openalex.org/W4285264221","doi":"https://doi.org/10.1109/tcsi.2022.3184115"},"language":"en","primary_location":{"id":"doi:10.1109/tcsi.2022.3184115","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsi.2022.3184115","pdf_url":null,"source":{"id":"https://openalex.org/S116977442","display_name":"IEEE Transactions on Circuits and Systems I Regular Papers","issn_l":"1549-8328","issn":["1549-8328","1558-0806"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Circuits and Systems I: Regular Papers","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/A5021484887","display_name":"Yang Wang","orcid":"https://orcid.org/0000-0002-8293-8881"},"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":true,"raw_author_name":"Yang Wang","raw_affiliation_strings":["Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5037648551","display_name":"Dazheng Deng","orcid":"https://orcid.org/0009-0006-4229-4985"},"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":"Dazheng Deng","raw_affiliation_strings":["Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100358856","display_name":"Leibo Liu","orcid":"https://orcid.org/0000-0001-7548-4116"},"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":"Leibo Liu","raw_affiliation_strings":["Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036023084","display_name":"Shaojun Wei","orcid":"https://orcid.org/0000-0001-5117-7920"},"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":"Shaojun Wei","raw_affiliation_strings":["Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5054524841","display_name":"Shouyi Yin","orcid":"https://orcid.org/0000-0003-2309-572X"},"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":"Shouyi Yin","raw_affiliation_strings":["Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Beijing Innovation Center for Future Chip and the Beijing National Research Center for Information Science and Technology, School of Integrated Circuits, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5021484887"],"corresponding_institution_ids":["https://openalex.org/I99065089"],"apc_list":null,"apc_paid":null,"fwci":1.9329,"has_fulltext":false,"cited_by_count":25,"citation_normalized_percentile":{"value":0.85823544,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":100},"biblio":{"volume":"69","issue":"10","first_page":"4042","last_page":"4055"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9998999834060669,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9998999834060669,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9997000098228455,"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/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.9988999962806702,"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/computer-science","display_name":"Computer science","score":0.6791177988052368},{"id":"https://openalex.org/keywords/dataflow","display_name":"Dataflow","score":0.6284868717193604},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.5934224128723145},{"id":"https://openalex.org/keywords/domain","display_name":"Domain (mathematical analysis)","score":0.46044716238975525},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.45739421248435974},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.4513595700263977},{"id":"https://openalex.org/keywords/logarithm","display_name":"Logarithm","score":0.4441553056240082},{"id":"https://openalex.org/keywords/edge-device","display_name":"Edge device","score":0.4408068060874939},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.4327367842197418},{"id":"https://openalex.org/keywords/floating-point","display_name":"Floating point","score":0.4268222153186798},{"id":"https://openalex.org/keywords/energy","display_name":"Energy (signal processing)","score":0.4229903519153595},{"id":"https://openalex.org/keywords/operand","display_name":"Operand","score":0.4100439250469208},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.39840370416641235},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.35111117362976074},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.26851898431777954},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.25575897097587585},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.18788790702819824},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.1537427306175232}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6791177988052368},{"id":"https://openalex.org/C96324660","wikidata":"https://www.wikidata.org/wiki/Q205446","display_name":"Dataflow","level":2,"score":0.6284868717193604},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.5934224128723145},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.46044716238975525},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.45739421248435974},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.4513595700263977},{"id":"https://openalex.org/C39927690","wikidata":"https://www.wikidata.org/wiki/Q11197","display_name":"Logarithm","level":2,"score":0.4441553056240082},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.4408068060874939},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.4327367842197418},{"id":"https://openalex.org/C84211073","wikidata":"https://www.wikidata.org/wiki/Q117879","display_name":"Floating point","level":2,"score":0.4268222153186798},{"id":"https://openalex.org/C186370098","wikidata":"https://www.wikidata.org/wiki/Q442787","display_name":"Energy (signal processing)","level":2,"score":0.4229903519153595},{"id":"https://openalex.org/C55526617","wikidata":"https://www.wikidata.org/wiki/Q719375","display_name":"Operand","level":2,"score":0.4100439250469208},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.39840370416641235},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.35111117362976074},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.26851898431777954},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.25575897097587585},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.18788790702819824},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.1537427306175232},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.0},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tcsi.2022.3184115","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsi.2022.3184115","pdf_url":null,"source":{"id":"https://openalex.org/S116977442","display_name":"IEEE Transactions on Circuits and Systems I Regular Papers","issn_l":"1549-8328","issn":["1549-8328","1558-0806"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Circuits and Systems I: Regular Papers","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.9100000262260437,"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7"}],"awards":[{"id":"https://openalex.org/G338081660","display_name":null,"funder_award_id":"62125403","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6322215604","display_name":null,"funder_award_id":"Z191100007519016","funder_id":"https://openalex.org/F4320335843","funder_display_name":"Beijing Science and Technology Planning Project"},{"id":"https://openalex.org/G7017498754","display_name":null,"funder_award_id":"2018YFB2202600","funder_id":"https://openalex.org/F4320335777","funder_display_name":"National Key Research and Development Program of China"},{"id":"https://openalex.org/G7234679820","display_name":null,"funder_award_id":"92164301","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7875222260","display_name":null,"funder_award_id":"U19B2041","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320335777","display_name":"National Key Research and Development Program of China","ror":null},{"id":"https://openalex.org/F4320335843","display_name":"Beijing Science and Technology Planning Project","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":49,"referenced_works":["https://openalex.org/W1903029394","https://openalex.org/W1987609958","https://openalex.org/W1989669570","https://openalex.org/W2053987390","https://openalex.org/W2067523571","https://openalex.org/W2108598243","https://openalex.org/W2115613939","https://openalex.org/W2119144962","https://openalex.org/W2138026948","https://openalex.org/W2147698585","https://openalex.org/W2164309468","https://openalex.org/W2285660444","https://openalex.org/W2743322459","https://openalex.org/W2794478957","https://openalex.org/W2887397219","https://openalex.org/W2888885397","https://openalex.org/W2899063892","https://openalex.org/W2920954974","https://openalex.org/W2921918777","https://openalex.org/W2922220370","https://openalex.org/W2945146780","https://openalex.org/W2955474694","https://openalex.org/W2966207885","https://openalex.org/W2991012113","https://openalex.org/W3015729306","https://openalex.org/W3016829100","https://openalex.org/W3034039317","https://openalex.org/W3040850704","https://openalex.org/W3048705738","https://openalex.org/W3093933627","https://openalex.org/W3118608800","https://openalex.org/W3160017297","https://openalex.org/W3170460529","https://openalex.org/W3172967059","https://openalex.org/W3185047697","https://openalex.org/W3193784739","https://openalex.org/W6677580257","https://openalex.org/W6679349572","https://openalex.org/W6745245109","https://openalex.org/W6748319235","https://openalex.org/W6753770798","https://openalex.org/W6756007670","https://openalex.org/W6757555829","https://openalex.org/W6761694480","https://openalex.org/W6766752034","https://openalex.org/W6767032739","https://openalex.org/W6771680064","https://openalex.org/W6787972765","https://openalex.org/W6797209661"],"related_works":["https://openalex.org/W2293118914","https://openalex.org/W2998381397","https://openalex.org/W4236419692","https://openalex.org/W4232549588","https://openalex.org/W2037018251","https://openalex.org/W2544512391","https://openalex.org/W3033233036","https://openalex.org/W3091978438","https://openalex.org/W2020888328","https://openalex.org/W562550760"],"abstract_inverted_index":{"Edge":[0],"device":[1],"deep":[2],"neural":[3],"network":[4,116],"(DNN)":[5],"training":[6,32,60,82,155,198,223],"is":[7,188],"practical":[8],"to":[9,64,80,117,135,143],"improve":[10,118],"model":[11],"adaptivity":[12],"for":[13,122,152,190],"unfamiliar":[14],"datasets":[15],"while":[16,139],"avoiding":[17],"privacy":[18],"disclosure":[19],"and":[20,37,48,90,174,212],"huge":[21],"communication":[22],"cost.":[23],"Nevertheless,":[24],"apart":[25],"from":[26,51],"feed-forward":[27],"(FF)":[28],"as":[29],"inference,":[30],"DNN":[31,59],"still":[33],"requires":[34],"back-propagation":[35],"(BP)":[36],"weight":[38],"gradient":[39],"(WG),":[40],"introducing":[41],"power-consuming":[42],"floating-point":[43],"computing":[44],"requirements,":[45],"hardware":[46,124],"underutilization,":[47],"energy":[49,92,156,186,210],"bottleneck":[50],"excessive":[52],"memory":[53,150],"access.":[54],"This":[55],"paper":[56],"proposes":[57],"a":[58,73,86,103,113,127,167],"processor":[61],"named":[62],"PL-NPU":[63,165,200],"solve":[65],"the":[66,109,119,149,154,163,196],"above":[67],"challenges":[68],"with":[69,85,112,159,172,195],"three":[70],"innovations.":[71],"First,":[72],"posit-based":[74],"logarithm-domain":[75],"processing":[76],"element":[77],"(PE)":[78],"adapts":[79],"various":[81],"data":[83,137],"requirements":[84],"low":[87],"bit-width":[88],"format":[89,138],"reduces":[91],"by":[93],"transferring":[94],"complicated":[95],"arithmetics":[96],"into":[97],"simple":[98],"logarithm":[99],"domain":[100],"operation.":[101],"Second,":[102],"reconfigurable":[104],"inter-intra-channel-reuse":[105],"dataflow":[106],"dynamically":[107],"adjusts":[108],"PE":[110],"mapping":[111],"regrouping":[114],"omega":[115],"operands":[120],"reuse":[121],"higher":[123,209],"utilization.":[125],"Third,":[126],"pointed-stake-shaped":[128],"codec":[129],"unit":[130],"adaptively":[131],"compresses":[132],"small":[133],"values":[134,142],"variable-length":[136],"compressing":[140],"large":[141],"fixed-length":[144],"8b":[145],"posit":[146],"format,":[147],"reducing":[148],"access":[151],"breaking":[153],"bottleneck.":[157],"Simulated":[158],"28nm":[160],"CMOS":[161],"technology,":[162],"proposed":[164],"achieves":[166],"maximum":[168],"frequency":[169],"of":[170],"1040MHz":[171],"343mW":[173],"5.28mm":[175],"<inline-formula":[176,202,214],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[177,203,215],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[178,204,216],"<tex-math":[179,205,217],"notation=\"LaTeX\">$\\mathbf":[180],"{^{2}}$":[181],"</tex-math></inline-formula>":[182,208,220],".":[183],"The":[184],"peak":[185],"efficiency":[187,211],"3.87TFLOPS/W":[189],"0.6V":[191],"at":[192],"60MHz.":[193],"Compared":[194],"state-of-the-art":[197],"processor,":[199],"reaches":[201],"notation=\"LaTeX\">$3.75\\times":[206],"$":[207,219],"offers":[213],"notation=\"LaTeX\">$1.68\\times":[218],"speedup":[221],"when":[222],"ResNet18.":[224]},"counts_by_year":[{"year":2026,"cited_by_count":4},{"year":2025,"cited_by_count":11},{"year":2024,"cited_by_count":4},{"year":2023,"cited_by_count":5},{"year":2022,"cited_by_count":1}],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-10-10T00:00:00"}