{"id":"https://openalex.org/W4285117917","doi":"https://doi.org/10.1109/tcsi.2022.3184175","title":"SWPU: A 126.04 TFLOPS/W Edge-Device Sparse DNN Training Processor With Dynamic Sub-Structured Weight Pruning","display_name":"SWPU: A 126.04 TFLOPS/W Edge-Device Sparse DNN Training Processor With Dynamic Sub-Structured Weight Pruning","publication_year":2022,"publication_date":"2022-06-23","ids":{"openalex":"https://openalex.org/W4285117917","doi":"https://doi.org/10.1109/tcsi.2022.3184175"},"language":"en","primary_location":{"id":"doi:10.1109/tcsi.2022.3184175","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsi.2022.3184175","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":false,"raw_author_name":"Yang Wang","raw_affiliation_strings":["School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-8293-8881","affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5071432313","display_name":"Yubin Qin","orcid":"https://orcid.org/0000-0001-5530-5416"},"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":"Yubin Qin","raw_affiliation_strings":["School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-5530-5416","affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, 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":["School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-7548-4116","affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, 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":["School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-5117-7920","affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, 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":["School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-2309-572X","affiliations":[{"raw_affiliation_string":"School of Integrated Circuits, the Beijing Innovation Center for Future Chip, and the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I99065089"],"apc_list":null,"apc_paid":null,"fwci":1.664,"has_fulltext":false,"cited_by_count":18,"citation_normalized_percentile":{"value":0.85040193,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":99},"biblio":{"volume":"69","issue":"10","first_page":"4014","last_page":"4027"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998999834060669,"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":0.9998999834060669,"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":0.9998999834060669,"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":0.9984999895095825,"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":0.7909533977508545},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.6750692129135132},{"id":"https://openalex.org/keywords/edge-device","display_name":"Edge device","score":0.5893387794494629},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.5848286151885986},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5242921710014343},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.49788594245910645},{"id":"https://openalex.org/keywords/operand","display_name":"Operand","score":0.45602720975875854},{"id":"https://openalex.org/keywords/granularity","display_name":"Granularity","score":0.45012521743774414},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.43226441740989685},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.4274868369102478},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.419575959444046},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.41817498207092285},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.33644455671310425},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.27397122979164124},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.2624739408493042}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7909533977508545},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6750692129135132},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.5893387794494629},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5848286151885986},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5242921710014343},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.49788594245910645},{"id":"https://openalex.org/C55526617","wikidata":"https://www.wikidata.org/wiki/Q719375","display_name":"Operand","level":2,"score":0.45602720975875854},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.45012521743774414},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.43226441740989685},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.4274868369102478},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.419575959444046},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.41817498207092285},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.33644455671310425},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.27397122979164124},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.2624739408493042},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0},{"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/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tcsi.2022.3184175","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsi.2022.3184175","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":[],"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/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}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":36,"referenced_works":["https://openalex.org/W2194775991","https://openalex.org/W2657126969","https://openalex.org/W2788653909","https://openalex.org/W2808133870","https://openalex.org/W2899481200","https://openalex.org/W2906043559","https://openalex.org/W2906663849","https://openalex.org/W2920954974","https://openalex.org/W2921918777","https://openalex.org/W2962843773","https://openalex.org/W2963037989","https://openalex.org/W2963163009","https://openalex.org/W2963446712","https://openalex.org/W2990613095","https://openalex.org/W3015729306","https://openalex.org/W3034971973","https://openalex.org/W3048842964","https://openalex.org/W3091592563","https://openalex.org/W3102587717","https://openalex.org/W3109309915","https://openalex.org/W3118608800","https://openalex.org/W3135859967","https://openalex.org/W3185047697","https://openalex.org/W4212774754","https://openalex.org/W6637373629","https://openalex.org/W6638632666","https://openalex.org/W6684191040","https://openalex.org/W6726275242","https://openalex.org/W6745668070","https://openalex.org/W6748522439","https://openalex.org/W6751979845","https://openalex.org/W6757914502","https://openalex.org/W6762718338","https://openalex.org/W6766978945","https://openalex.org/W6779101013","https://openalex.org/W6787972765"],"related_works":["https://openalex.org/W4293869292","https://openalex.org/W2988454569","https://openalex.org/W2949961122","https://openalex.org/W2233116163","https://openalex.org/W4297853509","https://openalex.org/W3098411449","https://openalex.org/W3094360146","https://openalex.org/W4287635891","https://openalex.org/W3164228065","https://openalex.org/W4285117917"],"abstract_inverted_index":{"When":[0,307],"deploying":[1],"deep":[2],"neural":[3],"networks":[4],"(DNNs),":[5],"edge":[6,30],"devices":[7],"training":[8,25,36,43,71,296,308,335],"is":[9,27,85,291],"practical":[10],"to":[11,74,126],"improve":[12],"model":[13],"adaptivity":[14],"for":[15,29,130],"various":[16],"user-specific":[17],"scenarios":[18],"while":[19],"avoiding":[20,238],"privacy":[21],"disclosure.":[22],"However,":[23],"the":[24,39,56,59,62,75,105,128,132,137,158,162,173,183,190,200,239,294],"computation":[26,44,95,184],"intolerable":[28],"devices.":[31],"It":[32,261],"inspires":[33],"sparse":[34,204,221,334],"DNN":[35],"(SDT)":[37],"into":[38],"limelight,":[40],"which":[41,210],"reduces":[42,313],"by":[45,198,223,298],"dynamic":[46],"weight":[47,167],"pruning.":[48,284],"Generally,":[49],"SDT":[50],"has":[51,88,99],"two":[52],"strategies":[53],"based":[54],"on":[55],"pruning":[57,78,91,102,142,266,270],"granularity:":[58],"structured":[60,269],"or":[61],"unstructured.":[63],"Unfortunately,":[64],"both":[65],"of":[66,134,202,258,289],"them":[67],"suffer":[68],"from":[69],"limited":[70,94],"efficiency":[72,133,281,288],"due":[73],"gap":[76,129],"between":[77],"granularity":[79],"and":[80,110,148,156,215,231,271,322],"hardware":[81,163],"implementation.":[82],"The":[83,97,285],"former":[84],"hardware-friendly":[86,159],"but":[87,104],"a":[89,100,122,140,152,165,207,255,263,309],"low":[90],"ratio,":[92,103],"indicating":[93],"reduction.":[96],"latter":[98],"high":[101,153],"unbalanced":[106],"workload":[107],"decreases":[108],"utilization":[109],"irregular":[111,191],"sparsity":[112,116,154,175,216],"distribution":[113],"causes":[114],"considerable":[115],"processing":[117,168],"overhead.":[118],"This":[119],"paper":[120],"proposes":[121],"software-hardware":[123],"co-":[124],"design":[125],"bridge":[127],"improving":[131],"SDT.":[135],"On":[136,161],"algorithm":[138],"side,":[139,164],"sub-structured":[141,166,203],"method,":[143],"achieved":[144],"with":[145,176,186,226,235,244,254],"hybrid":[146,174],"shape-wise":[147],"line-wise":[149],"pruning,":[150],"generates":[151],"ratio":[155,267],"keeps":[157],"property.":[160],"unit":[169],"(SWPU)":[170],"effectively":[171],"handles":[172],"three":[177],"techniques.":[178],"First,":[179],"SWPU":[180,194,219,248,290,312],"dynamically":[181],"reorders":[182],"sequence":[185],"hamming-distance-based":[187],"clustering,":[188],"balancing":[189],"workload.":[192],"Second,":[193],"performs":[195,220],"runtime":[196],"scheduling":[197],"exploiting":[199],"feature":[201],"convolution":[205,222],"through":[206],"detect-before-load":[208],"controller,":[209],"skips":[211],"redundant":[212],"memory":[213],"access":[214],"processing.":[217],"Third,":[218],"compressing":[224],"operands":[225],"spatial":[227],"disconnect":[228],"log-based":[229],"routing":[230,241],"recovers":[232],"their":[233],"location":[234],"bi-directional":[236],"switching,":[237],"power-consumed":[240],"logic.":[242],"Synthesized":[243],"28nm":[245],"CMOS":[246],"technology,":[247],"can":[249],"enable":[250],"0.56V-to-1.0V":[251],"supply":[252],"voltage":[253],"maximum":[256],"frequency":[257],"675":[259],"MHz.":[260],"achieves":[262],"50.1%":[264],"higher":[265,279],"than":[268,282,332],"<inline-formula":[272,299,314,324],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[273,300,315,325],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[274,301,316,326],"<tex-math":[275,302,317,327],"notation=\"LaTeX\">$1.53\\times":[276],"$":[277,304,319,329],"</tex-math></inline-formula>":[278,305,320,330],"energy":[280,287,321],"unstructured":[283],"peak":[286],"126.04TFLOPS/W,":[292],"outperforming":[293],"state-of-the-art":[295],"processor":[297],"notation=\"LaTeX\">$1.67\\times":[303],".":[306],"ResNet-18":[310],"model,":[311],"notation=\"LaTeX\">$3.72\\times":[318],"offers":[323],"notation=\"LaTeX\">$4.69\\times":[328],"speedup":[331],"previous":[333],"processors.":[336]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":8},{"year":2024,"cited_by_count":4},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":2}],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-10-10T00:00:00"}