{"id":"https://openalex.org/W3012178976","doi":"https://doi.org/10.1109/lca.2020.2979965","title":"Systolic Tensor Array: An Efficient Structured-Sparse GEMM Accelerator for Mobile CNN Inference","display_name":"Systolic Tensor Array: An Efficient Structured-Sparse GEMM Accelerator for Mobile CNN Inference","alternate_titles":[],"publication_year":2020,"publication_date":"2020-01-01","ids":{"openalex":"https://openalex.org/W3012178976","doi":"https://doi.org/10.1109/lca.2020.2979965","mag":"3012178976"},"language":"en","primary_location":{"id":"doi:10.1109/lca.2020.2979965","doi":"https://doi.org/10.1109/lca.2020.2979965","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2020.2979965","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","dongbi-b","norway-1","poland-70","russia-white-list-2"],"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 Computer Architecture Letters","raw_type":"journal-article"},"parent_works":[],"lineage":["https://openalex.org/W3012178976"],"child_works_count":0,"type":"article","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2005.08098","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5071942908","display_name":"Zhi-Gang Liu","orcid":"https://orcid.org/0000-0002-0552-495X","observed_orcids":["https://orcid.org/0000-0002-0552-495X"]},"institutions":[],"countries":[],"is_corresponding":false,"contributor_roles":[],"raw_author_name":"Zhi-Gang Liu","raw_affiliation_strings":["Arm ML Research Lab, Boston, USA"],"raw_orcid":"https://orcid.org/0000-0002-0552-495X","affiliations":[{"raw_affiliation_string":"Arm ML Research Lab, Boston, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103266146","display_name":"Paul N. Whatmough","orcid":"https://orcid.org/0000-0002-1865-6492","observed_orcids":["https://orcid.org/0000-0002-1865-6492","https://orcid.org/0000-0001-6222-4286"]},"institutions":[],"countries":[],"is_corresponding":false,"contributor_roles":[],"raw_author_name":"Paul N. Whatmough","raw_affiliation_strings":["Arm ML Research Lab, Boston, USA"],"raw_orcid":"https://orcid.org/0000-0001-6222-4286","affiliations":[{"raw_affiliation_string":"Arm ML Research Lab, Boston, USA","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5090422974","display_name":"Matthew Mattina","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"contributor_roles":[],"raw_author_name":"Matthew Mattina","raw_affiliation_strings":["Arm ML Research Lab, Boston, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Arm ML Research Lab, Boston, USA","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":{"value":2045,"currency":"USD","value_usd":2045},"apc_paid":null,"fwci":4.8084,"has_fulltext":true,"cited_by_count":130,"citation_normalized_percentile":{"value":0.96254207,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":88,"max":100},"biblio":{"volume":"19","issue":"1","first_page":"34","last_page":"37"},"is_retracted":false,"is_pulled":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.6104000210762024,"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":0.6104000210762024,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.3059000074863434,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10363","display_name":"Low-power high-performance VLSI design","score":0.03849999979138374,"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/systolic-arrays","display_name":"systolic arrays","score":0.9449999928474426},{"id":"https://openalex.org/keywords/structured-sparsity","display_name":"structured sparsity","score":0.7749999761581421},{"id":"https://openalex.org/keywords/low-precision-arithmetic","display_name":"low-precision arithmetic","score":0.8009999990463257},{"id":"https://openalex.org/keywords/power-and-area-efficiency","display_name":"area and power efficiency","score":0.5910000205039978},{"id":"https://openalex.org/keywords/microarchitecture-design","display_name":"microarchitecture design","score":0.5149999856948853}],"concepts":[{"id":"https://openalex.org/C55526617","wikidata":"https://www.wikidata.org/wiki/Q719375","display_name":"Operand","level":2,"score":0.774446964263916},{"id":"https://openalex.org/C150741067","wikidata":"https://www.wikidata.org/wiki/Q2377218","display_name":"Systolic array","level":3,"score":0.748036801815033},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7370812892913818},{"id":"https://openalex.org/C2781198647","wikidata":"https://www.wikidata.org/wiki/Q1633673","display_name":"Datapath","level":2,"score":0.7202332019805908},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.605694055557251},{"id":"https://openalex.org/C17349429","wikidata":"https://www.wikidata.org/wiki/Q1049914","display_name":"Matrix multiplication","level":3,"score":0.4920575022697449},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.48185718059539795},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.4786316454410553},{"id":"https://openalex.org/C2777210771","wikidata":"https://www.wikidata.org/wiki/Q4927124","display_name":"Block (permutation group theory)","level":2,"score":0.469363272190094},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4485420882701874},{"id":"https://openalex.org/C96324660","wikidata":"https://www.wikidata.org/wiki/Q205446","display_name":"Dataflow","level":2,"score":0.43908417224884033},{"id":"https://openalex.org/C56372850","wikidata":"https://www.wikidata.org/wiki/Q1050404","display_name":"Sparse matrix","level":3,"score":0.41639289259910583},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3828250765800476},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.36789992451667786},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.35727426409721375},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.20888912677764893},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.17573386430740356},{"id":"https://openalex.org/C14580979","wikidata":"https://www.wikidata.org/wiki/Q876049","display_name":"Very-large-scale integration","level":2,"score":0.17228388786315918},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.13438114523887634},{"id":"https://openalex.org/C555944384","wikidata":"https://www.wikidata.org/wiki/Q249","display_name":"Wireless","level":2,"score":0.09023675322532654},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","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/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C84114770","wikidata":"https://www.wikidata.org/wiki/Q46344","display_name":"Quantum","level":2,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/lca.2020.2979965","doi":"https://doi.org/10.1109/lca.2020.2979965","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2020.2979965","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","dongbi-b","norway-1","poland-70","russia-white-list-2"],"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 Computer Architecture Letters","raw_type":"journal-article"},{"id":"pmh:oai:arXiv.org:2005.08098","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2005.08098","pdf_url":"https://arxiv.org/pdf/2005.08098","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2005.08098","doi":"https://doi.org/10.48550/arxiv.2005.08098","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2005.08098","pdf_url":"https://arxiv.org/pdf/2005.08098","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2005.08098","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2005.08098","pdf_url":"https://arxiv.org/pdf/2005.08098","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W3012178976.pdf","grobid_xml":"https://content.openalex.org/works/W3012178976.grobid-xml"},"referenced_works_count":12,"referenced_works":["https://openalex.org/W2285660444","https://openalex.org/W2289252105","https://openalex.org/W2513419314","https://openalex.org/W2606722458","https://openalex.org/W2625457103","https://openalex.org/W2767785892","https://openalex.org/W2773134966","https://openalex.org/W2900228909","https://openalex.org/W2915654079","https://openalex.org/W2949660525","https://openalex.org/W2950656546","https://openalex.org/W2953212265"],"related_works":["https://openalex.org/W3083448369","https://openalex.org/W3185702163","https://openalex.org/W4224612674","https://openalex.org/W3207820301","https://openalex.org/W4410583693","https://openalex.org/W4382203333","https://openalex.org/W4320712936","https://openalex.org/W7162681955","https://openalex.org/W4387302793","https://openalex.org/W4400232637"],"abstract_inverted_index":{"Convolutional":[0],"neural":[1],"network":[2],"(CNN)":[3],"inference":[4],"on":[5],"mobile":[6],"devices":[7],"demands":[8],"efficient":[9,34],"hardware":[10],"acceleration":[11],"of":[12,28,83,108],"low-precision":[13],"(INT8)":[14],"general":[15],"matrix":[16],"multiplication":[17],"(GEMM).":[18],"The":[19,90],"systolic":[20],"array":[21,27],"(SA)":[22],"is":[23],"a":[24,75,81,133,146],"pipelined":[25],"2D":[26],"processing":[29,163],"elements":[30],"(PEs),":[31],"with":[32,123,172],"very":[33],"local":[35],"data":[36,136],"movement,":[37],"well":[38],"suited":[39],"to":[40,56,61,80,117,131],"accelerating":[41],"GEMM,":[42],"and":[43,97,104,113,148,159],"widely":[44],"deployed":[45],"in":[46,101,157],"industry.":[47],"In":[48],"this":[49,129],"letter,":[50],"we":[51,68,127],"describe":[52],"two":[53],"significant":[54],"improvements":[55],"the":[57,70,118,152],"traditional":[58,71],"SA":[59,120,153],"architecture,":[60],"specifically":[62],"optimize":[63],"for":[64],"CNN":[65],"inference.":[66],"First,":[67],"generalize":[69],"scalar":[72],"PE,":[73],"into":[74],"Tensor-PE,":[76],"which":[77],"gives":[78],"rise":[79],"family":[82,92],"new":[84],"Systolic":[85],"Tensor":[86],"Array":[87],"(STA)":[88],"microarchitectures.":[89],"STA":[91],"increases":[93],"intra-PE":[94],"operand":[95],"reuse":[96],"datapath":[98],"efficiency,":[99],"resulting":[100],"circuit":[102],"area":[103,158],"power":[105,160],"dissipation":[106],"reduction":[107],"as":[109,111],"much":[110],"2.08\u00d7":[112],"1.36\u00d7":[114],"respectively,":[115,161],"compared":[116],"conventional":[119],"at":[121,155],"iso-throughput":[122,156],"INT8":[124],"operands.":[125],"Second,":[126],"extend":[128],"design":[130],"support":[132],"novel":[134],"block-sparse":[135],"format":[137],"called":[138],"density-bound":[139],"block":[140],"(DBB).":[141],"This":[142],"variant":[143],"(STA-DBB)":[144],"achieves":[145],"3.14\u00d7":[147],"1.97\u00d7":[149],"improvement":[150],"over":[151],"baseline":[154],"when":[162],"specially-trained":[164],"DBB-sparse":[165],"models,":[166],"while":[167],"remaining":[168],"fully":[169],"backwards":[170],"compatible":[171],"dense":[173],"models.":[174]},"cited_by_api_url":"https://api.openalex.org/works?filter=cites:W3012178976","counts_by_year":[{"year":2026,"cited_by_count":18},{"year":2025,"cited_by_count":25},{"year":2024,"cited_by_count":24},{"year":2023,"cited_by_count":25},{"year":2022,"cited_by_count":15},{"year":2021,"cited_by_count":17},{"year":2020,"cited_by_count":5},{"year":2019,"cited_by_count":1}],"updated_date":"2026-11-20T13:11:24.684918","created_date":"2020-03-23T00:00:00"}
