{"id":"https://openalex.org/W4309672181","doi":"https://doi.org/10.1109/tpds.2022.3217824","title":"Dissecting Tensor Cores via Microbenchmarks: Latency, Throughput and Numeric Behaviors","display_name":"Dissecting Tensor Cores via Microbenchmarks: Latency, Throughput and Numeric Behaviors","publication_year":2022,"publication_date":"2022-11-04","ids":{"openalex":"https://openalex.org/W4309672181","doi":"https://doi.org/10.1109/tpds.2022.3217824"},"language":"en","primary_location":{"id":"doi:10.1109/tpds.2022.3217824","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2022.3217824","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"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 Parallel and Distributed Systems","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://www.osti.gov/servlets/purl/1902259","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5059737143","display_name":"Wei Sun","orcid":"https://orcid.org/0000-0003-2873-6418"},"institutions":[{"id":"https://openalex.org/I83019370","display_name":"Eindhoven University of Technology","ror":"https://ror.org/02c2kyt77","country_code":"NL","type":"education","lineage":["https://openalex.org/I83019370"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Wei Sun","raw_affiliation_strings":["Electronic System Group, Eindhoven University of Technology, Eindhoven, AZ, The Netherlands"],"raw_orcid":"https://orcid.org/0000-0003-2873-6418","affiliations":[{"raw_affiliation_string":"Electronic System Group, Eindhoven University of Technology, Eindhoven, AZ, The Netherlands","institution_ids":["https://openalex.org/I83019370"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100413657","display_name":"Ang Li","orcid":"https://orcid.org/0000-0003-3734-9137"},"institutions":[{"id":"https://openalex.org/I142606810","display_name":"Pacific Northwest National Laboratory","ror":"https://ror.org/05h992307","country_code":"US","type":"facility","lineage":["https://openalex.org/I1325736334","https://openalex.org/I1330989302","https://openalex.org/I142606810","https://openalex.org/I39565521"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ang Li","raw_affiliation_strings":["Physical and Computational Sciences Directorate, Pacific Northwest National Laboratory, Richland, WA, USA"],"raw_orcid":"https://orcid.org/0000-0003-3734-9137","affiliations":[{"raw_affiliation_string":"Physical and Computational Sciences Directorate, Pacific Northwest National Laboratory, Richland, WA, USA","institution_ids":["https://openalex.org/I142606810"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078443672","display_name":"Tong Geng","orcid":"https://orcid.org/0000-0002-3644-2922"},"institutions":[{"id":"https://openalex.org/I142606810","display_name":"Pacific Northwest National Laboratory","ror":"https://ror.org/05h992307","country_code":"US","type":"facility","lineage":["https://openalex.org/I1325736334","https://openalex.org/I1330989302","https://openalex.org/I142606810","https://openalex.org/I39565521"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Tong Geng","raw_affiliation_strings":["Physical and Computational Sciences Directorate, Pacific Northwest National Laboratory, Richland, WA, USA"],"raw_orcid":"https://orcid.org/0000-0002-3644-2922","affiliations":[{"raw_affiliation_string":"Physical and Computational Sciences Directorate, Pacific Northwest National Laboratory, Richland, WA, USA","institution_ids":["https://openalex.org/I142606810"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032436846","display_name":"Sander Stuijk","orcid":"https://orcid.org/0000-0002-2518-6847"},"institutions":[{"id":"https://openalex.org/I83019370","display_name":"Eindhoven University of Technology","ror":"https://ror.org/02c2kyt77","country_code":"NL","type":"education","lineage":["https://openalex.org/I83019370"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Sander Stuijk","raw_affiliation_strings":["Electronic System Group, Eindhoven University of Technology, Eindhoven, AZ, The Netherlands"],"raw_orcid":"https://orcid.org/0000-0002-2518-6847","affiliations":[{"raw_affiliation_string":"Electronic System Group, Eindhoven University of Technology, Eindhoven, AZ, The Netherlands","institution_ids":["https://openalex.org/I83019370"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5081768631","display_name":"Henk Corporaal","orcid":"https://orcid.org/0000-0003-4506-5732"},"institutions":[{"id":"https://openalex.org/I83019370","display_name":"Eindhoven University of Technology","ror":"https://ror.org/02c2kyt77","country_code":"NL","type":"education","lineage":["https://openalex.org/I83019370"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Henk Corporaal","raw_affiliation_strings":["Electronic System Group, Eindhoven University of Technology, Eindhoven, AZ, The Netherlands"],"raw_orcid":"https://orcid.org/0000-0003-4506-5732","affiliations":[{"raw_affiliation_string":"Electronic System Group, Eindhoven University of Technology, Eindhoven, AZ, The Netherlands","institution_ids":["https://openalex.org/I83019370"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":8.6773,"has_fulltext":true,"cited_by_count":51,"citation_normalized_percentile":{"value":0.98626926,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":99,"max":100},"biblio":{"volume":"34","issue":"1","first_page":"246","last_page":"261"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9994999766349792,"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"}},"topics":[{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9994999766349792,"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/T13650","display_name":"Computational Physics and Python Applications","score":0.9961000084877014,"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"}},{"id":"https://openalex.org/T11206","display_name":"Model Reduction and Neural Networks","score":0.9739999771118164,"subfield":{"id":"https://openalex.org/subfields/3109","display_name":"Statistical and Nonlinear Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7090193033218384},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.6118135452270508},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.603569507598877},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.37212300300598145},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.08997783064842224}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7090193033218384},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.6118135452270508},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.603569507598877},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.37212300300598145},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.08997783064842224},{"id":"https://openalex.org/C555944384","wikidata":"https://www.wikidata.org/wiki/Q249","display_name":"Wireless","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}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/tpds.2022.3217824","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2022.3217824","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"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 Parallel and Distributed Systems","raw_type":"journal-article"},{"id":"pmh:oai:osti.gov:1902259","is_oa":true,"landing_page_url":"https://www.osti.gov/biblio/1902259","pdf_url":"https://www.osti.gov/servlets/purl/1902259","source":{"id":"https://openalex.org/S4306402487","display_name":"OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I139351228","host_organization_name":"Office of Scientific and Technical Information","host_organization_lineage":["https://openalex.org/I139351228"],"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":null},{"id":"pmh:oai:pure.tue.nl:openaire_cris_publications/44b46a8d-4c5a-4c51-90ca-1cd62811902f","is_oa":true,"landing_page_url":"https://research.tue.nl/en/publications/44b46a8d-4c5a-4c51-90ca-1cd62811902f","pdf_url":"https://pure.tue.nl/ws/files/328202544/Dissecting_Tensor_Cores_via_Microbenchmarks_Latency_Throughput_and_Numeric_Behaviors.pdf","source":{"id":"https://openalex.org/S4406922641","display_name":"TU/e Research Portal","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Sun, W, Li, A, Geng, T, Stuijk, S & Corporaal, H 2023, 'Dissecting Tensor Cores via Microbenchmarks : Latency, Throughput and Numeric Behaviors', IEEE Transactions on Parallel and Distributed Systems, vol. 34, no. 1, 9931992, pp. 246-261. https://doi.org/10.1109/TPDS.2022.3217824","raw_type":"info:eu-repo/semantics/article"}],"best_oa_location":{"id":"pmh:oai:osti.gov:1902259","is_oa":true,"landing_page_url":"https://www.osti.gov/biblio/1902259","pdf_url":"https://www.osti.gov/servlets/purl/1902259","source":{"id":"https://openalex.org/S4306402487","display_name":"OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I139351228","host_organization_name":"Office of Scientific and Technical Information","host_organization_lineage":["https://openalex.org/I139351228"],"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":null},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.46000000834465027}],"awards":[{"id":"https://openalex.org/G1791724952","display_name":null,"funder_award_id":"78284","funder_id":"https://openalex.org/F4320337506","funder_display_name":"Advanced Scientific Computing Research"},{"id":"https://openalex.org/G1899721271","display_name":null,"funder_award_id":"AC05-76RL01830","funder_id":"https://openalex.org/F4320306250","funder_display_name":"Battelle"},{"id":"https://openalex.org/G1967746503","display_name":null,"funder_award_id":"DE-AC05-76RL01830","funder_id":"https://openalex.org/F4320338354","funder_display_name":"Pacific Northwest National Laboratory"},{"id":"https://openalex.org/G2014121706","display_name":null,"funder_award_id":"ZERO-ARM P3","funder_id":"https://openalex.org/F4320321800","funder_display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek"},{"id":"https://openalex.org/G2506945562","display_name":null,"funder_award_id":"76RL01830","funder_id":"https://openalex.org/F4320306084","funder_display_name":"U.S. Department of Energy"},{"id":"https://openalex.org/G2510423872","display_name":null,"funder_award_id":"AC05-76RL01830","funder_id":"https://openalex.org/F4320306084","funder_display_name":"U.S. Department of Energy"},{"id":"https://openalex.org/G2865512842","display_name":null,"funder_award_id":"DE-AC05-76RL01830","funder_id":"https://openalex.org/F4320337506","funder_display_name":"Advanced Scientific Computing Research"},{"id":"https://openalex.org/G3620080081","display_name":null,"funder_award_id":"78284","funder_id":"https://openalex.org/F4320306084","funder_display_name":"U.S. Department of Energy"},{"id":"https://openalex.org/G5200347334","display_name":null,"funder_award_id":"DE-AC05-76RL01830","funder_id":"https://openalex.org/F4320306084","funder_display_name":"U.S. Department of Energy"},{"id":"https://openalex.org/G5321382714","display_name":null,"funder_award_id":"76RL01830","funder_id":"https://openalex.org/F4320338354","funder_display_name":"Pacific Northwest National Laboratory"},{"id":"https://openalex.org/G5726405315","display_name":null,"funder_award_id":"DE-AC05","funder_id":"https://openalex.org/F4320306250","funder_display_name":"Battelle"},{"id":"https://openalex.org/G6664334341","display_name":null,"funder_award_id":"DE-AC05-76RL01830","funder_id":"https://openalex.org/F4320306250","funder_display_name":"Battelle"},{"id":"https://openalex.org/G7125981805","display_name":null,"funder_award_id":"DE-AC05-76RL01830","funder_id":"https://openalex.org/F4320332359","funder_display_name":"Office of Science"},{"id":"https://openalex.org/G7995982022","display_name":null,"funder_award_id":"DE-AC05","funder_id":"https://openalex.org/F4320306084","funder_display_name":"U.S. Department of Energy"}],"funders":[{"id":"https://openalex.org/F4320306084","display_name":"U.S. Department of Energy","ror":"https://ror.org/01bj3aw27"},{"id":"https://openalex.org/F4320306250","display_name":"Battelle","ror":"https://ror.org/01h5tnr73"},{"id":"https://openalex.org/F4320321800","display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek","ror":"https://ror.org/04jsz6e67"},{"id":"https://openalex.org/F4320332359","display_name":"Office of Science","ror":"https://ror.org/00mmn6b08"},{"id":"https://openalex.org/F4320337506","display_name":"Advanced Scientific Computing Research","ror":"https://ror.org/0012c7r22"},{"id":"https://openalex.org/F4320338354","display_name":"Pacific Northwest National Laboratory","ror":"https://ror.org/05h992307"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4309672181.pdf","grobid_xml":"https://content.openalex.org/works/W4309672181.grobid-xml"},"referenced_works_count":44,"referenced_works":["https://openalex.org/W1762731526","https://openalex.org/W1979527452","https://openalex.org/W1980282429","https://openalex.org/W2110195531","https://openalex.org/W2163687928","https://openalex.org/W2172654076","https://openalex.org/W2410461480","https://openalex.org/W2580538010","https://openalex.org/W2606722458","https://openalex.org/W2618168961","https://openalex.org/W2763421725","https://openalex.org/W2791673912","https://openalex.org/W2796649226","https://openalex.org/W2891325938","https://openalex.org/W2895305554","https://openalex.org/W2907701003","https://openalex.org/W2914086833","https://openalex.org/W2918037051","https://openalex.org/W2920798074","https://openalex.org/W2926767350","https://openalex.org/W2962729196","https://openalex.org/W2963989532","https://openalex.org/W2998957070","https://openalex.org/W3000305333","https://openalex.org/W3021029305","https://openalex.org/W3031930374","https://openalex.org/W3043303806","https://openalex.org/W3047486355","https://openalex.org/W3101543398","https://openalex.org/W3101708369","https://openalex.org/W3117765988","https://openalex.org/W3127067080","https://openalex.org/W3132532188","https://openalex.org/W3175878603","https://openalex.org/W3204140704","https://openalex.org/W3205226039","https://openalex.org/W4220912491","https://openalex.org/W4287327026","https://openalex.org/W6745245109","https://openalex.org/W6750448596","https://openalex.org/W6760272376","https://openalex.org/W6789946416","https://openalex.org/W6790943123","https://openalex.org/W6803607599"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Tensor":[0,23,52,75,107,133],"Cores":[1,108,134],"have":[2,26],"been":[3,86],"an":[4],"important":[5],"unit":[6],"to":[7,27],"accelerate":[8],"Fused":[9],"Matrix":[10],"Multiplication":[11],"Accumulation":[12],"(MMA)":[13],"in":[14,152,158],"all":[15],"NVIDIA":[16],"GPUs":[17],"since":[18],"Volta":[19],"Architecture.":[20],"To":[21],"program":[22],"Cores,":[24],"users":[25],"use":[28],"either":[29],"legacy":[30],"wmma":[31,38,55],"APIs":[32,39,56],"or":[33],"current":[34,81,122],"mma":[35],"APIs.":[36,124],"Legacy":[37],"are":[40,109],"more":[41],"easy-to-use":[42],"but":[43],"can":[44,62,155],"only":[45],"exploit":[46],"limited":[47],"features":[48],"and":[49,61,100,119,136,147],"power":[50],"of":[51,71,80,94,121,132,144],"Cores.":[53,76],"Specifically,":[54],"support":[57],"fewer":[58],"operand":[59],"shapes":[60],"not":[63,85],"leverage":[64],"the":[65,72,78,90,104,117,129,138],"new":[66],"sparse":[67],"matrix":[68],"multiplication":[69],"feature":[70],"newest":[73,105],"Ampere":[74,106],"However,":[77],"performance":[79],"programming":[82,123],"interface":[83],"has":[84],"well":[87],"explored.":[88],"Furthermore,":[89],"computation":[91],"numeric":[92,130],"behaviors":[93,131],"low-precision":[95],"floating":[96],"points":[97],"(TF32,":[98],"BF16,":[99],"FP16)":[101],"supported":[102],"by":[103],"also":[110,126],"mysterious.":[111],"In":[112],"this":[113,153],"paper,":[114],"we":[115],"explore":[116],"throughput":[118],"latency":[120],"We":[125],"intuitively":[127],"study":[128],"MMA":[135],"profile":[137],"intermediate":[139],"operations":[140],"including":[141],"multiplication,":[142],"addition":[143],"inner":[145],"product,":[146],"accumulation.":[148],"All":[149],"codes":[150],"used":[151],"work":[154],"be":[156],"found":[157],"<uri":[159],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[160],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">https://github.com/sunlex0717/DissectingTensorCores</uri>":[161],".":[162]},"counts_by_year":[{"year":2026,"cited_by_count":8},{"year":2025,"cited_by_count":20},{"year":2024,"cited_by_count":14},{"year":2023,"cited_by_count":9}],"updated_date":"2026-07-18T07:39:51.176621","created_date":"2022-11-29T00:00:00"}