{"id":"https://openalex.org/W4281790033","doi":"https://doi.org/10.1145/3470496.3527382","title":"Themis","display_name":"Themis","publication_year":2022,"publication_date":"2022-05-31","ids":{"openalex":"https://openalex.org/W4281790033","doi":"https://doi.org/10.1145/3470496.3527382"},"language":"en","primary_location":{"id":"doi:10.1145/3470496.3527382","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3470496.3527382","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3470496.3527382","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 49th Annual International Symposium on Computer Architecture","raw_type":"proceedings-article"},"type":"preprint","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3470496.3527382","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5088899364","display_name":"Saeed Rashidi","orcid":"https://orcid.org/0000-0002-6472-9920"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Saeed Rashidi","raw_affiliation_strings":["Georgia Institute of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5068175490","display_name":"William Won","orcid":"https://orcid.org/0000-0002-1715-9144"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"William Won","raw_affiliation_strings":["Georgia Institute of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065143008","display_name":"Sudarshan Srinivasan","orcid":"https://orcid.org/0009-0002-8662-5820"},"institutions":[{"id":"https://openalex.org/I4210146682","display_name":"Intel (India)","ror":"https://ror.org/04f2n1245","country_code":"IN","type":"company","lineage":["https://openalex.org/I1343180700","https://openalex.org/I4210146682"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Sudarshan Srinivasan","raw_affiliation_strings":["Intel, Bangalore, Karnataka, India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Intel, Bangalore, Karnataka, India","institution_ids":["https://openalex.org/I4210146682"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103751730","display_name":"Srinivas Sridharan","orcid":null},"institutions":[{"id":"https://openalex.org/I4210099336","display_name":"Menlo School","ror":"https://ror.org/01240pn49","country_code":"US","type":"education","lineage":["https://openalex.org/I4210099336"]},{"id":"https://openalex.org/I4210146682","display_name":"Intel (India)","ror":"https://ror.org/04f2n1245","country_code":"IN","type":"company","lineage":["https://openalex.org/I1343180700","https://openalex.org/I4210146682"]},{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["IN","US"],"is_corresponding":false,"raw_author_name":"Srinivas Sridharan","raw_affiliation_strings":["Meta","AG,C1","AG, C2","Meta Menlo Park, California, USA","Intel Bangalore, Karnataka, India","Georgia Institute of Technology Atlanta, Georgia, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Meta","institution_ids":[]},{"raw_affiliation_string":"AG,C1","institution_ids":[]},{"raw_affiliation_string":"AG, C2","institution_ids":[]},{"raw_affiliation_string":"Meta Menlo Park, California, USA","institution_ids":["https://openalex.org/I4210099336"]},{"raw_affiliation_string":"Intel Bangalore, Karnataka, India","institution_ids":["https://openalex.org/I4210146682"]},{"raw_affiliation_string":"Georgia Institute of Technology Atlanta, Georgia, USA","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5034089074","display_name":"Tushar Krishna","orcid":"https://orcid.org/0000-0001-5738-6942"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Tushar Krishna","raw_affiliation_strings":["Georgia Institute of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology","institution_ids":["https://openalex.org/I130701444"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5088899364"],"corresponding_institution_ids":["https://openalex.org/I130701444"],"apc_list":null,"apc_paid":null,"fwci":2.4939,"has_fulltext":true,"cited_by_count":32,"citation_normalized_percentile":{"value":0.89389305,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"581","last_page":"596"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.998199999332428,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.998199999332428,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9980000257492065,"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":0.9980000257492065,"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":0.7992531061172485},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.6270597577095032},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.6223844289779663},{"id":"https://openalex.org/keywords/leverage","display_name":"Leverage (statistics)","score":0.5594719648361206},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.4385409653186798},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.33334869146347046},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.2153383493423462},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.11691996455192566},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.08723661303520203},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.07125279307365417}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7992531061172485},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.6270597577095032},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.6223844289779663},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.5594719648361206},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.4385409653186798},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.33334869146347046},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2153383493423462},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.11691996455192566},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.08723661303520203},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.07125279307365417}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3470496.3527382","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3470496.3527382","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3470496.3527382","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 49th Annual International Symposium on Computer Architecture","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3470496.3527382","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3470496.3527382","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3470496.3527382","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 49th Annual International Symposium on Computer Architecture","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320307102","display_name":"Intel Corporation","ror":"https://ror.org/01ek73717"},{"id":"https://openalex.org/F4320319290","display_name":"Meta","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4281790033.pdf","grobid_xml":"https://content.openalex.org/works/W4281790033.grobid-xml"},"referenced_works_count":29,"referenced_works":["https://openalex.org/W1937799376","https://openalex.org/W2009571103","https://openalex.org/W2057332538","https://openalex.org/W2099030945","https://openalex.org/W2128543830","https://openalex.org/W2131613942","https://openalex.org/W2136041296","https://openalex.org/W2142763683","https://openalex.org/W2150871235","https://openalex.org/W2170796499","https://openalex.org/W2625200202","https://openalex.org/W2949161920","https://openalex.org/W2969388332","https://openalex.org/W2980104813","https://openalex.org/W2981114289","https://openalex.org/W3016395792","https://openalex.org/W3043522163","https://openalex.org/W3072623287","https://openalex.org/W3085163101","https://openalex.org/W3092501829","https://openalex.org/W3096425133","https://openalex.org/W3114882619","https://openalex.org/W3129488589","https://openalex.org/W3129831491","https://openalex.org/W3188065709","https://openalex.org/W3197720002","https://openalex.org/W3206778761","https://openalex.org/W4210984760","https://openalex.org/W4225898610"],"related_works":["https://openalex.org/W2787993192","https://openalex.org/W2158269427","https://openalex.org/W4381280689","https://openalex.org/W2847365777","https://openalex.org/W3128025644","https://openalex.org/W2355048207","https://openalex.org/W2952348651","https://openalex.org/W2111375262","https://openalex.org/W1991205173","https://openalex.org/W2750422482"],"abstract_inverted_index":{"Distributed":[0],"training":[1,8,21,45,145],"is":[2],"a":[3,62,94],"solution":[4],"to":[5,30,106],"reduce":[6],"DNN":[7],"time":[9],"by":[10,137,158],"splitting":[11],"the":[12,26,32,38,73,77,108,116,129,134,143],"task":[13],"across":[14,111],"multiple":[15],"NPUs":[16,27,48],"(e.g.,":[17],"GPU/TPU).":[18],"However,":[19],"distributed":[20],"adds":[22],"communication":[23,87,109],"overhead":[24],"between":[25],"in":[28],"order":[29],"synchronize":[31],"gradients":[33],"and/or":[34],"activation,":[35],"depending":[36],"on":[37,88,124],"parallelization":[39],"strategy.":[40],"In":[41],"next-generation":[42],"platforms":[43],"for":[44,85],"at":[46],"scale,":[47],"will":[49],"be":[50],"connected":[51],"through":[52],"multidimensional":[53],"networks":[54],"with":[55],"diverse,":[56],"heterogeneous":[57],"bandwidths.":[58],"This":[59],"work":[60],"identifies":[61],"looming":[63],"challenge":[64],"of":[65,133,148],"keeping":[66],"all":[67,112],"network":[68,74,117,130],"dimensions":[69],"busy":[70],"and":[71,141,156,168],"maximizing":[72],"BW":[75,118,131],"within":[76],"hybrid":[78],"environment":[79],"if":[80],"we":[81],"leverage":[82],"scheduling":[83,97],"techniques":[84],"collective":[86,96],"systems":[89],"today.":[90],"We":[91],"propose":[92],"Themis,":[93],"novel":[95],"scheme":[98],"that":[99,123],"dynamically":[100],"schedules":[101],"collectives":[102],"(divided":[103],"into":[104],"chunks)":[105],"balance":[107],"loads":[110],"dimensions,":[113],"further":[114],"improving":[115],"utilization.":[119],"Our":[120],"results":[121],"show":[122],"average,":[125],"Themis":[126],"can":[127],"improve":[128,142],"utilization":[132],"single":[135],"All-Reduce":[136],"1.72\u00d7":[138],"(2.70\u00d7":[139],"max),":[140,161,164,167,171],"end-to-end":[144],"iteration":[146],"performance":[147],"real":[149],"workloads":[150],"such":[151],"as":[152],"ResNet-152,":[153],"GNMT,":[154],"DLRM,":[155],"Transformer-1T":[157],"1.49\u00d7":[159],"(2.25\u00d7":[160],"1.30\u00d7":[162,165],"(1.78\u00d7":[163],"(1.77\u00d7":[166],"1.25\u00d7":[169],"(1.53\u00d7":[170],"respectively.":[172]},"counts_by_year":[{"year":2026,"cited_by_count":5},{"year":2025,"cited_by_count":13},{"year":2024,"cited_by_count":10},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":1}],"updated_date":"2026-04-25T08:17:42.794288","created_date":"2022-06-13T00:00:00"}