{"id":"https://openalex.org/W4388544075","doi":"https://doi.org/10.1109/tpds.2023.3331372","title":"US-Byte: An Efficient Communication Framework for Scheduling Unequal-Sized Tensor Blocks in Distributed Deep Learning","display_name":"US-Byte: An Efficient Communication Framework for Scheduling Unequal-Sized Tensor Blocks in Distributed Deep Learning","publication_year":2023,"publication_date":"2023-11-09","ids":{"openalex":"https://openalex.org/W4388544075","doi":"https://doi.org/10.1109/tpds.2023.3331372"},"language":"en","primary_location":{"id":"doi:10.1109/tpds.2023.3331372","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2023.3331372","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":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5011127712","display_name":"Yunqi Gao","orcid":"https://orcid.org/0009-0003-0599-8383"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Yunqi Gao","raw_affiliation_strings":["Zhejiang University, Hangzhou, China"],"affiliations":[{"raw_affiliation_string":"Zhejiang University, Hangzhou, China","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102834373","display_name":"Bing Hu","orcid":"https://orcid.org/0000-0002-0594-8998"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bing Hu","raw_affiliation_strings":["Zhejiang University, Hangzhou, China"],"affiliations":[{"raw_affiliation_string":"Zhejiang University, Hangzhou, China","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028862760","display_name":"Mahdi Boloursaz Mashhadi","orcid":"https://orcid.org/0000-0001-9948-9165"},"institutions":[{"id":"https://openalex.org/I28290843","display_name":"University of Surrey","ror":"https://ror.org/00ks66431","country_code":"GB","type":"education","lineage":["https://openalex.org/I28290843"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Mahdi Boloursaz Mashhadi","raw_affiliation_strings":["5GIC & 6GIC, Institute for Communication Systems (ICS), University of Surrey, Guildford, U.K"],"affiliations":[{"raw_affiliation_string":"5GIC & 6GIC, Institute for Communication Systems (ICS), University of Surrey, Guildford, U.K","institution_ids":["https://openalex.org/I28290843"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020103599","display_name":"A-Long Jin","orcid":"https://orcid.org/0000-0002-9680-2811"},"institutions":[{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"A-Long Jin","raw_affiliation_strings":["University of Hong Kong, Pokfulam, Hong Kong"],"affiliations":[{"raw_affiliation_string":"University of Hong Kong, Pokfulam, Hong Kong","institution_ids":["https://openalex.org/I889458895"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5002566516","display_name":"Pei Xiao","orcid":"https://orcid.org/0000-0002-7886-5878"},"institutions":[{"id":"https://openalex.org/I28290843","display_name":"University of Surrey","ror":"https://ror.org/00ks66431","country_code":"GB","type":"education","lineage":["https://openalex.org/I28290843"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Pei Xiao","raw_affiliation_strings":["5GIC & 6GIC, Institute for Communication Systems (ICS), University of Surrey, Guildford, U.K"],"affiliations":[{"raw_affiliation_string":"5GIC & 6GIC, Institute for Communication Systems (ICS), University of Surrey, Guildford, U.K","institution_ids":["https://openalex.org/I28290843"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5078198240","display_name":"Chunming Wu","orcid":"https://orcid.org/0000-0001-7958-9687"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chunming Wu","raw_affiliation_strings":["Zhejiang University, Hangzhou, China"],"affiliations":[{"raw_affiliation_string":"Zhejiang University, Hangzhou, China","institution_ids":["https://openalex.org/I76130692"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5011127712"],"corresponding_institution_ids":["https://openalex.org/I76130692"],"apc_list":null,"apc_paid":null,"fwci":1.3207,"has_fulltext":false,"cited_by_count":11,"citation_normalized_percentile":{"value":0.83218034,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":99},"biblio":{"volume":"35","issue":"1","first_page":"123","last_page":"139"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9987000226974487,"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":0.9987000226974487,"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/T12303","display_name":"Tensor decomposition and applications","score":0.9977999925613403,"subfield":{"id":"https://openalex.org/subfields/2605","display_name":"Computational Mathematics"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.996399998664856,"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":0.889446496963501},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.7312447428703308},{"id":"https://openalex.org/keywords/byte","display_name":"Byte","score":0.6917054057121277},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.6057835817337036},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5884861946105957},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.5660686492919922},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.5174096822738647},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.43195950984954834},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.38917264342308044},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.38882672786712646},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.3218406140804291},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.11639872193336487}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.889446496963501},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.7312447428703308},{"id":"https://openalex.org/C43364308","wikidata":"https://www.wikidata.org/wiki/Q8799","display_name":"Byte","level":2,"score":0.6917054057121277},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.6057835817337036},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5884861946105957},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5660686492919922},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.5174096822738647},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.43195950984954834},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.38917264342308044},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.38882672786712646},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3218406140804291},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.11639872193336487},{"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/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.0},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/tpds.2023.3331372","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2023.3331372","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:alma.44SUR_INST:11195490990002346","is_oa":false,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4210197018","display_name":"View","issn_l":"2688-268X","issn":["2688-268X","2688-3988"],"is_oa":false,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320595","host_organization_name":"Wiley","host_organization_lineage":["https://openalex.org/P4310320595"],"host_organization_lineage_names":["Wiley"],"type":"journal"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":""}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":75,"referenced_works":["https://openalex.org/W1686810756","https://openalex.org/W2057332538","https://openalex.org/W2060393849","https://openalex.org/W2097117768","https://openalex.org/W2108598243","https://openalex.org/W2113459411","https://openalex.org/W2132737349","https://openalex.org/W2168231600","https://openalex.org/W2183341477","https://openalex.org/W2186615578","https://openalex.org/W2194775991","https://openalex.org/W2498764059","https://openalex.org/W2604783387","https://openalex.org/W2617766261","https://openalex.org/W2622263826","https://openalex.org/W2745734086","https://openalex.org/W2769644379","https://openalex.org/W2774000609","https://openalex.org/W2896457183","https://openalex.org/W2920397365","https://openalex.org/W2953384591","https://openalex.org/W2962758826","https://openalex.org/W2963163009","https://openalex.org/W2963786636","https://openalex.org/W2964324519","https://openalex.org/W2964412985","https://openalex.org/W2966527647","https://openalex.org/W2968108410","https://openalex.org/W2975712713","https://openalex.org/W2982464076","https://openalex.org/W2982475424","https://openalex.org/W2994894455","https://openalex.org/W3047002910","https://openalex.org/W3047537431","https://openalex.org/W3101036738","https://openalex.org/W3106914464","https://openalex.org/W3118608800","https://openalex.org/W3124352525","https://openalex.org/W3174954936","https://openalex.org/W3176364684","https://openalex.org/W3177554119","https://openalex.org/W3193250980","https://openalex.org/W3203336515","https://openalex.org/W4206504654","https://openalex.org/W4283207077","https://openalex.org/W4283217358","https://openalex.org/W4288093768","https://openalex.org/W4288357791","https://openalex.org/W4295312788","https://openalex.org/W4301239768","https://openalex.org/W4312138951","https://openalex.org/W4313417913","https://openalex.org/W4320713383","https://openalex.org/W4385367428","https://openalex.org/W6637373629","https://openalex.org/W6676984168","https://openalex.org/W6679393576","https://openalex.org/W6684859321","https://openalex.org/W6686509673","https://openalex.org/W6713134421","https://openalex.org/W6736413256","https://openalex.org/W6738460352","https://openalex.org/W6739622702","https://openalex.org/W6739693220","https://openalex.org/W6746200960","https://openalex.org/W6746839373","https://openalex.org/W6748645090","https://openalex.org/W6754818105","https://openalex.org/W6755207826","https://openalex.org/W6762211661","https://openalex.org/W6766978945","https://openalex.org/W6769475105","https://openalex.org/W6785294077","https://openalex.org/W6787972765","https://openalex.org/W6797162769"],"related_works":["https://openalex.org/W2058965144","https://openalex.org/W2164382479","https://openalex.org/W98480971","https://openalex.org/W2150291671","https://openalex.org/W2891987081","https://openalex.org/W2011430815","https://openalex.org/W2499279132","https://openalex.org/W4321606653","https://openalex.org/W2149404148","https://openalex.org/W1981130628"],"abstract_inverted_index":{"The":[0],"communication":[1,13,23,55,95,129,139,226],"bottleneck":[2],"severely":[3],"constrains":[4],"the":[5,41,54,59,69,77,81,111,116,125,135,146,150,156,165,205,224],"scalability":[6],"of":[7,44,119,127,137,200,209],"distributed":[8,16],"deep":[9],"learning,":[10],"and":[11,22,35,130,133,140,153,187,193],"efficient":[12,34,94],"scheduling":[14,60],"accelerates":[15],"DNN":[17],"training":[18,112],"by":[19,121],"overlapping":[20],"computation":[21],"tasks.":[24],"However,":[25],"existing":[26],"approaches":[27],"based":[28],"on":[29,168,173],"tensor":[30,45,65,103],"partitioning":[31],"are":[32],"not":[33,51,86],"suffer":[36],"from":[37],"two":[38,122,174],"challenges:":[39],"1)":[40,124],"fixed":[42],"number":[43],"blocks":[46,66,104],"transferred":[47],"in":[48,76,105],"parallel":[49],"can":[50,72,100,182,216],"necessarily":[52],"minimize":[53,110],"overheads;":[56],"2)":[57,134],"although":[58],"order":[61,108],"that":[62,180,214],"preferentially":[63],"transmits":[64],"close":[67],"to":[68,109,185,191,203,219,223],"input":[70],"layer":[71],"start":[73],"forward":[74,141],"propagation":[75],"next":[78],"iteration":[79],"earlier,":[80],"shortest":[82],"per-iteration":[83],"time":[84],"is":[85],"obtained.":[87],"In":[88],"this":[89],"paper,":[90],"we":[91],"propose":[92],"an":[93],"framework":[96],"called":[97],"US-Byte.":[98,210],"It":[99],"schedule":[101],"unequal-sized":[102],"a":[106,160],"near-optimal":[107],"time.":[113],"We":[114,143,163,196],"build":[115],"mathematical":[117],"model":[118],"US-Byte":[120,166,181,215],"phases:":[123],"overlap":[126,136],"gradient":[128,138],"backward":[131],"propagation,":[132],"propagation.":[142],"theoretically":[144],"derive":[145],"optimal":[147],"solution":[148],"for":[149],"second":[151],"phase":[152,158],"efficiently":[154],"solve":[155],"first":[157],"with":[159],"low-complexity":[161],"algorithm.":[162],"implement":[164],"architecture":[167],"PyTorch":[169],"framework.":[170,227],"Extensive":[171],"experiments":[172],"different":[175],"8-node":[176],"GPU":[177],"clusters":[178],"demonstrate":[179],"achieve":[183,217],"up":[184,218],"1.26x":[186],"1.56x":[188],"speedup":[189,221],"compared":[190,222],"ByteScheduler":[192],"WFBP,":[194],"respectively.":[195],"further":[197],"exploit":[198],"simulations":[199],"128":[201],"GPUs":[202],"verify":[204],"potential":[206],"scaling":[207],"performance":[208],"Simulation":[211],"results":[212],"show":[213],"1.69x":[220],"state-of-the-art":[225]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":4}],"updated_date":"2026-04-02T15:55:50.835912","created_date":"2025-10-10T00:00:00"}