{"id":"https://openalex.org/W4416649506","doi":"https://doi.org/10.1145/3768985","title":"Harmonics: Scalable Collective Scheduling in Multi-Tenant GPU Clusters","display_name":"Harmonics: Scalable Collective Scheduling in Multi-Tenant GPU Clusters","publication_year":2025,"publication_date":"2025-11-24","ids":{"openalex":"https://openalex.org/W4416649506","doi":"https://doi.org/10.1145/3768985"},"language":"en","primary_location":{"id":"doi:10.1145/3768985","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3768985","pdf_url":null,"source":{"id":"https://openalex.org/S4387290834","display_name":"Proceedings of the ACM on Networking","issn_l":"2834-5509","issn":["2834-5509"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"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 ACM on Networking","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1145/3768985","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5076940594","display_name":"Hossein Shafieirad","orcid":"https://orcid.org/0000-0002-0633-404X"},"institutions":[{"id":"https://openalex.org/I4210115038","display_name":"Huawei Technologies (Canada)","ror":"https://ror.org/026venb53","country_code":"CA","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210115038"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Hossein Shafieirad","raw_affiliation_strings":["Huawei Technologies Canada, Toronto, ON, Canada"],"raw_orcid":"https://orcid.org/0000-0002-0633-404X","affiliations":[{"raw_affiliation_string":"Huawei Technologies Canada, Toronto, ON, Canada","institution_ids":["https://openalex.org/I4210115038"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5073778515","display_name":"Amir Shani","orcid":"https://orcid.org/0000-0002-4614-5315"},"institutions":[{"id":"https://openalex.org/I4210115038","display_name":"Huawei Technologies (Canada)","ror":"https://ror.org/026venb53","country_code":"CA","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210115038"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Amir Shani","raw_affiliation_strings":["Huawei Technologies Canada, Toronto, ON, Canada"],"raw_orcid":"https://orcid.org/0009-0006-7833-8652","affiliations":[{"raw_affiliation_string":"Huawei Technologies Canada, Toronto, ON, Canada","institution_ids":["https://openalex.org/I4210115038"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044077636","display_name":"Manaf Bin-Yahya","orcid":"https://orcid.org/0000-0003-3688-1594"},"institutions":[{"id":"https://openalex.org/I4210115038","display_name":"Huawei Technologies (Canada)","ror":"https://ror.org/026venb53","country_code":"CA","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210115038"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Manaf Bin-Yahya","raw_affiliation_strings":["Huawei Technologies Canada, Toronto, ON, Canada"],"raw_orcid":"https://orcid.org/0000-0003-3688-1594","affiliations":[{"raw_affiliation_string":"Huawei Technologies Canada, Toronto, ON, Canada","institution_ids":["https://openalex.org/I4210115038"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120427430","display_name":"Seyed Hossein Mortazavi","orcid":"https://orcid.org/0000-0002-8058-4127"},"institutions":[{"id":"https://openalex.org/I4210115038","display_name":"Huawei Technologies (Canada)","ror":"https://ror.org/026venb53","country_code":"CA","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210115038"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Seyed Hossein Mortazavi","raw_affiliation_strings":["Huawei Technologies Canada, Toronto, ON, Canada"],"raw_orcid":"https://orcid.org/0000-0002-8058-4127","affiliations":[{"raw_affiliation_string":"Huawei Technologies Canada, Toronto, ON, Canada","institution_ids":["https://openalex.org/I4210115038"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079348889","display_name":"Geng Li","orcid":"https://orcid.org/0000-0002-5467-548X"},"institutions":[{"id":"https://openalex.org/I4210136246","display_name":"China Telecom (China)","ror":"https://ror.org/03jgnzt20","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210136246"]},{"id":"https://openalex.org/I4387153335","display_name":"China Telecom","ror":"https://ror.org/05p67dv18","country_code":null,"type":"company","lineage":["https://openalex.org/I4387153335"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Geng Li","raw_affiliation_strings":["China Telecom Cloud Computing Research Institute, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-5467-548X","affiliations":[{"raw_affiliation_string":"China Telecom Cloud Computing Research Institute, Beijing, China","institution_ids":["https://openalex.org/I4210136246","https://openalex.org/I4387153335"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076874267","display_name":"Xinle Du","orcid":"https://orcid.org/0000-0001-8918-9580"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xinle Du","raw_affiliation_strings":["Huawei Technologies China, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-8918-9580","affiliations":[{"raw_affiliation_string":"Huawei Technologies China, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Tao Su","orcid":"https://orcid.org/0009-0007-3919-3734"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tao Su","raw_affiliation_strings":["Huawei Technologies China, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0007-3919-3734","affiliations":[{"raw_affiliation_string":"Huawei Technologies China, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101425222","display_name":"Wei Wang","orcid":"https://orcid.org/0009-0002-1231-5539"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wei Wang","raw_affiliation_strings":["Huawei Technologies China, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-1231-5539","affiliations":[{"raw_affiliation_string":"Huawei Technologies China, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086356074","display_name":"Jingbin Zhou","orcid":"https://orcid.org/0009-0005-4781-1876"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jingbin Zhou","raw_affiliation_strings":["Huawei Technologies China, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0005-4781-1876","affiliations":[{"raw_affiliation_string":"Huawei Technologies China, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5081502954","display_name":"Majid Ghaderi","orcid":"https://orcid.org/0000-0002-3783-4346"},"institutions":[{"id":"https://openalex.org/I168635309","display_name":"University of Calgary","ror":"https://ror.org/03yjb2x39","country_code":"CA","type":"education","lineage":["https://openalex.org/I168635309"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Majid Ghaderi","raw_affiliation_strings":["University of Calgary, Calgary, Canada"],"raw_orcid":"https://orcid.org/0000-0002-3783-4346","affiliations":[{"raw_affiliation_string":"University of Calgary, Calgary, Canada","institution_ids":["https://openalex.org/I168635309"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":10,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.45458538,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"3","issue":"CoNEXT4","first_page":"1","last_page":"20"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T14347","display_name":"Big Data and Digital Economy","score":0.25440001487731934,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T14347","display_name":"Big Data and Digital Economy","score":0.25440001487731934,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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":0.2531999945640564,"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/T10101","display_name":"Cloud Computing and Resource Management","score":0.20990000665187836,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/scalability","display_name":"Scalability","score":0.6930999755859375},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.6769999861717224},{"id":"https://openalex.org/keywords/fair-share-scheduling","display_name":"Fair-share scheduling","score":0.46619999408721924},{"id":"https://openalex.org/keywords/dynamic-priority-scheduling","display_name":"Dynamic priority scheduling","score":0.45320001244544983},{"id":"https://openalex.org/keywords/two-level-scheduling","display_name":"Two-level scheduling","score":0.4255000054836273},{"id":"https://openalex.org/keywords/cloud-computing","display_name":"Cloud computing","score":0.4117000102996826},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.3968000113964081},{"id":"https://openalex.org/keywords/fixed-priority-pre-emptive-scheduling","display_name":"Fixed-priority pre-emptive scheduling","score":0.36880001425743103}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8574000000953674},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.7027999758720398},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6930999755859375},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.6769999861717224},{"id":"https://openalex.org/C31689143","wikidata":"https://www.wikidata.org/wiki/Q733809","display_name":"Fair-share scheduling","level":3,"score":0.46619999408721924},{"id":"https://openalex.org/C107568181","wikidata":"https://www.wikidata.org/wiki/Q5319000","display_name":"Dynamic priority scheduling","level":3,"score":0.45320001244544983},{"id":"https://openalex.org/C119948110","wikidata":"https://www.wikidata.org/wiki/Q7858726","display_name":"Two-level scheduling","level":4,"score":0.4255000054836273},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.4117000102996826},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3968000113964081},{"id":"https://openalex.org/C122141398","wikidata":"https://www.wikidata.org/wiki/Q5456330","display_name":"Fixed-priority pre-emptive scheduling","level":5,"score":0.36880001425743103},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.358599990606308},{"id":"https://openalex.org/C201515029","wikidata":"https://www.wikidata.org/wiki/Q5520966","display_name":"Gang scheduling","level":5,"score":0.35839998722076416},{"id":"https://openalex.org/C2781335571","wikidata":"https://www.wikidata.org/wiki/Q2633544","display_name":"GPU cluster","level":3,"score":0.3441999852657318},{"id":"https://openalex.org/C127456818","wikidata":"https://www.wikidata.org/wiki/Q238879","display_name":"Rate-monotonic scheduling","level":4,"score":0.3343999981880188},{"id":"https://openalex.org/C175893541","wikidata":"https://www.wikidata.org/wiki/Q1196582","display_name":"Round-robin scheduling","level":4,"score":0.3273000121116638},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.3248000144958496},{"id":"https://openalex.org/C55416958","wikidata":"https://www.wikidata.org/wiki/Q6206757","display_name":"Job shop scheduling","level":3,"score":0.31850001215934753},{"id":"https://openalex.org/C2989134064","wikidata":"https://www.wikidata.org/wiki/Q288510","display_name":"Execution time","level":2,"score":0.2939000129699707},{"id":"https://openalex.org/C188414643","wikidata":"https://www.wikidata.org/wiki/Q3001183","display_name":"Harmonics","level":3,"score":0.26649999618530273},{"id":"https://openalex.org/C32310161","wikidata":"https://www.wikidata.org/wiki/Q666365","display_name":"Earliest deadline first scheduling","level":5,"score":0.2565999925136566}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3768985","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3768985","pdf_url":null,"source":{"id":"https://openalex.org/S4387290834","display_name":"Proceedings of the ACM on Networking","issn_l":"2834-5509","issn":["2834-5509"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"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 ACM on Networking","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1145/3768985","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3768985","pdf_url":null,"source":{"id":"https://openalex.org/S4387290834","display_name":"Proceedings of the ACM on Networking","issn_l":"2834-5509","issn":["2834-5509"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"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 ACM on Networking","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":9,"referenced_works":["https://openalex.org/W2774772673","https://openalex.org/W4235732943","https://openalex.org/W4255119735","https://openalex.org/W4281931618","https://openalex.org/W4386768656","https://openalex.org/W4391136507","https://openalex.org/W4399426455","https://openalex.org/W4401176822","https://openalex.org/W4410708363"],"related_works":[],"abstract_inverted_index":{"Distributed":[0],"machine":[1],"learning":[2],"(DML),":[3],"such":[4,133],"as":[5,134],"large":[6],"language":[7],"model":[8],"(LLM)":[9],"training,":[10],"has":[11],"become":[12],"one":[13],"of":[14,50,61],"the":[15,48,59],"most":[16],"critical":[17],"services":[18],"in":[19,175],"multi-tenant":[20],"cloud":[21],"computing.":[22],"However,":[23],"communication":[24,108,151],"contention":[25,109],"among":[26],"concurrent":[27],"DML":[28],"jobs":[29],"significantly":[30],"degrades":[31],"overall":[32],"GPU":[33,127],"utilization,":[34],"leading":[35],"to":[36,92,130,148,155,167],"inefficient":[37],"training":[38,144,171],"cluster":[39],"performance.":[40,63,80],"Existing":[41],"approaches":[42],"either":[43],"achieve":[44],"high":[45],"performance":[46],"at":[47,58],"cost":[49],"long":[51],"scheduling":[52,56,70,77,163],"runtime":[53],"or":[54],"reduce":[55],"time":[57,145,152,164,172],"expense":[60],"poor":[62],"We":[64,114],"present":[65],"Harmonics,":[66],"a":[67,74,88,126],"novel":[68],"two-tier":[69],"framework":[71],"that":[72,160],"strikes":[73],"balance":[75],"between":[76,84],"latency":[78],"and":[79,87,95,110,116,140,150],"It":[81],"coordinates":[82],"decisions":[83],"Local":[85],"Schedulers":[86],"lightweight":[89],"Global":[90],"Coordinator":[91],"enable":[93],"scalable":[94],"adaptive":[96],"scheduling.":[97],"By":[98],"combining":[99],"rack-level":[100],"epoch-based":[101],"optimization":[102],"with":[103],"global":[104],"coordination,":[105],"Harmonics":[106,118,142],"alleviates":[107],"improves":[111],"resource":[112],"efficiency.":[113],"implement":[115],"evaluate":[117],"on":[119,125],"real":[120],"distributed":[121],"ML":[122],"workloads":[123],"running":[124],"testbed.":[128],"Compared":[129],"state-of-the-art":[131],"methods":[132],"fair":[135],"sharing,":[136],"optimal":[137],"scheduling,":[138],"Crux,":[139],"Cassini,":[141],"reduces":[143,162],"by":[146,153,165,173],"up":[147,154,166],"33%":[149],"48%.":[156],"Large-scale":[157],"simulations":[158],"show":[159],"it":[161],"91\u00d7":[168],"while":[169],"improving":[170],"26%":[174],"large-cluster":[176],"settings.":[177]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-11-25T00:00:00"}