{"id":"https://openalex.org/W7155494105","doi":"https://doi.org/10.48550/arxiv.2604.21275","title":"Optimizing High-Throughput Distributed Data Pipelines for Reproducible Deep Learning at Scale","display_name":"Optimizing High-Throughput Distributed Data Pipelines for Reproducible Deep Learning at Scale","publication_year":2026,"publication_date":"2026-04-23","ids":{"openalex":"https://openalex.org/W7155494105","doi":"https://doi.org/10.48550/arxiv.2604.21275"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.21275","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.21275","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.21275","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134482504","display_name":"Kashish Mittal","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mittal, Kashish","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134527164","display_name":"Di Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Di","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5026180984","display_name":"Roozbeh Ketabi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ketabi, Roozbeh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088698221","display_name":"Arushi Arora","orcid":"https://orcid.org/0000-0001-8913-3544"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Arora, Arushi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134477215","display_name":"Brendon Lapp","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lapp, Brendon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5092114915","display_name":"P. Zhang","orcid":"https://orcid.org/0000-0002-9177-6108"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Peng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.826200008392334,"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":0.826200008392334,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.025200000032782555,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T14347","display_name":"Big Data and Digital Economy","score":0.0203000009059906,"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/terabyte","display_name":"Terabyte","score":0.7856000065803528},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.7164999842643738},{"id":"https://openalex.org/keywords/pipeline-transport","display_name":"Pipeline transport","score":0.579200029373169},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.5534999966621399},{"id":"https://openalex.org/keywords/scale","display_name":"Scale (ratio)","score":0.4880000054836273},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.48649999499320984},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.4810999929904938},{"id":"https://openalex.org/keywords/big-data","display_name":"Big data","score":0.47920000553131104},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.44119998812675476},{"id":"https://openalex.org/keywords/data-modeling","display_name":"Data modeling","score":0.4212000072002411}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8051999807357788},{"id":"https://openalex.org/C199683683","wikidata":"https://www.wikidata.org/wiki/Q8799","display_name":"Terabyte","level":2,"score":0.7856000065803528},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.7164999842643738},{"id":"https://openalex.org/C175309249","wikidata":"https://www.wikidata.org/wiki/Q725864","display_name":"Pipeline transport","level":2,"score":0.579200029373169},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.5534999966621399},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.5060999989509583},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.4880000054836273},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.48649999499320984},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.4810999929904938},{"id":"https://openalex.org/C75684735","wikidata":"https://www.wikidata.org/wiki/Q858810","display_name":"Big data","level":2,"score":0.47920000553131104},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4666999876499176},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.44119998812675476},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.4212000072002411},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.383899986743927},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.38280001282691956},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3734000027179718},{"id":"https://openalex.org/C160403385","wikidata":"https://www.wikidata.org/wiki/Q220543","display_name":"Queue","level":2,"score":0.3700999915599823},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.34929999709129333},{"id":"https://openalex.org/C93996380","wikidata":"https://www.wikidata.org/wiki/Q44127","display_name":"Server","level":2,"score":0.3287000060081482},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.32600000500679016},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.322299987077713},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.31459999084472656},{"id":"https://openalex.org/C70061542","wikidata":"https://www.wikidata.org/wiki/Q989016","display_name":"Distributed database","level":2,"score":0.30790001153945923},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.30550000071525574},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.298799991607666},{"id":"https://openalex.org/C193415008","wikidata":"https://www.wikidata.org/wiki/Q639681","display_name":"Network architecture","level":2,"score":0.2948000133037567},{"id":"https://openalex.org/C557945733","wikidata":"https://www.wikidata.org/wiki/Q389772","display_name":"Data transmission","level":2,"score":0.2824999988079071},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.28049999475479126},{"id":"https://openalex.org/C138827492","wikidata":"https://www.wikidata.org/wiki/Q6661985","display_name":"Data processing","level":2,"score":0.2791999876499176},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2768000066280365},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.27480000257492065},{"id":"https://openalex.org/C47487241","wikidata":"https://www.wikidata.org/wiki/Q5227230","display_name":"Data access","level":2,"score":0.2669999897480011},{"id":"https://openalex.org/C158379750","wikidata":"https://www.wikidata.org/wiki/Q214111","display_name":"Network packet","level":2,"score":0.2669000029563904},{"id":"https://openalex.org/C2776397876","wikidata":"https://www.wikidata.org/wiki/Q1450531","display_name":"Cyberinfrastructure","level":2,"score":0.2662999927997589},{"id":"https://openalex.org/C75165309","wikidata":"https://www.wikidata.org/wiki/Q2258979","display_name":"Search engine indexing","level":2,"score":0.2628999948501587},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.26010000705718994},{"id":"https://openalex.org/C149810388","wikidata":"https://www.wikidata.org/wiki/Q5374873","display_name":"Emulation","level":2,"score":0.2587999999523163},{"id":"https://openalex.org/C2779280203","wikidata":"https://www.wikidata.org/wiki/Q17121211","display_name":"Small data","level":2,"score":0.25850000977516174},{"id":"https://openalex.org/C86111242","wikidata":"https://www.wikidata.org/wiki/Q859595","display_name":"Coprocessor","level":2,"score":0.2556999921798706}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.21275","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.21275","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.21275","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.21275","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/8","score":0.4085155129432678,"display_name":"Decent work and economic growth"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Training":[0],"massive-scale":[1],"deep":[2],"learning":[3],"models":[4],"on":[5],"datasets":[6],"spanning":[7],"tens":[8],"of":[9],"terabytes":[10],"presents":[11],"critical":[12],"challenges":[13],"in":[14,101],"hardware":[15],"utilization":[16,62,140],"and":[17,25,40,52,90,110,144,152],"training":[18,33,94,130],"reproducibility.":[19],"In":[20],"this":[21],"paper,":[22],"we":[23,47,71,97],"identify":[24],"resolve":[26],"profound":[27],"data-loading":[28],"bottlenecks":[29],"within":[30],"distributed":[31],"GPU":[32,61,139],"pipelines":[34],"using":[35],"the":[36],"Petastorm":[37],"data":[38,54,120],"loader":[39],"Apache":[41],"Parquet":[42],"datasets.":[43],"Through":[44],"systematic":[45],"profiling,":[46],"demonstrate":[48],"that":[49,76],"network":[50],"I/O":[51,89],"CPU-bound":[53],"transformations":[55,80],"(e.g.,":[56],"PyArrow":[57],"to":[58,63,135,141],"NumPy)":[59],"constrain":[60],"as":[64,66],"low":[65],"10-15%.":[67],"To":[68],"address":[69],"this,":[70],"propose":[72],"an":[73],"optimized":[74],"architecture":[75],"features":[77],"push-down":[78],"worker-level":[79],"coupled":[81],"with":[82],"local-disk":[83],"caching":[84],"via":[85],"Fanout-Cache,":[86],"minimizing":[87],"redundant":[88],"CPU":[91],"overhead":[92],"across":[93],"epochs.":[95],"Furthermore,":[96],"eliminate":[98],"race":[99],"conditions":[100],"multi-worker":[102],"shared":[103],"queues":[104],"by":[105],"implementing":[106],"dedicated":[107],"round-robin":[108],"ventilator":[109],"result":[111],"queues,":[112],"alongside":[113],"modernized":[114],"RNG":[115],"handling,":[116],"achieving":[117],"strict":[118],"deterministic":[119],"loading.":[121],"Our":[122],"optimizations":[123],"yield":[124],"a":[125],"6x":[126],"speedup,":[127],"reducing":[128,146],"end-to-end":[129],"time":[131],"from":[132],"22":[133],"hours":[134],"3":[136],"hours,":[137],"increasing":[138],"over":[142],"60%,":[143],"drastically":[145],"run-to-run":[147],"variance,":[148],"enabling":[149],"robust,":[150],"high-throughput,":[151],"reproducible":[153],"large-scale":[154],"model":[155],"training.":[156]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-25T00:00:00"}