{"id":"https://openalex.org/W7162815760","doi":"https://doi.org/10.1145/3745756.3809214","title":"FBLayout: Optimizing Memory Layout for Efficient LLM Finetuning on Mobile GPUs","display_name":"FBLayout: Optimizing Memory Layout for Efficient LLM Finetuning on Mobile GPUs","publication_year":2026,"publication_date":"2026-05-29","ids":{"openalex":"https://openalex.org/W7162815760","doi":"https://doi.org/10.1145/3745756.3809214"},"language":null,"primary_location":{"id":"doi:10.1145/3745756.3809214","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3745756.3809214","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 24th Annual International Conference on Mobile Systems, Applications and Services","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2607.21624","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5053448652","display_name":"Kahou Tam","orcid":"https://orcid.org/0000-0001-5816-6837"},"institutions":[{"id":"https://openalex.org/I204512498","display_name":"University of Macau","ror":"https://ror.org/01r4q9n85","country_code":"MO","type":"education","lineage":["https://openalex.org/I204512498"]}],"countries":["MO"],"is_corresponding":false,"raw_author_name":"Kahou Tam","raw_affiliation_strings":["State Key Laboratory of IoTSC, University of Macau, Macau, Macao"],"raw_orcid":"https://orcid.org/0000-0001-5816-6837","affiliations":[{"raw_affiliation_string":"State Key Laboratory of IoTSC, University of Macau, Macau, Macao","institution_ids":["https://openalex.org/I204512498"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128745653","display_name":"Wei Niu","orcid":null},"institutions":[{"id":"https://openalex.org/I165733156","display_name":"University of Georgia","ror":"https://ror.org/00te3t702","country_code":"US","type":"education","lineage":["https://openalex.org/I165733156"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wei Niu","raw_affiliation_strings":["University of Georgia, Athens, GA, USA"],"raw_orcid":"https://orcid.org/0000-0002-2697-7042","affiliations":[{"raw_affiliation_string":"University of Georgia, Athens, GA, USA","institution_ids":["https://openalex.org/I165733156"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137389173","display_name":"Yu Bao","orcid":"https://orcid.org/0009-0009-3744-0580"},"institutions":[{"id":"https://openalex.org/I204512498","display_name":"University of Macau","ror":"https://ror.org/01r4q9n85","country_code":"MO","type":"education","lineage":["https://openalex.org/I204512498"]}],"countries":["MO"],"is_corresponding":false,"raw_author_name":"Yu Bao","raw_affiliation_strings":["State Key Laboratory of IoTSC, University of Macau, Macau, Macao"],"raw_orcid":"https://orcid.org/0009-0009-3744-0580","affiliations":[{"raw_affiliation_string":"State Key Laboratory of IoTSC, University of Macau, Macau, Macao","institution_ids":["https://openalex.org/I204512498"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137381217","display_name":"Xiaomin Ouyang","orcid":"https://orcid.org/0000-0003-0710-0963"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Xiaomin Ouyang","raw_affiliation_strings":["Hong Kong University of Science and Technology, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0003-0710-0963","affiliations":[{"raw_affiliation_string":"Hong Kong University of Science and Technology, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137332138","display_name":"ChengZhong Xu","orcid":"https://orcid.org/0000-0001-9480-0356"},"institutions":[{"id":"https://openalex.org/I204512498","display_name":"University of Macau","ror":"https://ror.org/01r4q9n85","country_code":"MO","type":"education","lineage":["https://openalex.org/I204512498"]}],"countries":["MO"],"is_corresponding":false,"raw_author_name":"ChengZhong Xu","raw_affiliation_strings":["State Key Laboratory of IoTSC, University of Macau, Macau, Macao"],"raw_orcid":"https://orcid.org/0000-0001-9480-0356","affiliations":[{"raw_affiliation_string":"State Key Laboratory of IoTSC, University of Macau, Macau, Macao","institution_ids":["https://openalex.org/I204512498"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5032899353","display_name":"L Li","orcid":null},"institutions":[{"id":"https://openalex.org/I204512498","display_name":"University of Macau","ror":"https://ror.org/01r4q9n85","country_code":"MO","type":"education","lineage":["https://openalex.org/I204512498"]}],"countries":["MO"],"is_corresponding":false,"raw_author_name":"Li Li","raw_affiliation_strings":["State Key Laboratory of IoTSC, University of Macau, Macau, Macao"],"raw_orcid":"https://orcid.org/0000-0002-2044-8289","affiliations":[{"raw_affiliation_string":"State Key Laboratory of IoTSC, University of Macau, Macau, Macao","institution_ids":["https://openalex.org/I204512498"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.52113309,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"432","last_page":"447"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.753000020980835,"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":0.753000020980835,"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/T12326","display_name":"Network Packet Processing and Optimization","score":0.033799998462200165,"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/T11181","display_name":"Advanced Data Storage Technologies","score":0.021199999377131462,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/mobile-device","display_name":"Mobile device","score":0.3560999929904938},{"id":"https://openalex.org/keywords/memory-management","display_name":"Memory management","score":0.3330000042915344},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2924000024795532},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.2757999897003174}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7556999921798706},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4596000015735626},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3977999985218048},{"id":"https://openalex.org/C186967261","wikidata":"https://www.wikidata.org/wiki/Q5082128","display_name":"Mobile device","level":2,"score":0.3560999929904938},{"id":"https://openalex.org/C176649486","wikidata":"https://www.wikidata.org/wiki/Q2308807","display_name":"Memory management","level":3,"score":0.3330000042915344},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.32120001316070557},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2924000024795532},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.2757999897003174},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.25760000944137573}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3745756.3809214","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3745756.3809214","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 24th Annual International Conference on Mobile Systems, Applications and Services","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2607.21624","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2607.21624","pdf_url":"https://arxiv.org/pdf/2607.21624","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2607.21624","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2607.21624","pdf_url":"https://arxiv.org/pdf/2607.21624","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[{"score":0.7587751746177673,"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy"}],"awards":[{"id":"https://openalex.org/G180185047","display_name":null,"funder_award_id":"GRG2024","funder_id":"https://openalex.org/F4320322841","funder_display_name":"Universidade de Macau"}],"funders":[{"id":"https://openalex.org/F4320311133","display_name":"United Mitochondrial Disease Foundation","ror":"https://ror.org/0528q0t18"},{"id":"https://openalex.org/F4320322841","display_name":"Universidade de Macau","ror":"https://ror.org/01r4q9n85"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7162815760.pdf","grobid_xml":"https://content.openalex.org/works/W7162815760.grobid-xml"},"referenced_works_count":15,"referenced_works":["https://openalex.org/W2194775991","https://openalex.org/W2963163009","https://openalex.org/W3174529902","https://openalex.org/W4282960063","https://openalex.org/W4306178406","https://openalex.org/W4312933868","https://openalex.org/W4380925616","https://openalex.org/W4387064014","https://openalex.org/W4387835442","https://openalex.org/W4394994587","https://openalex.org/W4395065134","https://openalex.org/W4399534536","https://openalex.org/W4400606525","https://openalex.org/W4402670099","https://openalex.org/W4402774099"],"related_works":[],"abstract_inverted_index":{"Transformer-based":[0],"models":[1,16,134],"have":[2],"enabled":[3],"unprecedented":[4],"capabilities":[5],"across":[6,108,135],"language,":[7],"vision,":[8],"and":[9,35,54,63,120,142,155,162],"multimodal":[10],"tasks.":[11],"On-device":[12],"fine-tuning":[13],"of":[14],"transformer":[15,133],"offers":[17],"a":[18,87,101],"privacy-preserving":[19],"path":[20],"to":[21,31,59,115,125],"personalized":[22],"AI,":[23],"yet":[24],"remains":[25],"inefficient":[26],"on":[27,72,131],"mobile":[28,45,95,137],"GPUs":[29],"due":[30],"severe":[32],"memory":[33,61,164],"constraints":[34],"frequent":[36],"layout":[37,74,104,123],"transformations":[38],"in":[39],"attention":[40],"mechanism":[41],"during":[42,67],"training.":[43],"Existing":[44],"training":[46],"frameworks":[47],"either":[48],"use":[49],"unified":[50,102],"layouts":[51,128],"for":[52,105],"forward":[53],"backward":[55],"passes":[56],"--":[57,69],"leading":[58],"fragmented":[60],"access":[62],"poor":[64],"GPU":[65,96],"utilization":[66],"backpropagation":[68],"or":[70],"rely":[71],"explicit":[73],"conversions,":[75],"which":[76],"introduce":[77],"significant":[78],"transformation":[79,114],"overhead.":[80],"To":[81],"overcome":[82],"this,":[83],"we":[84],"propose":[85],"FBLayout,":[86],"layout-aware":[88],"framework":[89],"that":[90,147],"co-designs":[91],"tensor":[92],"organization":[93],"with":[94],"platforms.":[97],"FBLayout":[98,148],"introduces:":[99],"(1)":[100],"R-Tile":[103],"multi-dimensional":[106],"reductions":[107],"forward/backward":[109],"passes;":[110],"(2)":[111],"tile-based":[112],"index":[113],"eliminate":[116],"physical":[117],"data":[118],"movement;":[119],"(3)":[121],"activation-guided":[122],"selection":[124],"propagate":[126],"efficient":[127],"globally.":[129],"Evaluations":[130],"seven":[132],"different":[136],"phones":[138],"(including":[139],"ARM":[140],"Mali":[141],"Qualcomm":[143],"Adreno":[144],"GPUs)":[145],"show":[146],"achieves":[149],"2.2-5.7x":[150],"speedup":[151],"over":[152],"MNN,":[153],"TFLite,":[154],"TVM,":[156],"while":[157],"significantly":[158],"improving":[159],"cache":[160],"efficiency":[161],"reducing":[163],"footprint,":[165],"enabling":[166],"practical":[167],"on-device":[168],"large":[169],"model":[170],"fine-tuning.":[171]},"counts_by_year":[],"updated_date":"2026-08-08T01:25:22.217667","created_date":"2026-05-30T00:00:00"}
