{"id":"https://openalex.org/W7220401992","doi":"https://doi.org/10.1145/3845814.3850072","title":"EdgeAgent: Orchestrating On-Device LLM inference for End-User Multi-Agent Systems on CPU-GPU Unified Memory Architectures","display_name":"EdgeAgent: Orchestrating On-Device LLM inference for End-User Multi-Agent Systems on CPU-GPU Unified Memory Architectures","publication_year":2026,"publication_date":"2026-11-04","ids":{"openalex":"https://openalex.org/W7220401992","doi":"https://doi.org/10.1145/3845814.3850072"},"language":"en","primary_location":{"id":"doi:10.1145/3845814.3850072","doi":"https://doi.org/10.1145/3845814.3850072","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3845814.3850072","pdf_url":null,"source":{"id":"https://openalex.org/S7407086669","display_name":"International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 32nd ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 1","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1145/3845814.3850072","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139792797","display_name":"Yuhai Long","orcid":"https://orcid.org/0009-0007-6736-6372","observed_orcids":["https://orcid.org/0009-0007-6736-6372"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuhai Long","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0009-0007-6736-6372","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5144564252","display_name":"Yuanxin Wei","orcid":"https://orcid.org/0009-0000-6551-9274","observed_orcids":["https://orcid.org/0009-0000-6551-9274"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuanxin Wei","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0009-0000-6551-9274","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5152538202","display_name":"Kai Wu","orcid":"https://orcid.org/0009-0002-3749-6268","observed_orcids":["https://orcid.org/0009-0002-3749-6268"]},"institutions":[{"id":"https://openalex.org/I180662265","display_name":"China Mobile (China)","ror":"https://ror.org/05gftfe97","country_code":"CN","type":"company","lineage":["https://openalex.org/I180662265"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kai Wu","raw_affiliation_strings":["China Mobile Internet Company Ltd., Guangzhou, China"],"raw_orcid":"https://orcid.org/0009-0002-3749-6268","affiliations":[{"raw_affiliation_string":"China Mobile Internet Company Ltd., Guangzhou, China","institution_ids":["https://openalex.org/I180662265"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5157625786","display_name":"Jinhui Wei","orcid":"https://orcid.org/0000-0002-9850-8384","observed_orcids":["https://orcid.org/0000-0002-9850-8384"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jinhui Wei","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0002-9850-8384","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5157434792","display_name":"Dan Huang","orcid":"https://orcid.org/0000-0001-5582-1031","observed_orcids":["https://orcid.org/0000-0001-5582-1031"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Dan Huang","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-5582-1031","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5139378114","display_name":"Jiangsu Du","orcid":"https://orcid.org/0000-0003-4707-9492","observed_orcids":["https://orcid.org/0000-0003-4707-9492"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiangsu Du","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0003-4707-9492","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"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":0.0,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"283","last_page":"297"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9761000275611877,"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":0.9761000275611877,"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/T10273","display_name":"IoT and Edge/Fog Computing","score":0.006899999920278788,"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"}},{"id":"https://openalex.org/T10904","display_name":"Embedded Systems Design Techniques","score":0.0032999999821186066,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/multi-agent-llm-systems","display_name":"multi-agent LLM systems","score":0.8019999861717224},{"id":"https://openalex.org/keywords/unified-memory-architecture","display_name":"unified memory architecture","score":0.7900000214576721},{"id":"https://openalex.org/keywords/speculative-decoding","display_name":"speculative decoding","score":0.9649999737739563},{"id":"https://openalex.org/keywords/tensor-parallelism","display_name":"tensor parallelism","score":0.8579999804496765},{"id":"https://openalex.org/keywords/edge-deployment","display_name":"edge deployment","score":0.5249999761581421},{"id":"https://openalex.org/keywords/privacy-preserving-inference","display_name":"privacy-preserving inference","score":0.6880000233650208}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8522999882698059},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.7777000069618225},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.602400004863739},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5906000137329102},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.520799994468689},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4456000030040741},{"id":"https://openalex.org/C197640229","wikidata":"https://www.wikidata.org/wiki/Q2534066","display_name":"Predictability","level":2,"score":0.4189999997615814},{"id":"https://openalex.org/C193702766","wikidata":"https://www.wikidata.org/wiki/Q1414548","display_name":"Concurrency","level":2,"score":0.41690000891685486},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.38040000200271606},{"id":"https://openalex.org/C206952183","wikidata":"https://www.wikidata.org/wiki/Q1193100","display_name":"Preemption","level":2,"score":0.3594000041484833},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.35589998960494995},{"id":"https://openalex.org/C2984822820","wikidata":"https://www.wikidata.org/wiki/Q1123036","display_name":"Processor scheduling","level":3,"score":0.31949999928474426},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.3003000020980835},{"id":"https://openalex.org/C188045654","wikidata":"https://www.wikidata.org/wiki/Q17148339","display_name":"Memory bandwidth","level":2,"score":0.2919999957084656},{"id":"https://openalex.org/C176649486","wikidata":"https://www.wikidata.org/wiki/Q2308807","display_name":"Memory management","level":3,"score":0.28519999980926514},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.27889999747276306},{"id":"https://openalex.org/C147297375","wikidata":"https://www.wikidata.org/wiki/Q6674930","display_name":"Look-ahead","level":2,"score":0.27790001034736633},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.2775999903678894},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.2703999876976013},{"id":"https://openalex.org/C34388435","wikidata":"https://www.wikidata.org/wiki/Q2267362","display_name":"Bounded function","level":2,"score":0.26989999413490295},{"id":"https://openalex.org/C12186640","wikidata":"https://www.wikidata.org/wiki/Q6815743","display_name":"Memory model","level":3,"score":0.26899999380111694},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.26170000433921814},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.25870001316070557},{"id":"https://openalex.org/C68387754","wikidata":"https://www.wikidata.org/wiki/Q7271585","display_name":"Schedule","level":2,"score":0.25589999556541443},{"id":"https://openalex.org/C2776834041","wikidata":"https://www.wikidata.org/wiki/Q25346349","display_name":"Execution model","level":2,"score":0.2502000033855438}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3845814.3850072","doi":"https://doi.org/10.1145/3845814.3850072","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3845814.3850072","pdf_url":null,"source":{"id":"https://openalex.org/S7407086669","display_name":"International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 32nd ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 1","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2610.03394","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.03394","pdf_url":"https://arxiv.org/pdf/2610.03394","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,"listed_in":[],"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","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2610.03394","doi":"https://doi.org/10.48550/arxiv.2610.03394","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.03394","pdf_url":null,"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,"listed_in":[],"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","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.1145/3845814.3850072","doi":"https://doi.org/10.1145/3845814.3850072","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3845814.3850072","pdf_url":null,"source":{"id":"https://openalex.org/S7407086669","display_name":"International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 32nd ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 1","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G2142148654","display_name":null,"funder_award_id":"GuikeAB25069495","funder_id":"https://openalex.org/F4320336589","funder_display_name":"Guangxi Key Research and Development Program"},{"id":"https://openalex.org/G4022804212","display_name":null,"funder_award_id":"23xkjc016","funder_id":"https://openalex.org/F4320335787","funder_display_name":"Fundamental Research Funds for the Central Universities"},{"id":"https://openalex.org/G4098926025","display_name":null,"funder_award_id":"23xkjc016","funder_id":"https://openalex.org/F4320321160","funder_display_name":"Sun Yat-sen University"},{"id":"https://openalex.org/G63103377","display_name":null,"funder_award_id":"62402534","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320318558","display_name":"Peng Cheng Laboratory","ror":"https://ror.org/03qdqbt06"},{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320321160","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71"},{"id":"https://openalex.org/F4320335787","display_name":"Fundamental Research Funds for the Central Universities","ror":null},{"id":"https://openalex.org/F4320336589","display_name":"Guangxi Key Research and Development Program","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7220401992.pdf","grobid_xml":"https://content.openalex.org/works/W7220401992.grobid-xml"},"referenced_works_count":23,"referenced_works":["https://openalex.org/W3122286897","https://openalex.org/W3187255235","https://openalex.org/W3195140297","https://openalex.org/W4387321091","https://openalex.org/W4394998532","https://openalex.org/W4396815229","https://openalex.org/W4401306886","https://openalex.org/W4402670152","https://openalex.org/W4405717632","https://openalex.org/W4406163700","https://openalex.org/W4414281281","https://openalex.org/W4414736038","https://openalex.org/W4416146691","https://openalex.org/W4416524577","https://openalex.org/W4417403165","https://openalex.org/W7104000765","https://openalex.org/W7117112653","https://openalex.org/W7117558955","https://openalex.org/W7128776460","https://openalex.org/W7134839917","https://openalex.org/W7155545501","https://openalex.org/W7166642611","https://openalex.org/W7196954301"],"related_works":["https://openalex.org/W7117737180","https://openalex.org/W7168414276","https://openalex.org/W7118005808","https://openalex.org/W7214403271","https://openalex.org/W7164818176","https://openalex.org/W7168167998","https://openalex.org/W4416524577","https://openalex.org/W7161131589","https://openalex.org/W7117558955","https://openalex.org/W7161039234"],"abstract_inverted_index":{"Emerging":[0],"multi-agent":[1,38,85],"LLMs":[2],"demand":[3],"privacy-preserving":[4],"edge":[5,82],"deployment,":[6],"yet":[7],"current":[8],"inference":[9,77],"systems":[10],"struggle":[11],"with":[12],"these":[13],"collaborative":[14],"workflows.":[15],"Specifically,":[16],"the":[17,88,116,159,172,176],"memory-bound":[18],"decode":[19],"phase":[20],"causes":[21],"severe":[22],"bus":[23],"contention":[24],"on":[25,125,152],"unified":[26],"memory":[27,104],"architectures":[28],"(UMA),":[29],"paralyzing":[30],"naive":[31],"CPU-GPU":[32],"co-execution.":[33],"Furthermore,":[34],"speculative":[35,169],"decoding":[36],"in":[37,43],"workloads":[39],"faces":[40],"extreme":[41,185],"variance":[42],"drafting":[44],"difficulty,":[45],"alternating":[46],"between":[47],"complex":[48],"reasoning":[49],"and":[50,66,84,111],"predictable":[51],"structured":[52],"generation.":[53],"Compounded":[54],"by":[55],"frequent":[56],"tool-induced":[57],"stalls,":[58],"this":[59],"highly":[60],"fragmented":[61],"execution":[62,161],"severely":[63],"underutilizes":[64],"hardware":[65,144],"defeats":[67],"traditional":[68],"static":[69],"batching.":[70],"We":[71],"present":[72],"EdgeAgent":[73,178],",":[74],"a":[75,164,181],"cross-layer":[76],"system":[78,179],"explicitly":[79],"co-designed":[80],"for":[81],"UMA":[83],"workloads.":[86],"At":[87,115],"micro-architectural":[89],"level,":[90,118],"it":[91,119],"bypasses":[92],"rigid":[93],"graph-compiler":[94],"constraints":[95],"to":[96,106,129,180],"enable":[97],"zero-copy":[98],"UMA-aware":[99,160],"tensor":[100],"parallelism,":[101],"utilizing":[102],"asymmetric":[103],"layouts":[105],"fully":[107],"saturate":[108],"both":[109],"CPU":[110],"GPU":[112],"compute":[113],"units.":[114],"scheduling":[117,174],"dynamically":[120],"allocates":[121],"draft":[122],"budgets":[123],"based":[124],"real-time":[126],"sequence":[127],"predictability":[128],"bound":[130],"bandwidth":[131],"waste.":[132],"Concurrently,":[133],"an":[134,153],"asynchronous":[135],"suspend-and-yield":[136],"mechanism":[137],"actively":[138],"evicts":[139],"stalled":[140],"agents,":[141],"ensuring":[142],"continuous":[143],"saturation":[145],"during":[146],"unpredictable":[147],"tool":[148],"invocations.":[149],"Extensive":[150],"evaluations":[151],"Apple":[154],"M4":[155],"SoC":[156],"demonstrate":[157],"that":[158],"alone":[162],"contributes":[163],"1.29\u00d7":[165],"speedup":[166,183],"over":[167],"batched":[168],"decoding.":[170],"Adding":[171],"agent-aware":[173],"lifts":[175],"full":[177],"1.77\u00d7":[182],"under":[184],"tool-use":[186],"latencies.":[187]},"counts_by_year":[],"updated_date":"2026-11-10T11:39:43.021295","created_date":"2026-10-06T00:00:00"}
