{"id":"https://openalex.org/W4399534502","doi":"https://doi.org/10.1145/3662006.3662066","title":"WiP: Efficient LLM Prefilling with Mobile NPU","display_name":"WiP: Efficient LLM Prefilling with Mobile NPU","publication_year":2024,"publication_date":"2024-06-03","ids":{"openalex":"https://openalex.org/W4399534502","doi":"https://doi.org/10.1145/3662006.3662066"},"language":"en","primary_location":{"id":"doi:10.1145/3662006.3662066","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3662006.3662066","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3662006.3662066","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Workshop on Edge and Mobile Foundation Models","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3662006.3662066","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5034382142","display_name":"Daliang Xu","orcid":"https://orcid.org/0000-0002-6775-0688"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Daliang Xu","raw_affiliation_strings":["Peking University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-6775-0688","affiliations":[{"raw_affiliation_string":"Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Hao Zhang","orcid":"https://orcid.org/0009-0007-1107-4688"},"institutions":[{"id":"https://openalex.org/I21193070","display_name":"Beijing Jiaotong University","ror":"https://ror.org/01yj56c84","country_code":"CN","type":"education","lineage":["https://openalex.org/I21193070"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hao Zhang","raw_affiliation_strings":["Beijing jiaotong Univsersity, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0007-1107-4688","affiliations":[{"raw_affiliation_string":"Beijing jiaotong Univsersity, Beijing, China","institution_ids":["https://openalex.org/I21193070"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101655748","display_name":"Liming Yang","orcid":"https://orcid.org/0009-0009-1163-028X"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Liming Yang","raw_affiliation_strings":["Peking University, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0009-1163-028X","affiliations":[{"raw_affiliation_string":"Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010428676","display_name":"Ruiqi Liu","orcid":"https://orcid.org/0009-0004-3188-0976"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ruiqi Liu","raw_affiliation_strings":["Peking University, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0004-3188-0976","affiliations":[{"raw_affiliation_string":"Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089690212","display_name":"Mengwei Xu","orcid":"https://orcid.org/0000-0001-6271-6993"},"institutions":[{"id":"https://openalex.org/I139759216","display_name":"Beijing University of Posts and Telecommunications","ror":"https://ror.org/04w9fbh59","country_code":"CN","type":"education","lineage":["https://openalex.org/I139759216"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Mengwei Xu","raw_affiliation_strings":["Beijing University of Posts and Telecommunications, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-6271-6993","affiliations":[{"raw_affiliation_string":"Beijing University of Posts and Telecommunications, Beijing, China","institution_ids":["https://openalex.org/I139759216"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5052249316","display_name":"Xuanzhe Liu","orcid":"https://orcid.org/0000-0002-7908-8484"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xuanzhe Liu","raw_affiliation_strings":["Peking University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-7908-8484","affiliations":[{"raw_affiliation_string":"Peking University, Beijing, China","institution_ids":["https://openalex.org/I20231570"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5034382142"],"corresponding_institution_ids":["https://openalex.org/I20231570"],"apc_list":null,"apc_paid":null,"fwci":1.0092,"has_fulltext":false,"cited_by_count":3,"citation_normalized_percentile":{"value":0.76418838,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"33","last_page":"35"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11158","display_name":"Wireless Networks and Protocols","score":0.9987000226974487,"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"}},"topics":[{"id":"https://openalex.org/T11158","display_name":"Wireless Networks and Protocols","score":0.9987000226974487,"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/T10575","display_name":"Wireless Communication Networks Research","score":0.9962999820709229,"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/T12146","display_name":"Power Line Communications and Noise","score":0.9926000237464905,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6119725108146667}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6119725108146667}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3662006.3662066","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3662006.3662066","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3662006.3662066","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Workshop on Edge and Mobile Foundation Models","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3662006.3662066","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3662006.3662066","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3662006.3662066","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Workshop on Edge and Mobile Foundation Models","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G3791636283","display_name":null,"funder_award_id":"62325201","funder_id":"https://openalex.org/F4320323817","funder_display_name":"Universitas Brawijaya"}],"funders":[{"id":"https://openalex.org/F4320323817","display_name":"Universitas Brawijaya","ror":"https://ror.org/01wk3d929"}],"has_content":{"pdf":true,"grobid_xml":false},"content_urls":{"pdf":"https://content.openalex.org/works/W4399534502.pdf"},"referenced_works_count":2,"referenced_works":["https://openalex.org/W4306178406","https://openalex.org/W4366549767"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052","https://openalex.org/W2382290278","https://openalex.org/W4395014643"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"play":[4],"a":[5,24],"crucial":[6],"role":[7],"in":[8],"various":[9],"Natural":[10],"Language":[11],"Processing":[12,53],"(NLP)":[13],"tasks,":[14],"prompting":[15],"their":[16],"deployment":[17],"on":[18],"mobile":[19],"devices":[20],"for":[21,34],"inference.":[22],"However,":[23],"significant":[25],"challenge":[26],"arises":[27],"due":[28],"to":[29,65,83,96],"high":[30],"waiting":[31],"latency,":[32],"especially":[33],"long":[35],"prompts.":[36],"This":[37],"paper":[38],"introduces":[39],"mllm-NPU,":[40],"the":[41,57],"first":[42],"system":[43],"enabling":[44],"efficient":[45],"on-device":[46],"LLM":[47,66],"prefilling":[48,67],"acceleration":[49,90],"using":[50],"on-chip":[51],"Neural":[52],"Units":[54],"(NPUs).":[55],"Despite":[56],"impressive":[58],"compute":[59,86],"capabilities":[60],"of":[61],"NPUs,":[62],"direct":[63],"application":[64],"often":[68],"falls":[69],"short.":[70],"To":[71],"this":[72],"end,":[73],"mllm-NPU":[74],"incorporates":[75],"two":[76],"key":[77],"techniques:":[78],"(1)":[79],"chunk-wise":[80],"CPU-NPU":[81],"co-scheduling":[82],"handle":[84],"static":[85,99],"graphs":[87],"and":[88],"INT8-only":[89],"problems.":[91],"(2)":[92],"dynamic":[93],"outlier":[94],"inference":[95],"deal":[97],"with":[98],"activation":[100],"quantization":[101],"sacrificing":[102],"accuracy":[103],"problem.":[104]},"counts_by_year":[{"year":2025,"cited_by_count":3}],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-10-10T00:00:00"}
