{"id":"https://openalex.org/W4416932162","doi":"https://doi.org/10.1145/3754598.3754671","title":"LLaMCAT: Optimizing Large Language Model Inference with Cache Arbitration and Throttling","display_name":"LLaMCAT: Optimizing Large Language Model Inference with Cache Arbitration and Throttling","publication_year":2025,"publication_date":"2025-09-08","ids":{"openalex":"https://openalex.org/W4416932162","doi":"https://doi.org/10.1145/3754598.3754671"},"language":"en","primary_location":{"id":"doi:10.1145/3754598.3754671","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3754598.3754671","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3754598.3754671?download=true","source":null,"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 54th International Conference on Parallel Processing","raw_type":"proceedings-article"},"type":"article","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3754598.3754671?download=true","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5028367760","display_name":"Zhongchun Zhou","orcid":"https://orcid.org/0009-0000-7037-7418"},"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":true,"raw_author_name":"Zhongchun Zhou","raw_affiliation_strings":["The Hong Kong University of Science and Technology, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0009-0000-7037-7418","affiliations":[{"raw_affiliation_string":"The 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/A5101722744","display_name":"Chengtao Lai","orcid":null},"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":"Chengtao Lai","raw_affiliation_strings":["The Hong Kong University of Science and Technology, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0002-9547-9653","affiliations":[{"raw_affiliation_string":"The Hong Kong University of Science and Technology, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"last","author":{"id":null,"display_name":"Wei Zhang","orcid":"https://orcid.org/0000-0002-7622-6714"},"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":"Wei Zhang","raw_affiliation_strings":["The Hong Kong University of Science and Technology, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0002-7622-6714","affiliations":[{"raw_affiliation_string":"The Hong Kong University of Science and Technology, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5028367760"],"corresponding_institution_ids":["https://openalex.org/I200769079"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.41069409,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"460","last_page":"469"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.5705000162124634,"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":0.5705000162124634,"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/T14347","display_name":"Big Data and Digital Economy","score":0.07980000227689743,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.047600001096725464,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.744700014591217},{"id":"https://openalex.org/keywords/cache-algorithms","display_name":"Cache algorithms","score":0.592199981212616},{"id":"https://openalex.org/keywords/cache-invalidation","display_name":"Cache invalidation","score":0.47540000081062317},{"id":"https://openalex.org/keywords/cache-pollution","display_name":"Cache pollution","score":0.4609000086784363},{"id":"https://openalex.org/keywords/smart-cache","display_name":"Smart Cache","score":0.3919999897480011},{"id":"https://openalex.org/keywords/memory-bandwidth","display_name":"Memory bandwidth","score":0.3880999982357025},{"id":"https://openalex.org/keywords/cache-coloring","display_name":"Cache coloring","score":0.38679999113082886},{"id":"https://openalex.org/keywords/callback","display_name":"Callback","score":0.36390000581741333},{"id":"https://openalex.org/keywords/bandwidth-throttling","display_name":"Bandwidth throttling","score":0.34929999709129333}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8583999872207642},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.744700014591217},{"id":"https://openalex.org/C38556500","wikidata":"https://www.wikidata.org/wiki/Q13404475","display_name":"Cache algorithms","level":4,"score":0.592199981212616},{"id":"https://openalex.org/C25536678","wikidata":"https://www.wikidata.org/wiki/Q5015977","display_name":"Cache invalidation","level":5,"score":0.47540000081062317},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.46209999918937683},{"id":"https://openalex.org/C113166858","wikidata":"https://www.wikidata.org/wiki/Q5015981","display_name":"Cache pollution","level":5,"score":0.4609000086784363},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.45019999146461487},{"id":"https://openalex.org/C167713795","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"Smart Cache","level":5,"score":0.3919999897480011},{"id":"https://openalex.org/C188045654","wikidata":"https://www.wikidata.org/wiki/Q17148339","display_name":"Memory bandwidth","level":2,"score":0.3880999982357025},{"id":"https://openalex.org/C201148951","wikidata":"https://www.wikidata.org/wiki/Q5015976","display_name":"Cache coloring","level":4,"score":0.38679999113082886},{"id":"https://openalex.org/C204495577","wikidata":"https://www.wikidata.org/wiki/Q1205349","display_name":"Callback","level":2,"score":0.36390000581741333},{"id":"https://openalex.org/C173061102","wikidata":"https://www.wikidata.org/wiki/Q478819","display_name":"Bandwidth throttling","level":3,"score":0.34929999709129333},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.34450000524520874},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.31700000166893005},{"id":"https://openalex.org/C138101251","wikidata":"https://www.wikidata.org/wiki/Q213092","display_name":"Thread (computing)","level":2,"score":0.3151000142097473},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.310699999332428},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.3073999881744385},{"id":"https://openalex.org/C3720319","wikidata":"https://www.wikidata.org/wiki/Q5015937","display_name":"Cache-only memory architecture","level":5,"score":0.3057999908924103},{"id":"https://openalex.org/C25344961","wikidata":"https://www.wikidata.org/wiki/Q192726","display_name":"Virtual machine","level":2,"score":0.3041999936103821},{"id":"https://openalex.org/C51185590","wikidata":"https://www.wikidata.org/wiki/Q1017228","display_name":"Bus sniffing","level":5,"score":0.3034000098705292},{"id":"https://openalex.org/C76399640","wikidata":"https://www.wikidata.org/wiki/Q189401","display_name":"Virtual memory","level":4,"score":0.30329999327659607},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.2827000021934509},{"id":"https://openalex.org/C133371097","wikidata":"https://www.wikidata.org/wiki/Q868014","display_name":"Non-uniform memory access","level":5,"score":0.263700008392334},{"id":"https://openalex.org/C12186640","wikidata":"https://www.wikidata.org/wiki/Q6815743","display_name":"Memory model","level":3,"score":0.2628999948501587},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.2621000111103058},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.25929999351501465},{"id":"https://openalex.org/C59687516","wikidata":"https://www.wikidata.org/wiki/Q5015938","display_name":"Cache-oblivious algorithm","level":5,"score":0.257999986410141},{"id":"https://openalex.org/C36340418","wikidata":"https://www.wikidata.org/wiki/Q7124288","display_name":"Page cache","level":5,"score":0.2578999996185303},{"id":"https://openalex.org/C2779602883","wikidata":"https://www.wikidata.org/wiki/Q15544750","display_name":"Memory architecture","level":2,"score":0.2538999915122986}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3754598.3754671","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3754598.3754671","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3754598.3754671?download=true","source":null,"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 54th International Conference on Parallel Processing","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2512.00083","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2512.00083","pdf_url":"https://arxiv.org/pdf/2512.00083","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-168387","is_oa":false,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-168387","pdf_url":null,"source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Conference paper"}],"best_oa_location":{"id":"doi:10.1145/3754598.3754671","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3754598.3754671","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3754598.3754671?download=true","source":null,"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 54th International Conference on Parallel Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416932162.pdf","grobid_xml":"https://content.openalex.org/works/W4416932162.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"have":[4],"achieved":[5],"unprecedented":[6],"success":[7],"across":[8],"various":[9],"applications,":[10],"but":[11],"their":[12],"substantial":[13],"memory":[14,21,98],"requirements":[15],"pose":[16],"significant":[17],"challenges":[18],"to":[19,49,71],"current":[20],"system":[22],"designs,":[23],"especially":[24],"during":[25],"inference.":[26,55],"Our":[27],"work":[28],"targets":[29],"last-level":[30],"cache":[31,66,78],"(LLC)":[32],"based":[33],"architectures,":[34],"including":[35],"GPUs":[36],"(e.g.,":[37],"NVIDIA":[38],"GPUs)":[39],"and":[40,63,76,103],"AI":[41],"accelerators.":[42],"We":[43,84],"introduce":[44],"LLaMCAT,":[45],"a":[46,87],"novel":[47],"approach":[48],"optimize":[50],"the":[51],"LLC":[52],"for":[53],"LLM":[54],"LLaMCAT":[56],"combines":[57],"Miss":[58],"Status":[59],"Holding":[60],"Register":[61],"(MSHR)-":[62],"load":[64],"balance-aware":[65],"arbitration":[67],"with":[68,94],"thread":[69],"throttling":[70],"address":[72],"stringent":[73],"bandwidth":[74],"demands":[75],"minimize":[77],"stalls":[79],"in":[80],"KV":[81],"Cache":[82],"access.":[83],"also":[85],"propose":[86],"hybrid":[88],"simulation":[89],"framework":[90],"integrating":[91],"analytical":[92],"models":[93],"cycle-level":[95],"simulators":[96],"via":[97],"traces,":[99],"balancing":[100],"architecture":[101],"detail":[102],"efficiency.":[104]},"counts_by_year":[],"updated_date":"2026-03-14T06:41:57.775601","created_date":"2025-12-03T00:00:00"}