{"id":"https://openalex.org/W7225174378","doi":"https://doi.org/10.5281/zenodo.23915485","title":"PAKP: Predictive Asymmetric KV-Paging via Centered Bounding-Box Lookahead for Sub-6GB Edge GPUs","display_name":"PAKP: Predictive Asymmetric KV-Paging via Centered Bounding-Box Lookahead for Sub-6GB Edge GPUs","publication_year":2026,"publication_date":"2026-11-03","ids":{"openalex":"https://openalex.org/W7225174378","doi":"https://doi.org/10.5281/zenodo.23915485"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.23915485","doi":"https://doi.org/10.5281/zenodo.23915485","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23915485","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},"volume":null,"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.23915485","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5160441854","display_name":"Akhil M","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Akhil M","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5160441854"],"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9897000193595886,"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":0.9897000193595886,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.0012000000569969416,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T10028","display_name":"Topic Modeling","score":0.0007999999797903001,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/large-language-models","display_name":"large language models","score":0.8299999833106995}],"concepts":[{"id":"https://openalex.org/C64270927","wikidata":"https://www.wikidata.org/wiki/Q206924","display_name":"PCI Express","level":3,"score":0.8420000076293945},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8001000285148621},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.4799000024795532},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.44600000977516174},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.42100000381469727},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.39169999957084656},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.3677999973297119},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.36419999599456787},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.35280001163482666},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.3474999964237213},{"id":"https://openalex.org/C126831891","wikidata":"https://www.wikidata.org/wiki/Q221673","display_name":"Host (biology)","level":2,"score":0.33480000495910645},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.3215000033378601},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.3165999948978424},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.30889999866485596},{"id":"https://openalex.org/C513985346","wikidata":"https://www.wikidata.org/wiki/Q270471","display_name":"Virtualization","level":3,"score":0.2985999882221222},{"id":"https://openalex.org/C55803922","wikidata":"https://www.wikidata.org/wiki/Q1755472","display_name":"SBus","level":4,"score":0.29589998722076416},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.2867000102996826},{"id":"https://openalex.org/C186967261","wikidata":"https://www.wikidata.org/wiki/Q5082128","display_name":"Mobile device","level":2,"score":0.2800000011920929},{"id":"https://openalex.org/C130795937","wikidata":"https://www.wikidata.org/wiki/Q2561570","display_name":"Remote direct memory access","level":2,"score":0.27900001406669617},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.2687000036239624},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.2639999985694885},{"id":"https://openalex.org/C172173386","wikidata":"https://www.wikidata.org/wiki/Q79984","display_name":"Ethernet","level":2,"score":0.2621000111103058},{"id":"https://openalex.org/C2911091166","wikidata":"https://www.wikidata.org/wiki/Q106419912","display_name":"Transplantation","level":2,"score":0.26080000400543213},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.25999999046325684},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C133588205","wikidata":"https://www.wikidata.org/wiki/Q28455645","display_name":"Instruction prefetch","level":3,"score":0.25870001316070557},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.2515999972820282}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.23915485","doi":"https://doi.org/10.5281/zenodo.23915485","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23915485","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},{"id":"doi:10.5281/zenodo.23915486","doi":"https://doi.org/10.5281/zenodo.23915486","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23915486","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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:10.5281/zenodo.23915485","doi":"https://doi.org/10.5281/zenodo.23915485","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23915485","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7172201241","https://openalex.org/W7203467034","https://openalex.org/W7127413828","https://openalex.org/W7126020484","https://openalex.org/W6929557363","https://openalex.org/W7163635936","https://openalex.org/W7106118842","https://openalex.org/W7160783786","https://openalex.org/W7151394422","https://openalex.org/W7213578301"],"abstract_inverted_index":{"Deploying":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"with":[5,85,143],"extended":[6],"context":[7],"windows":[8],"(32k\u201364k+":[9],"tokens)":[10],"on":[11,66,148],"consumer-tier":[12],"edge":[13],"hardware":[14],"(e.g.,":[15,59],"NVIDIA":[16],"RTX":[17],"4050":[18,154],"6GB":[19,155],"VRAM)":[20],"is":[21],"fundamentally":[22],"bottlenecked":[23],"by":[24,172],"the":[25,30],"linear":[26],"memory":[27,78],"growth":[28],"of":[29,163],"Key-Value":[31],"(KV)":[32],"cache":[33],"and":[34,104,182],"severe":[35],"PCIe":[36,51,140,164],"bus":[37,141],"bandwidth":[38],"constraints.":[39],"Naive":[40],"host":[41,180],"offloading":[42],"degrades":[43],"token":[44],"generation":[45],"throughput":[46],"to":[47,124],"<3.2":[48],"tok/s":[49],"over":[50],"Gen4":[52],"(11.81":[53],"GB/s),":[54],"while":[55],"static":[56],"eviction":[57],"rules":[58],"sliding":[60],"windows)":[61],"suffer":[62],"catastrophic":[63],"recall":[64,177],"failure":[65],"historical":[67],"needle":[68],"tokens.":[69],"We":[70],"present":[71],"PAKP":[72,92,160],"(Predictive":[73],"Asymmetric":[74],"KV-Paging),":[75],"a":[76,86],"two-tier":[77],"engine":[79],"that":[80,100,138],"pairs":[81],"active":[82,168],"GPU":[83,144],"VRAM":[84,99,170],"32GB":[87,157],"page-locked":[88],"Host":[89],"RAM":[90],"pool.":[91],"introduces:1.":[93],"Centered":[94],"Subspace":[95],"Bounding-Box":[96],"Descriptors":[97],"in":[98],"eliminate":[101],"key-space":[102],"anisotropy":[103],"retain":[105],"single-token":[106],"passkey":[107,176],"spikes":[108],"across":[109],"64-token":[110],"pages.2.":[111],"RoPE-Aligned":[112],"Cross-Layer":[113],"Lookahead":[114],"Prefetching":[115],"leveraging":[116],"residual":[117],"stream":[118],"continuity":[119],"(cos(h_L,":[120],"h_{L+1})":[121],"=":[122],"0.9842)":[123],"project":[125],"Layer":[126,131],"L+1":[127],"query":[128],"access":[129],"during":[130],"L":[132],"execution.3.":[133],"Dual-Stream":[134],"Asynchronous":[135],"DMA":[136],"transfers":[137],"overlap":[139],"latency":[142],"compute.":[145],"Empirically":[146],"evaluated":[147],"an":[149],"ASUS":[150],"TUF":[151],"F15":[152],"(RTX":[153],"VRAM,":[156],"DDR5":[158],"RAM),":[159],"hides":[161],"90.5%":[162],"transfer":[165],"latency,":[166],"reduces":[167],"KV":[169],"consumption":[171],"79.2%,":[173],"maintains":[174],"100%":[175],"from":[178],"evicted":[179],"memory,":[181],"sustains":[183],"27.1\u201339.2":[184],"tokens/sec.":[185]},"counts_by_year":[],"updated_date":"2026-11-05T12:25:35.769452","created_date":"2026-11-04T00:00:00"}
