{"id":"https://openalex.org/W7225455683","doi":"https://doi.org/10.5281/zenodo.24026971","title":"Reload or Recompute? A Calibrated Rule for KV-Cache Reuse Across Commodity GPUs","display_name":"Reload or Recompute? A Calibrated Rule for KV-Cache Reuse Across Commodity GPUs","publication_year":2026,"publication_date":"2026-11-07","ids":{"openalex":"https://openalex.org/W7225455683","doi":"https://doi.org/10.5281/zenodo.24026971"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.24026971","doi":"https://doi.org/10.5281/zenodo.24026971","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24026971","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","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.24026971","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5160466287","display_name":"Mohith M","orcid":"https://orcid.org/0009-0003-2978-3045","observed_orcids":["https://orcid.org/0009-0003-2978-3045"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"M, Mohith","raw_affiliation_strings":["Chinmaya Vishwa Vidyapeeth"],"raw_orcid":"https://orcid.org/0009-0003-2978-3045","affiliations":[{"raw_affiliation_string":"Chinmaya Vishwa Vidyapeeth","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5160466287"],"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":0.9711999893188477,"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":0.9711999893188477,"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":0.00559999980032444,"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":0.002300000051036477,"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/time-to-first-token","display_name":"time-to-first-token","score":0.8809999823570251},{"id":"https://openalex.org/keywords/gpu-memory-bandwidth","display_name":"GPU memory bandwidth","score":0.503000020980835},{"id":"https://openalex.org/keywords/calibration","display_name":"calibration","score":0.5260000228881836},{"id":"https://openalex.org/keywords/large-language-model-inference","display_name":"large language model inference","score":0.4090000092983246},{"id":"https://openalex.org/keywords/long-context-inference","display_name":"long-context inference","score":0.46399998664855957},{"id":"https://openalex.org/keywords/qwen2-5","display_name":"Qwen2.5","score":0.6039999723434448}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7576000094413757},{"id":"https://openalex.org/C141603448","wikidata":"https://www.wikidata.org/wiki/Q134830","display_name":"Prefix","level":2,"score":0.7232000231742859},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.5450000166893005},{"id":"https://openalex.org/C132459708","wikidata":"https://www.wikidata.org/wiki/Q744069","display_name":"Extrapolation","level":2,"score":0.5029000043869019},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.45260000228881836},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.41280001401901245},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.39340001344680786},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.3781999945640564},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.3716000020503998},{"id":"https://openalex.org/C2779439359","wikidata":"https://www.wikidata.org/wiki/Q317088","display_name":"Commodity","level":2,"score":0.36649999022483826},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.34549999237060547},{"id":"https://openalex.org/C45235069","wikidata":"https://www.wikidata.org/wiki/Q278425","display_name":"Table (database)","level":2,"score":0.31610000133514404},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.31529998779296875},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3149999976158142},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3140000104904175},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.3116999864578247},{"id":"https://openalex.org/C150202949","wikidata":"https://www.wikidata.org/wiki/Q107602","display_name":"Pointer (user interface)","level":2,"score":0.29809999465942383},{"id":"https://openalex.org/C113954288","wikidata":"https://www.wikidata.org/wiki/Q186885","display_name":"Timestamp","level":2,"score":0.29350000619888306},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.29100000858306885},{"id":"https://openalex.org/C2777200299","wikidata":"https://www.wikidata.org/wiki/Q52943","display_name":"Conversation","level":2,"score":0.272599995136261},{"id":"https://openalex.org/C49154492","wikidata":"https://www.wikidata.org/wiki/Q5300","display_name":"Central processing unit","level":2,"score":0.26330000162124634},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.25}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.24026971","doi":"https://doi.org/10.5281/zenodo.24026971","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24026971","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.24026972","doi":"https://doi.org/10.5281/zenodo.24026972","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24026972","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.24026971","doi":"https://doi.org/10.5281/zenodo.24026971","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24026971","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/W7225178476","https://openalex.org/W7212226223","https://openalex.org/W7162456234","https://openalex.org/W7222852259","https://openalex.org/W4403444130","https://openalex.org/W7222428471","https://openalex.org/W7204182843","https://openalex.org/W7213409226","https://openalex.org/W7154612387","https://openalex.org/W7165142542"],"abstract_inverted_index":{"When":[0],"a":[1,4,18,24,95,99,132],"conversation":[2],"returns,":[3],"server":[5],"must":[6],"choose":[7],"whether":[8],"reloading":[9],"its":[10],"key-value":[11],"(KV)":[12],"cache":[13],"or":[14],"recomputingits":[15],"prefix":[16,97],"gives":[17],"faster":[19],"first":[20,31],"token.":[21],"We":[22,124],"evaluate":[23],"break-even":[25,103],"bandwidth":[26],"rule":[27],"for":[28],"time":[29],"to":[30,110],"token":[32],"(TTFT)":[33],"thatincludes":[34],"resident-cache":[35],"continuation,":[36],"fixed":[37],"overhead,":[38],"and":[39,41,52,59,64,70,73,88,98,115,131],"CPU":[40],"transfer":[42],"costs.":[43],"Building":[44],"on":[45,62,82,85,113],"existing":[46],"calibrationand":[47],"compute/load":[48],"analyses,":[49],"we":[50],"timestamp":[51],"freeze":[53],"per-cell":[54],"predictions":[55,76],"after":[56],"short":[57,68],"calibration":[58],"evaluateseparate":[60],"sweeps":[61],"commodity":[63],"older":[65],"GPUs,":[66],"including":[67],"contexts":[69],"separate":[71],"CPU-memory":[72],"filepaths.":[74],"Frozen":[75],"correctly":[77],"classify":[78],"574/636":[79],"resolved":[80],"decisions":[81],"L4,":[83],"192/206":[84],"A10":[86,116],"x16,":[87],"398/426":[89],"onA10G":[90],"x8.":[91],"For":[92],"Qwen2.5-1.5B":[93],"with":[94],"1,024-token":[96],"32-token":[100],"new":[101],"turn,":[102],"simulated-bandwidthbrackets":[104],"rise":[105],"from":[106],"T4\u2019s":[107],"100\u2013400":[108],"MB/s":[109,112],"800\u20131,600":[111],"L4":[114],"x16.":[117],"Long-context":[118],"extrapolation":[119],"canproduce":[120],"large":[121],"TTFT":[122],"errors.":[123],"release":[125],"measurements,":[126],"frozen":[127],"predictions,":[128],"reference":[129],"code,":[130],"source-linkedclaim":[133],"ledger.":[134]},"counts_by_year":[],"updated_date":"2026-11-09T13:22:03.289127","created_date":"2026-11-08T00:00:00"}