{"id":"https://openalex.org/W7225828643","doi":"https://doi.org/10.5281/zenodo.24146465","title":"The Agent Memory Wall","display_name":"The Agent Memory Wall","publication_year":2026,"publication_date":"2026-11-11","ids":{"openalex":"https://openalex.org/W7225828643","doi":"https://doi.org/10.5281/zenodo.24146465"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.24146465","doi":"https://doi.org/10.5281/zenodo.24146465","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24146465","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.24146465","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137686721","display_name":"Bharat Bhushan sharma","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"sharma, Bharat Bhushan","raw_affiliation_strings":["PHENEX TECHNOLOGIES INC"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"PHENEX TECHNOLOGIES INC","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5137686721"],"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":0.8978999853134155,"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":0.8978999853134155,"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/T14098","display_name":"Energy Efficiency in Computing","score":0.03350000083446503,"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/T12238","display_name":"Green IT and Sustainability","score":0.010599999688565731,"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":[],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.763700008392334},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.6291000247001648},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.6227999925613403},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.5742999911308289},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.5565000176429749},{"id":"https://openalex.org/C186370098","wikidata":"https://www.wikidata.org/wiki/Q442787","display_name":"Energy (signal processing)","level":2,"score":0.5508000254631042},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.5177000164985657},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.49160000681877136},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.48330000042915344},{"id":"https://openalex.org/C2778348673","wikidata":"https://www.wikidata.org/wiki/Q739302","display_name":"Production (economics)","level":2,"score":0.39570000767707825},{"id":"https://openalex.org/C126831891","wikidata":"https://www.wikidata.org/wiki/Q221673","display_name":"Host (biology)","level":2,"score":0.3869999945163727},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.37070000171661377},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3614000082015991},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.35179999470710754},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.31949999928474426},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.30889999866485596},{"id":"https://openalex.org/C2780165032","wikidata":"https://www.wikidata.org/wiki/Q16869822","display_name":"Energy consumption","level":2,"score":0.2946000099182129},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.290800005197525},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.2825999855995178},{"id":"https://openalex.org/C3017489831","wikidata":"https://www.wikidata.org/wiki/Q2393193","display_name":"Running time","level":2,"score":0.28220000863075256},{"id":"https://openalex.org/C77265313","wikidata":"https://www.wikidata.org/wiki/Q879844","display_name":"Rest (music)","level":2,"score":0.2630999982357025},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.2572000026702881},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.25529998540878296},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.2526000142097473}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.24146465","doi":"https://doi.org/10.5281/zenodo.24146465","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24146465","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.24146466","doi":"https://doi.org/10.5281/zenodo.24146466","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24146466","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.24146465","doi":"https://doi.org/10.5281/zenodo.24146465","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24146465","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":[],"abstract_inverted_index":{"An":[0],"AI":[1],"agent":[2],"alternates":[3],"between":[4,137,147],"model":[5,235],"calls":[6],"and":[7,54,61,63,79,130,139,146,149,204,286,306,316],"tool":[8,12],"calls.":[9],"While":[10],"a":[11,108,126,156,177,188,220,233],"runs,":[13,277],"the":[14,18,28,34,38,100,105,134,168,181,186,194,201,212,224,237,258,264,294,308,317],"serving":[15],"engine":[16],"keeps":[17],"agent\u2019s":[19],"key-value":[20],"(KV)":[21],"cache":[22,106,309,318],"only":[23,256],"if":[24],"nothing":[25],"else":[26],"needs":[27],"memory;":[29],"when":[30,257],"it":[31,278],"is":[32,120,326],"evicted,":[33],"next":[35],"call":[36],"recomputes":[37],"whole":[39],"context.":[40],"We":[41,91],"measure":[42],"what":[43],"this":[44],"costs":[45],"in":[46,215,274],"GPU":[47,170,195,202,327],"energy":[48,117,171,240,268,288,325],"on":[49,96,176,219,248,255],"an":[50,251],"H200":[51],"with":[52,200,305],"vLLM":[53],"SGLang,":[55],"three":[56],"Qwen3":[57],"models":[58],"(32B,":[59],"30B-A3B":[60],"8B)":[62],"four":[64,93],"workloads:":[65],"synthetic":[66,97,249],"sessions,":[67,98,250],"recorded":[68],"SWE-agent":[69],"trajectories,":[70],"replayed":[71],"samples":[72],"of":[73,110,123,142,151,167,172,180,296,298,313],"production":[74,182],"tool-agent":[75],"traffic":[76],"from":[77,159],"Kimi,":[78],"graded":[80],"BFCL":[81],"multi-turn":[82],"tasks.":[83],"Every":[84],"prediction":[85],"was":[86],"registered":[87],"before":[88],"its":[89],"run.":[90],"report":[92],"results.":[94],"First,":[95],"once":[99],"agents\u2019":[101],"working":[102,259],"set":[103,260],"exceeds":[104,261],"by":[107,291,311],"factor":[109],"1.2-2.2,":[111],"hits":[112],"collapse;":[113],"past":[114],"that":[115,270],"point":[116],"per":[118,241,289,322],"turn":[119,290],"within":[121,244],"0.01-1.6%":[122],"running":[124],"without":[125,307],"cache.":[127],"With":[128],"Qwen3-32B":[129],"5":[131],"s":[132],"pauses":[133],"collapse":[135],"comes":[136],"16":[138,150],"32":[140],"sessions":[141],"about":[143,152,163,211,245],"11k":[144],"tokens":[145,209],"8":[148],"27k.":[153],"Second,":[154],"restoring":[155],"prompt\u2019s":[157],"KV":[158],"host":[160,191],"memory":[161],"takes":[162],"1/110":[164],"to":[165,243],"1/220":[166],"above-idle":[169],"recomputing":[173],"it.":[174],"Third,":[175],"1/4":[178],"sample":[179,223],"trace,":[183],"which":[184],"overloads":[185],"GPU,":[187],"160":[189],"GB":[190],"tier":[192,225],"let":[193],"finish":[196],"every":[197,280],"request":[198],"(85-86%":[199],"cache)":[203],"process":[205],"16-18%":[206],"more":[207,284],"prompt":[208],"for":[210],"same":[213],"energy,":[214],"two":[216,275],"time":[217],"windows;":[218],"lighter":[221],"1/8":[222],"saved":[226],"under":[227],"1%.":[228,246],"Given":[229],"measured":[230],"hit":[231],"rates,":[232],"one-parameter":[234],"predicts":[236],"busy":[238],"card\u2019s":[239],"token":[242],"Fourth,":[247],"admission":[252],"gate,":[253],"switched":[254],"1.2":[262],"times":[263],"cache,":[265],"cost":[266],"no":[267],"below":[269],"point;":[271],"above":[272],"it,":[273],"10-minute":[276],"served":[279],"session,":[281],"completed":[282],"48-62%":[283],"turns":[285],"cut":[287],"42-50%,":[292],"at":[293,300],"price":[295],"minutes":[297],"queueing":[299],"admission.":[301],"On":[302],"BFCL,":[303],"accuracy":[304],"differed":[310],"3":[312],"400":[314],"tasks,":[315],"doubled":[319],"correct":[320],"answers":[321],"kWh.":[323],"All":[324],"board":[328],"energy.":[329]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-12T00:00:00"}