{"id":"https://openalex.org/W7165365033","doi":"https://doi.org/10.48550/arxiv.2606.20373","title":"AutoPass: Evidence-Guided LLM Agents for Compiler Performance Tuning","display_name":"AutoPass: Evidence-Guided LLM Agents for Compiler Performance Tuning","publication_year":2026,"publication_date":"2026-06-18","ids":{"openalex":"https://openalex.org/W7165365033","doi":"https://doi.org/10.48550/arxiv.2606.20373"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.20373","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20373","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.20373","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139012669","display_name":"Zepeng Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zepeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138965740","display_name":"Jie Ren","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ren, Jie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069158506","display_name":"Zhanyong Tang","orcid":"https://orcid.org/0000-0002-4333-2334"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Zhanyong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139019633","display_name":"Jie Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Jie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138974816","display_name":"Zheng Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"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/T10181","display_name":"Natural Language Processing Techniques","score":0.20739999413490295,"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"}},"topics":[{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.20739999413490295,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.1639000028371811,"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/T10028","display_name":"Topic Modeling","score":0.10970000177621841,"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/compiler","display_name":"Compiler","score":0.9049999713897705},{"id":"https://openalex.org/keywords/compiler-correctness","display_name":"Compiler correctness","score":0.6794999837875366},{"id":"https://openalex.org/keywords/optimizing-compiler","display_name":"Optimizing compiler","score":0.6518999934196472},{"id":"https://openalex.org/keywords/loop-optimization","display_name":"Loop optimization","score":0.6348000168800354},{"id":"https://openalex.org/keywords/interprocedural-optimization","display_name":"Interprocedural optimization","score":0.6128000020980835},{"id":"https://openalex.org/keywords/heuristics","display_name":"Heuristics","score":0.5667999982833862},{"id":"https://openalex.org/keywords/intermediate-language","display_name":"Intermediate language","score":0.4668999910354614},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4032999873161316},{"id":"https://openalex.org/keywords/dynamic-compilation","display_name":"Dynamic compilation","score":0.38580000400543213}],"concepts":[{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.9049999713897705},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.894599974155426},{"id":"https://openalex.org/C68366613","wikidata":"https://www.wikidata.org/wiki/Q5156378","display_name":"Compiler correctness","level":3,"score":0.6794999837875366},{"id":"https://openalex.org/C190902152","wikidata":"https://www.wikidata.org/wiki/Q1325106","display_name":"Optimizing compiler","level":3,"score":0.6518999934196472},{"id":"https://openalex.org/C29331672","wikidata":"https://www.wikidata.org/wiki/Q3354468","display_name":"Loop optimization","level":4,"score":0.6348000168800354},{"id":"https://openalex.org/C111564260","wikidata":"https://www.wikidata.org/wiki/Q4288856","display_name":"Interprocedural optimization","level":5,"score":0.6128000020980835},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5903000235557556},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.5769000053405762},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.5667999982833862},{"id":"https://openalex.org/C77660490","wikidata":"https://www.wikidata.org/wiki/Q244916","display_name":"Intermediate language","level":3,"score":0.4668999910354614},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4032999873161316},{"id":"https://openalex.org/C8767382","wikidata":"https://www.wikidata.org/wiki/Q1058454","display_name":"Dynamic compilation","level":3,"score":0.38580000400543213},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.37369999289512634},{"id":"https://openalex.org/C133162039","wikidata":"https://www.wikidata.org/wiki/Q1061077","display_name":"Code generation","level":3,"score":0.3709999918937683},{"id":"https://openalex.org/C139571649","wikidata":"https://www.wikidata.org/wiki/Q1156793","display_name":"Program optimization","level":3,"score":0.36079999804496765},{"id":"https://openalex.org/C9957790","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler construction","level":3,"score":0.3560999929904938},{"id":"https://openalex.org/C2780870223","wikidata":"https://www.wikidata.org/wiki/Q1004415","display_name":"Runtime system","level":2,"score":0.3553999960422516},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.3100000023841858},{"id":"https://openalex.org/C128916667","wikidata":"https://www.wikidata.org/wiki/Q1343660","display_name":"Register allocation","level":3,"score":0.3082999885082245},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3059999942779541},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.2921999990940094},{"id":"https://openalex.org/C2989134064","wikidata":"https://www.wikidata.org/wiki/Q288510","display_name":"Execution time","level":2,"score":0.2888999879360199},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.27730000019073486},{"id":"https://openalex.org/C76782552","wikidata":"https://www.wikidata.org/wiki/Q110546","display_name":"Just-in-time compilation","level":3,"score":0.27559998631477356},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.2720000147819519},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.26989999413490295},{"id":"https://openalex.org/C2777138346","wikidata":"https://www.wikidata.org/wiki/Q1714153","display_name":"Performance tuning","level":2,"score":0.2524000108242035},{"id":"https://openalex.org/C200833197","wikidata":"https://www.wikidata.org/wiki/Q333707","display_name":"Compile time","level":3,"score":0.2513999938964844}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.20373","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20373","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.20373","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20373","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"show":[4],"promise":[5],"for":[6,34],"code":[7],"compilation":[8],"tasks,":[9],"but":[10],"applying":[11],"them":[12],"to":[13,20,44,67,72,82,97,123],"runtime":[14,26,42,95],"performance":[15,36],"tuning":[16,37],"is":[17],"difficult":[18],"due":[19],"complex":[21],"microarchitectural":[22],"effects":[23],"and":[24,41,77,100,111,126,135,141,149,158,165],"noisy":[25],"measurements.":[27],"We":[28,128],"present":[29],"AutoPass,":[30],"a":[31,55],"multi-agent":[32],"framework":[33],"compiler":[35,40,53,66,84,134],"that":[38],"uses":[39],"evidence":[43],"guide":[45,101],"LLM-generated":[46],"optimization":[47,75,91],"decisions.":[48],"Rather":[49],"than":[50],"treating":[51],"the":[52,65,68,79,132],"as":[54],"black":[56],"box":[57],"like":[58],"prior":[59],"auto-tuning":[60],"schemes,":[61],"AutoPass":[62,104,130,145],"opens":[63],"up":[64],"LLM,":[69],"enabling":[70],"it":[71,120,137],"query":[73],"compiler-internal":[74],"states":[76],"analyze":[78],"intermediate":[80],"representation":[81],"orchestrate":[83],"options.":[85],"The":[86],"search":[87],"process":[88],"iteratively":[89],"refines":[90],"configurations":[92],"using":[93],"measured":[94],"feedback":[96],"diagnose":[98],"regressions":[99],"latency-improving":[102],"edits.":[103],"operates":[105],"in":[106],"an":[107],"inference-only,":[108],"training-free":[109],"setting":[110],"requires":[112],"no":[113],"offline":[114],"training":[115],"or":[116],"task-specific":[117],"fine-tuning,":[118],"making":[119],"readily":[121],"applicable":[122],"new":[124],"benchmarks":[125],"platforms.":[127],"implement":[129],"on":[131,138,163],"LLVM":[133,161],"evaluate":[136],"server-grade":[139],"x86-64":[140,164],"embedded":[142],"ARM64":[143],"systems.":[144],"outperforms":[146],"expert-tuned":[147],"heuristics":[148],"classical":[150],"autotuning":[151],"methods,":[152],"achieving":[153],"geometric-mean":[154],"speedups":[155],"of":[156],"1.043x":[157],"1.117x":[159],"over":[160],"-O3":[162],"ARM64,":[166],"respectively.":[167]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-20T00:00:00"}