{"id":"https://openalex.org/W7133508609","doi":"https://doi.org/10.1109/hpca68181.2026.11408526","title":"Characterizing Cloud-Native LLM Inference at Bytedance and Exposing Optimization Challenges and Opportunities for Future AI Accelerators","display_name":"Characterizing Cloud-Native LLM Inference at Bytedance and Exposing Optimization Challenges and Opportunities for Future AI Accelerators","publication_year":2026,"publication_date":"2026-01-31","ids":{"openalex":"https://openalex.org/W7133508609","doi":"https://doi.org/10.1109/hpca68181.2026.11408526"},"language":null,"primary_location":{"id":"doi:10.1109/hpca68181.2026.11408526","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408526","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128068184","display_name":"Jingwei Cai","orcid":null},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Jingwei Cai","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128091138","display_name":"Dehao Kong","orcid":null},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Dehao Kong","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128127206","display_name":"Hantao Huang","orcid":null},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Hantao Huang","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128066297","display_name":"Zishan Jiang","orcid":null},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Zishan Jiang","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024162917","display_name":"Zixuan Ma","orcid":"https://orcid.org/0000-0002-1840-8907"},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Zixuan Ma","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128052693","display_name":"Qingyu Guo","orcid":null},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Qingyu Guo","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128046422","display_name":"Zhenxing Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Zhenxing Zhang","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128104699","display_name":"Guiming Shi","orcid":null},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guiming Shi","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126073255","display_name":"Mingyu Gao","orcid":null},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Mingyu Gao","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128128049","display_name":"Kaisheng Ma","orcid":null},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kaisheng Ma","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5128058525","display_name":"Minghui Yu","orcid":null},"institutions":[{"id":"https://openalex.org/I231086526","display_name":"University of Siedlce","ror":"https://ror.org/01wkb9987","country_code":"PL","type":"education","lineage":["https://openalex.org/I231086526"]}],"countries":["PL"],"is_corresponding":false,"raw_author_name":"Minghui Yu","raw_affiliation_strings":["ByteDance Seed"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance Seed","institution_ids":["https://openalex.org/I231086526"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":11,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.25833765,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"19"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10714","display_name":"Software-Defined Networks and 5G","score":0.22789999842643738,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T10714","display_name":"Software-Defined Networks and 5G","score":0.22789999842643738,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T10101","display_name":"Cloud Computing and Resource Management","score":0.1665000021457672,"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/T10273","display_name":"IoT and Edge/Fog Computing","score":0.0868000015616417,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/inference","display_name":"Inference","score":0.4219000041484833},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2980000078678131},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.2888000011444092},{"id":"https://openalex.org/keywords/identification","display_name":"Identification (biology)","score":0.24449999630451202},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.2433999925851822}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5942000150680542},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4219000041484833},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3700000047683716},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2980000078678131},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.29120001196861267},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.2888000011444092},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2711000144481659},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.24449999630451202},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2433999925851822},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.24050000309944153}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/hpca68181.2026.11408526","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408526","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":17,"referenced_works":["https://openalex.org/W2606722458","https://openalex.org/W2935331687","https://openalex.org/W3012493694","https://openalex.org/W3043571714","https://openalex.org/W3135242540","https://openalex.org/W3211226062","https://openalex.org/W4281708879","https://openalex.org/W4297097426","https://openalex.org/W4327930469","https://openalex.org/W4380881063","https://openalex.org/W4393592481","https://openalex.org/W4404954357","https://openalex.org/W4407857403","https://openalex.org/W4408362299","https://openalex.org/W4411485910","https://openalex.org/W4414736038","https://openalex.org/W7133227460"],"related_works":[],"abstract_inverted_index":{"As":[0],"a":[1],"major":[2],"provider":[3],"of":[4,22,50,75,103,170],"LLM":[5,25,107],"inference":[6,20],"services,":[7],"ByteDance":[8,105],"has":[9],"continuously":[10],"explored":[11],"diverse":[12],"accelerator":[13,158],"options":[14],"to":[15,33,67,133],"meet":[16],"the":[17,46,70,99,104,113,121],"rapidly":[18],"growing":[19],"demands":[21,72],"various":[23],"heterogeneous":[24],"scenarios":[26,52],"with":[27],"higher":[28],"cost-effectiveness,":[29],"thereby":[30],"enabling":[31],"LLMs":[32],"serve":[34],"more":[35],"people":[36],"worldwide.":[37],"However,":[38],"during":[39],"this":[40,88,92],"process,":[41],"we":[42,94,125,129,166,190],"have":[43,126,191],"found":[44],"that":[45],"complexity":[47],"and":[48,53,62,73,84,97,101,119,123,131,146,152,154,163,177,182,186],"opacity":[49],"cloud":[51,55,117],"corresponding":[54],"accelerators":[56],"make":[57],"it":[58],"difficult":[59],"for":[60],"academia":[61],"many":[63],"innovative":[64],"chip":[65],"startups":[66],"fully":[68],"understand":[69,115],"real":[71],"challenges":[74,122],"these":[76],"scenarios,":[77,118],"which":[78,140],"in":[79,87],"turn":[80],"severely":[81],"restricts":[82],"innovation":[83,188],"application":[85],"potential":[86],"field.":[89],"To":[90],"bridge":[91],"gap,":[93],"first":[95],"present":[96,167],"analyze":[98],"data":[100],"characteristics":[102],"Doubao":[106],"app":[108],"across":[109],"multiple":[110],"dimensions,":[111],"helping":[112],"community":[114],"real-world":[116],"detail":[120],"opportunities":[124,189],"identified.":[127],"Second,":[128],"propose":[130],"plan":[132],"open-source":[134],"our":[135],"multi-level":[136],"evaluation":[137],"framework,":[138],"XPU-Perf,":[139],"includes":[141],"benchmarks":[142],"spanning":[143],"instructions,":[144],"operators,":[145],"models.":[147],"This":[148],"framework":[149],"improves":[150],"interpretability":[151],"trustworthiness,":[153],"helps":[155],"promising":[156],"new":[157],"architectures":[159],"gain":[160],"wider":[161],"adoption":[162],"development.":[164],"Finally,":[165],"comparative":[168],"results":[169],"four":[171],"typical":[172],"accelerators,":[173],"summarize":[174],"their":[175],"shortcomings":[176],"challenges,":[178],"conduct":[179],"in-depth":[180],"analysis,":[181],"highlight":[183],"numerous":[184],"architectural":[185],"scheduling":[187],"observed.":[192]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-03-05T00:00:00"}