{"id":"https://openalex.org/W4413014945","doi":"https://doi.org/10.1145/3759441.3759444","title":"Efficient LLM Inference via Chunked Prefills","display_name":"Efficient LLM Inference via Chunked Prefills","publication_year":2025,"publication_date":"2025-08-04","ids":{"openalex":"https://openalex.org/W4413014945","doi":"https://doi.org/10.1145/3759441.3759444"},"language":"en","primary_location":{"id":"doi:10.1145/3759441.3759444","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3759441.3759444","pdf_url":null,"source":{"id":"https://openalex.org/S50071195","display_name":"ACM SIGOPS Operating Systems Review","issn_l":"0163-5980","issn":["0163-5980","1943-586X"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM SIGOPS Operating Systems Review","raw_type":"journal-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":null,"display_name":"Arney Agrawal","orcid":null},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Arney Agrawal","raw_affiliation_strings":["Georgia Institute of Technology, GA, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology, GA, USA","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5094077405","display_name":"Nitin Kedia","orcid":null},"institutions":[{"id":"https://openalex.org/I4210124949","display_name":"Microsoft Research (India)","ror":"https://ror.org/02w7f3w92","country_code":"IN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210124949"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Nitin Kedia","raw_affiliation_strings":["Microsoft Research India, India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research India, India","institution_ids":["https://openalex.org/I4210124949"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060296906","display_name":"Ashish Panwar","orcid":"https://orcid.org/0009-0007-0621-4412"},"institutions":[{"id":"https://openalex.org/I4210124949","display_name":"Microsoft Research (India)","ror":"https://ror.org/02w7f3w92","country_code":"IN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210124949"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Ashish Panwar","raw_affiliation_strings":["Microsoft Research India, India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research India, India","institution_ids":["https://openalex.org/I4210124949"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005136390","display_name":"Jayashree Mohan","orcid":"https://orcid.org/0009-0005-5260-3203"},"institutions":[{"id":"https://openalex.org/I4210124949","display_name":"Microsoft Research (India)","ror":"https://ror.org/02w7f3w92","country_code":"IN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210124949"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Jayashree Mohan","raw_affiliation_strings":["Microsoft Research India, India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research India, India","institution_ids":["https://openalex.org/I4210124949"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060166632","display_name":"Nipun Kwatra","orcid":"https://orcid.org/0000-0003-0354-6204"},"institutions":[{"id":"https://openalex.org/I4210124949","display_name":"Microsoft Research (India)","ror":"https://ror.org/02w7f3w92","country_code":"IN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210124949"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Nipun Kwatra","raw_affiliation_strings":["Microsoft Research India, India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research India, India","institution_ids":["https://openalex.org/I4210124949"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028750260","display_name":"Bhargav S. Gulavani","orcid":null},"institutions":[{"id":"https://openalex.org/I4210124949","display_name":"Microsoft Research (India)","ror":"https://ror.org/02w7f3w92","country_code":"IN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210124949"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Bhargav S. Gulavani","raw_affiliation_strings":["Microsoft Research India, India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research India, India","institution_ids":["https://openalex.org/I4210124949"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048451114","display_name":"Alexey Tumanov","orcid":"https://orcid.org/0009-0005-7862-1477"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Alexey Tumanov","raw_affiliation_strings":["Georgia Institute of Technology, GA, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology, GA, USA","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5015355530","display_name":"Ramachandran Ramjee","orcid":"https://orcid.org/0000-0003-0007-6040"},"institutions":[{"id":"https://openalex.org/I4210124949","display_name":"Microsoft Research (India)","ror":"https://ror.org/02w7f3w92","country_code":"IN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210124949"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Ramachandran Ramjee","raw_affiliation_strings":["Microsoft Research India, India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research India, India","institution_ids":["https://openalex.org/I4210124949"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":8,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I130701444"],"apc_list":null,"apc_paid":null,"fwci":4.2721,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.9435703,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":91,"max":98},"biblio":{"volume":"59","issue":"1","first_page":"9","last_page":"16"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9997000098228455,"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":0.9997000098228455,"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":0.9957000017166138,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9839000105857849,"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/inference","display_name":"Inference","score":0.622803270816803},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.4204910695552826},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.40413743257522583},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.19168692827224731}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.622803270816803},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4204910695552826},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.40413743257522583},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.19168692827224731}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3759441.3759444","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3759441.3759444","pdf_url":null,"source":{"id":"https://openalex.org/S50071195","display_name":"ACM SIGOPS Operating Systems Review","issn_l":"0163-5980","issn":["0163-5980","1943-586X"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM SIGOPS Operating Systems Review","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":20,"referenced_works":["https://openalex.org/W2612690371","https://openalex.org/W2901299405","https://openalex.org/W2963926728","https://openalex.org/W4224308101","https://openalex.org/W4288028629","https://openalex.org/W4377865306","https://openalex.org/W4384918448","https://openalex.org/W4386395487","https://openalex.org/W4386942223","https://openalex.org/W4387321091","https://openalex.org/W4389157038","https://openalex.org/W4389261323","https://openalex.org/W4392489911","https://openalex.org/W4396822035","https://openalex.org/W4401211704","https://openalex.org/W6739901393","https://openalex.org/W6772383348","https://openalex.org/W6778883912","https://openalex.org/W6838461927","https://openalex.org/W6858453470"],"related_works":["https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W4391375266","https://openalex.org/W1979597421","https://openalex.org/W2007980826","https://openalex.org/W2061531152","https://openalex.org/W3002753104","https://openalex.org/W2077600819","https://openalex.org/W2142036596","https://openalex.org/W2072657027"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Model":[2],"(LLM)":[3],"inference":[4],"serving":[5,124],"faces":[6],"a":[7,37],"fundamental":[8],"challenge":[9],"due":[10],"to":[11,36],"the":[12,33,79,112,117],"distinct":[13],"characteristics":[14],"of":[15,81],"its":[16,85],"two":[17],"phases:":[18],"compute-intensive":[19],"pre":[20],"fill":[21],"and":[22,43,100,133,142],"memory-intensive":[23],"decode.":[24],"Existing":[25],"scheduling":[26],"strategies":[27],"often":[28],"prioritize":[29],"one":[30],"phase":[31],"over":[32],"other,":[34],"leading":[35],"difficult":[38],"tradeoff":[39],"between":[40],"system":[41],"throughput":[42,49],"request":[44],"latency.":[45],"Prefill-prioritizing":[46],"schedulers":[47,64],"improve":[48],"but":[50,68],"introduce":[51],"significant":[52],"latency":[53,67,128],"jitter":[54],"(generation":[55],"stalls)":[56],"by":[57],"interfering":[58],"with":[59,103],"ongoing":[60],"decodes.":[61],"Conversely,":[62],"decode-prioritizing":[63],"maintain":[65],"low":[66,74],"underutilize":[69],"GPU":[70],"resources,":[71],"resulting":[72],"in":[73,87,116,137],"throughput.":[75],"This":[76,120],"paper":[77],"revisits":[78],"technique":[80],"chunked":[82],"prefills,":[83],"demonstrating":[84],"efficacy":[86],"mitigating":[88],"this":[89],"tradeoff.":[90],"By":[91],"splitting":[92],"large":[93],"prefill":[94],"computations":[95],"into":[96],"smaller,":[97],"manageable":[98],"chunks":[99],"interleaving":[101],"them":[102],"decode":[104,118],"operations":[105],"using":[106],"stall-free":[107],"batching,":[108],"we":[109],"can":[110],"leverage":[111],"compute":[113],"slack":[114],"inherent":[115],"phase.":[119],"approach":[121],"significantly":[122],"improves":[123],"capacity":[125],"under":[126],"strict":[127],"constraints,":[129],"minimizes":[130],"generation":[131],"stalls,":[132],"reduces":[134],"pipeline":[135],"bubbles":[136],"distributed":[138],"deployments,":[139],"enabling":[140],"efficient":[141],"responsive":[143],"inference.":[144]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":1}],"updated_date":"2026-04-30T09:15:22.047038","created_date":"2025-10-10T00:00:00"}