{"id":"https://openalex.org/W7134930092","doi":"https://doi.org/10.1145/3779212.3790135","title":"Bullet: Boosting GPU Utilization for LLM Serving via Dynamic Spatial-Temporal Orchestration","display_name":"Bullet: Boosting GPU Utilization for LLM Serving via Dynamic Spatial-Temporal Orchestration","publication_year":2026,"publication_date":"2026-03-10","ids":{"openalex":"https://openalex.org/W7134930092","doi":"https://doi.org/10.1145/3779212.3790135"},"language":null,"primary_location":{"id":"doi:10.1145/3779212.3790135","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3779212.3790135","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3779212.3790135","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128746625","display_name":"Zejia Lin","orcid":null},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zejia Lin","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0002-7205-4062","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100572697","display_name":"Hongxin Xu","orcid":null},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hongxin Xu","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0009-0001-9054-396X","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Guanyi Chen","orcid":"https://orcid.org/0000-0001-6850-4385"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guanyi Chen","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-6850-4385","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Zhiguang Chen","orcid":"https://orcid.org/0000-0002-9318-5715"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhiguang Chen","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0002-9318-5715","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128769633","display_name":"Yutong Lu","orcid":null},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yutong Lu","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-5315-3375","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5128786333","display_name":"Xianwei Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xianwei Zhang","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0003-3507-4299","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157773358"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"290","last_page":"306"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.45680001378059387,"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":0.45680001378059387,"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/T14347","display_name":"Big Data and Digital Economy","score":0.14589999616146088,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.08709999918937683,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/scheduling","display_name":"Scheduling (production processes)","score":0.5976999998092651},{"id":"https://openalex.org/keywords/boosting","display_name":"Boosting (machine learning)","score":0.548799991607666},{"id":"https://openalex.org/keywords/provisioning","display_name":"Provisioning","score":0.5166000127792358},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.4652999937534332},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.44440001249313354},{"id":"https://openalex.org/keywords/workflow","display_name":"Workflow","score":0.3345000147819519},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.3237999975681305},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.30640000104904175}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8621000051498413},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5976999998092651},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.548799991607666},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.5432000160217285},{"id":"https://openalex.org/C172191483","wikidata":"https://www.wikidata.org/wiki/Q1071806","display_name":"Provisioning","level":2,"score":0.5166000127792358},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.4652999937534332},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.44440001249313354},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.35199999809265137},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.3345000147819519},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.3237999975681305},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.3230000138282776},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.30640000104904175},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3050999939441681},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.2980000078678131},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.29179999232292175},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.2870999872684479},{"id":"https://openalex.org/C26713055","wikidata":"https://www.wikidata.org/wiki/Q245962","display_name":"Implementation","level":2,"score":0.27320000529289246},{"id":"https://openalex.org/C28034677","wikidata":"https://www.wikidata.org/wiki/Q17092530","display_name":"Interleaving","level":2,"score":0.27309998869895935},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.27300000190734863},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.27230000495910645},{"id":"https://openalex.org/C200789330","wikidata":"https://www.wikidata.org/wiki/Q7000834","display_name":"Network Functions Virtualization","level":3,"score":0.26489999890327454},{"id":"https://openalex.org/C2776190703","wikidata":"https://www.wikidata.org/wiki/Q488148","display_name":"Slicing","level":2,"score":0.26030001044273376},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.2572000026702881},{"id":"https://openalex.org/C67953723","wikidata":"https://www.wikidata.org/wiki/Q192525","display_name":"Workstation","level":2,"score":0.2549999952316284},{"id":"https://openalex.org/C142355369","wikidata":"https://www.wikidata.org/wiki/Q7698919","display_name":"Temporal isolation among virtual machines","level":4,"score":0.2549999952316284},{"id":"https://openalex.org/C513985346","wikidata":"https://www.wikidata.org/wiki/Q270471","display_name":"Virtualization","level":3,"score":0.2522999942302704},{"id":"https://openalex.org/C25344961","wikidata":"https://www.wikidata.org/wiki/Q192726","display_name":"Virtual machine","level":2,"score":0.2515000104904175}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3779212.3790135","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3779212.3790135","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3779212.3790135","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3779212.3790135","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":23,"referenced_works":["https://openalex.org/W2323909431","https://openalex.org/W2522765117","https://openalex.org/W2903485826","https://openalex.org/W2963926728","https://openalex.org/W3133730715","https://openalex.org/W4220728567","https://openalex.org/W4242509916","https://openalex.org/W4321446237","https://openalex.org/W4360831842","https://openalex.org/W4381885923","https://openalex.org/W4386436407","https://openalex.org/W4387321091","https://openalex.org/W4390097634","https://openalex.org/W4394944658","https://openalex.org/W4403883066","https://openalex.org/W4407196897","https://openalex.org/W4407217670","https://openalex.org/W4408049809","https://openalex.org/W4408867487","https://openalex.org/W4409657546","https://openalex.org/W4416028086","https://openalex.org/W4416203837","https://openalex.org/W6910470332"],"related_works":[],"abstract_inverted_index":{"Modern":[0],"large":[1],"language":[2],"model":[3],"(LLM)":[4],"serving":[5],"systems":[6],"confront":[7],"inefficient":[8,42],"GPU":[9,52,127,144],"utilization":[10,71,145],"due":[11,72],"to":[12,28,73,164],"the":[13,64,97],"fundamental":[14],"mismatch":[15],"between":[16],"compute-intensive":[17],"prefill":[18,65,120],"phase":[19,66,113],"and":[20,76,79,92,121,138],"memory-bound":[21],"decode":[22,122],"phase.":[23],"While":[24],"current":[25],"practices":[26],"attempt":[27],"address":[29],"this":[30],"by":[31],"organizing":[32],"these":[33,109],"phases":[34],"into":[35],"hybrid":[36,81],"batches,":[37],"such":[38],"solutions":[39],"create":[40],"an":[41],"tradeoff":[43],"that":[44,107,156],"sacrifices":[45],"either":[46],"throughput":[47,161],"or":[48],"latency,":[49],"leaving":[50],"substantial":[51],"resources":[53,91,128],"underutilized.":[54],"For":[55],"this,":[56],"we":[57,99],"identify":[58],"two":[59],"key":[60],"root":[61],"causes:":[62],"1)":[63],"suffers":[67],"from":[68],"suboptimal":[69],"compute":[70,90],"wave":[74],"quantization":[75],"attention":[77],"bottlenecks,":[78],"2)":[80],"batching":[82],"disproportionately":[83],"prioritizes":[84],"latency":[85,148,171],"over":[86,166],"throughput,":[87],"wasting":[88],"both":[89],"memory":[93],"bandwidth.":[94],"To":[95],"mitigate":[96],"issues,":[98],"present":[100],"Bullet,":[101],"a":[102],"novel":[103],"spatial-temporal":[104],"orchestration":[105],"system":[106],"eliminates":[108],"inefficiencies":[110],"through":[111],"fine-grained":[112],"coordination.":[114],"Bullet":[115,142,157],"enables":[116],"concurrent":[117],"execution":[118],"of":[119],"requests,":[123],"while":[124,168],"dynamically":[125],"provisioning":[126],"based":[129],"on":[130,152],"real-time":[131],"performance":[132],"modeling.":[133],"By":[134],"integrating":[135],"SLO-aware":[136],"scheduling":[137],"adaptive":[139],"resource":[140],"allocation,":[141],"maximizes":[143],"without":[146],"compromising":[147],"targets.":[149],"Experimental":[150],"evaluations":[151],"real-world":[153],"workloads":[154],"demonstrate":[155],"delivers":[158],"1.26\u00d7":[159],"average":[160],"gains":[162],"(up":[163],"1.55\u00d7)":[165],"state-of-the-arts,":[167],"consistently":[169],"meeting":[170],"constraints.":[172]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2026-03-12T00:00:00"}