{"id":"https://openalex.org/W7155413969","doi":"https://doi.org/10.1145/3777884.3797011","title":"SweetSpot: An Analytical Model for Predicting Energy Efficiency of LLM Inference","display_name":"SweetSpot: An Analytical Model for Predicting Energy Efficiency of LLM Inference","publication_year":2026,"publication_date":"2026-04-23","ids":{"openalex":"https://openalex.org/W7155413969","doi":"https://doi.org/10.1145/3777884.3797011"},"language":null,"primary_location":{"id":"doi:10.1145/3777884.3797011","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3777884.3797011","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 17th ACM/SPEC International Conference on Performance Engineering","raw_type":"proceedings-article"},"type":"article","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3777884.3797011","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5120514218","display_name":"Hiari Pizzini Cavagna","orcid":"https://orcid.org/0009-0005-2768-0418"},"institutions":[{"id":"https://openalex.org/I9360294","display_name":"University of Bologna","ror":"https://ror.org/01111rn36","country_code":"IT","type":"education","lineage":["https://openalex.org/I9360294"]}],"countries":["IT"],"is_corresponding":true,"raw_author_name":"Hiari Pizzini Cavagna","raw_affiliation_strings":["University of Bologna, Bologna, Italy"],"raw_orcid":"https://orcid.org/0009-0005-2768-0418","affiliations":[{"raw_affiliation_string":"University of Bologna, Bologna, Italy","institution_ids":["https://openalex.org/I9360294"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111397583","display_name":"Andrea Proia","orcid":null},"institutions":[{"id":"https://openalex.org/I9360294","display_name":"University of Bologna","ror":"https://ror.org/01111rn36","country_code":"IT","type":"education","lineage":["https://openalex.org/I9360294"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Andrea Proia","raw_affiliation_strings":["University of Bologna, Bologna, Italy"],"raw_orcid":"https://orcid.org/0009-0005-6367-762X","affiliations":[{"raw_affiliation_string":"University of Bologna, Bologna, Italy","institution_ids":["https://openalex.org/I9360294"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120661846","display_name":"Giacomo Madella","orcid":null},"institutions":[{"id":"https://openalex.org/I9360294","display_name":"University of Bologna","ror":"https://ror.org/01111rn36","country_code":"IT","type":"education","lineage":["https://openalex.org/I9360294"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Giacomo Madella","raw_affiliation_strings":["University of Bologna, Bologna, Italy"],"raw_orcid":"https://orcid.org/0009-0009-1289-6354","affiliations":[{"raw_affiliation_string":"University of Bologna, Bologna, Italy","institution_ids":["https://openalex.org/I9360294"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134461691","display_name":"Giovanni Battista Esposito","orcid":"https://orcid.org/0009-0002-9595-2962"},"institutions":[{"id":"https://openalex.org/I9360294","display_name":"University of Bologna","ror":"https://ror.org/01111rn36","country_code":"IT","type":"education","lineage":["https://openalex.org/I9360294"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Giovanni Battista Esposito","raw_affiliation_strings":["University of Bologna, Bologna, Italy"],"raw_orcid":"https://orcid.org/0009-0002-9595-2962","affiliations":[{"raw_affiliation_string":"University of Bologna, Bologna, Italy","institution_ids":["https://openalex.org/I9360294"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030560727","display_name":"Francesco Antici","orcid":"https://orcid.org/0000-0002-1125-0588"},"institutions":[{"id":"https://openalex.org/I9360294","display_name":"University of Bologna","ror":"https://ror.org/01111rn36","country_code":"IT","type":"education","lineage":["https://openalex.org/I9360294"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Francesco Antici","raw_affiliation_strings":["University of Bologna, Bologna, Italy"],"raw_orcid":"https://orcid.org/0000-0002-1125-0588","affiliations":[{"raw_affiliation_string":"University of Bologna, Bologna, Italy","institution_ids":["https://openalex.org/I9360294"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5002332287","display_name":"Daniele Cesarini","orcid":"https://orcid.org/0000-0003-1294-372X"},"institutions":[{"id":"https://openalex.org/I4210119786","display_name":"Cineca","ror":"https://ror.org/02f013h18","country_code":"IT","type":"other","lineage":["https://openalex.org/I4210119786"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Daniele Cesarini","raw_affiliation_strings":["Cineca, Casalecchio di Reno, BO, Italy"],"raw_orcid":"https://orcid.org/0000-0003-1294-372X","affiliations":[{"raw_affiliation_string":"Cineca, Casalecchio di Reno, BO, Italy","institution_ids":["https://openalex.org/I4210119786"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033859818","display_name":"Zeynep Kiziltan","orcid":"https://orcid.org/0000-0003-0412-4396"},"institutions":[{"id":"https://openalex.org/I9360294","display_name":"University of Bologna","ror":"https://ror.org/01111rn36","country_code":"IT","type":"education","lineage":["https://openalex.org/I9360294"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Zeynep Kiziltan","raw_affiliation_strings":["University of Bologna, Bologna, Italy"],"raw_orcid":"https://orcid.org/0000-0003-0412-4396","affiliations":[{"raw_affiliation_string":"University of Bologna, Bologna, Italy","institution_ids":["https://openalex.org/I9360294"]}]},{"author_position":"last","author":{"id":null,"display_name":"Andrea Bartolini","orcid":"https://orcid.org/0000-0002-1148-2450"},"institutions":[{"id":"https://openalex.org/I9360294","display_name":"University of Bologna","ror":"https://ror.org/01111rn36","country_code":"IT","type":"education","lineage":["https://openalex.org/I9360294"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Andrea Bartolini","raw_affiliation_strings":["University of Bologna, Bologna, Italy"],"raw_orcid":"https://orcid.org/0000-0002-1148-2450","affiliations":[{"raw_affiliation_string":"University of Bologna, Bologna, Italy","institution_ids":["https://openalex.org/I9360294"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5120514218"],"corresponding_institution_ids":["https://openalex.org/I9360294"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.96196205,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"83","last_page":"95"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12127","display_name":"Software System Performance and Reliability","score":0.19449999928474426,"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/T12127","display_name":"Software System Performance and Reliability","score":0.19449999928474426,"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/T12238","display_name":"Green IT and Sustainability","score":0.16279999911785126,"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"}},{"id":"https://openalex.org/T10101","display_name":"Cloud Computing and Resource Management","score":0.10769999772310257,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/efficient-energy-use","display_name":"Efficient energy use","score":0.6951000094413757},{"id":"https://openalex.org/keywords/energy-consumption","display_name":"Energy consumption","score":0.5446000099182129},{"id":"https://openalex.org/keywords/autoregressive-model","display_name":"Autoregressive model","score":0.5149999856948853},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5133000016212463},{"id":"https://openalex.org/keywords/function","display_name":"Function (biology)","score":0.46239998936653137},{"id":"https://openalex.org/keywords/energy","display_name":"Energy (signal processing)","score":0.438400000333786},{"id":"https://openalex.org/keywords/production","display_name":"Production (economics)","score":0.39250001311302185},{"id":"https://openalex.org/keywords/measure","display_name":"Measure (data warehouse)","score":0.3905999958515167}],"concepts":[{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.6951000094413757},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5450000166893005},{"id":"https://openalex.org/C2780165032","wikidata":"https://www.wikidata.org/wiki/Q16869822","display_name":"Energy consumption","level":2,"score":0.5446000099182129},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.5149999856948853},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5133000016212463},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.46239998936653137},{"id":"https://openalex.org/C186370098","wikidata":"https://www.wikidata.org/wiki/Q442787","display_name":"Energy (signal processing)","level":2,"score":0.438400000333786},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4154999852180481},{"id":"https://openalex.org/C2778348673","wikidata":"https://www.wikidata.org/wiki/Q739302","display_name":"Production (economics)","level":2,"score":0.39250001311302185},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.3905999958515167},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.36739999055862427},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.3546000123023987},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3244999945163727},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.32010000944137573},{"id":"https://openalex.org/C115051666","wikidata":"https://www.wikidata.org/wiki/Q6522493","display_name":"Ranging","level":2,"score":0.29429998993873596},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.2815000116825104},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.27869999408721924},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.27379998564720154},{"id":"https://openalex.org/C423512","wikidata":"https://www.wikidata.org/wiki/Q383973","display_name":"Electricity generation","level":3,"score":0.2718000113964081},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.26499998569488525},{"id":"https://openalex.org/C139876678","wikidata":"https://www.wikidata.org/wiki/Q524004","display_name":"Production function","level":3,"score":0.25600001215934753}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3777884.3797011","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3777884.3797011","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 17th ACM/SPEC International Conference on Performance Engineering","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2602.05695","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2602.05695","pdf_url":"https://arxiv.org/pdf/2602.05695","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.1145/3777884.3797011","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3777884.3797011","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 17th ACM/SPEC International Conference on Performance Engineering","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.9085177779197693}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":8,"referenced_works":["https://openalex.org/W2963910387","https://openalex.org/W3129831491","https://openalex.org/W4408943130","https://openalex.org/W4409047589","https://openalex.org/W4409537475","https://openalex.org/W4412889635","https://openalex.org/W4413019871","https://openalex.org/W7133227460"],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"inference":[4],"is":[5],"central":[6],"to":[7,18,144,163,189],"modern":[8],"AI":[9],"applications,":[10],"dominating":[11],"worldwide":[12],"datacenter":[13],"workloads,":[14],"making":[15],"it":[16],"critical":[17],"predict":[19],"its":[20],"energy":[21,26,59,68,127,186],"footprint.":[22],"Existing":[23],"approaches":[24],"estimate":[25],"consumption":[27,128],"as":[28,114],"a":[29,48,66,115,136,168],"simple":[30],"linear":[31],"function":[32,116],"of":[33,44,65,104,117,139,171],"input":[34,53,118,157],"and":[35,54,58,76,101,119,153,158,166,195],"output":[36,55,120,159],"sequence.":[37],"However,":[38],"by":[39],"analyzing":[40],"the":[41,63,99,105,111],"autoregressive":[42],"structure":[43],"Transformers,":[45],"which":[46,108],"implies":[47],"fundamentally":[49],"non-linear":[50],"relationship":[51],"between":[52],"sequence":[56,178],"lengths":[57,160,179],"consumption,":[60],"we":[61,91,125],"demonstrate":[62],"existence":[64],"generation":[67,197],"minima.":[69],"Peak":[70],"efficiency":[71,80,112,182],"occurs":[72],"with":[73,180],"short-to-moderate":[74],"inputs":[75,85],"medium-length":[77],"outputs,":[78],"while":[79],"drops":[81],"sharply":[82],"for":[83],"long":[84],"or":[86],"very":[87],"short":[88],"outputs.":[89],"Consequently,":[90],"propose":[92],"SweetSpot,":[93],"an":[94],"analytical":[95],"model":[96],"derived":[97],"from":[98,142,161],"computational":[100],"memory-access":[102],"complexity":[103],"Transformer":[106],"architecture,":[107],"accurately":[109],"characterizes":[110],"curve":[113],"lengths.":[121],"To":[122],"assess":[123],"accuracy,":[124],"measure":[126],"using":[129],"TensorRT-LLM":[130],"on":[131],"NVIDIA":[132],"H100":[133],"GPUs":[134],"across":[135],"diverse":[137],"set":[138],"LLMs":[140],"ranging":[141],"1B":[143],"9B":[145],"parameters,":[146],"including":[147],"OPT,":[148],"LLaMA,":[149],"Gemma,":[150],"Falcon,":[151],"Qwen2,":[152],"Granite.":[154],"We":[155],"test":[156],"64":[162],"4096":[164],"tokens":[165],"achieve":[167],"mean":[169],"MAPE":[170],"1.79%.":[172],"Our":[173],"results":[174],"show":[175],"that":[176],"aligning":[177],"these":[181],"''sweet":[183],"spots''":[184],"reduce":[185],"usage,":[187],"up":[188],"33.41x,":[190],"enabling":[191],"informed":[192],"truncation,":[193],"summarization,":[194],"adaptive":[196],"strategies":[198],"in":[199],"production":[200],"systems.":[201]},"counts_by_year":[],"updated_date":"2026-04-26T08:31:28.666265","created_date":"2026-04-24T00:00:00"}