{"id":"https://openalex.org/W4404057367","doi":"https://doi.org/10.1109/socc62300.2024.10737844","title":"Energy Cost Modelling for Optimizing Large Language Model Inference on Hardware Accelerators","display_name":"Energy Cost Modelling for Optimizing Large Language Model Inference on Hardware Accelerators","publication_year":2024,"publication_date":"2024-09-16","ids":{"openalex":"https://openalex.org/W4404057367","doi":"https://doi.org/10.1109/socc62300.2024.10737844"},"language":"en","primary_location":{"id":"doi:10.1109/socc62300.2024.10737844","is_oa":false,"landing_page_url":"https://doi.org/10.1109/socc62300.2024.10737844","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 IEEE 37th International System-on-Chip Conference (SOCC)","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/A5065013150","display_name":"Robin Geens","orcid":null},"institutions":[{"id":"https://openalex.org/I99464096","display_name":"KU Leuven","ror":"https://ror.org/05f950310","country_code":"BE","type":"education","lineage":["https://openalex.org/I99464096"]}],"countries":["BE"],"is_corresponding":true,"raw_author_name":"Robin Geens","raw_affiliation_strings":["KU Leuven,MICAS,Belgium"],"affiliations":[{"raw_affiliation_string":"KU Leuven,MICAS,Belgium","institution_ids":["https://openalex.org/I99464096"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5043501540","display_name":"Man Shi","orcid":"https://orcid.org/0000-0003-1675-6235"},"institutions":[{"id":"https://openalex.org/I99464096","display_name":"KU Leuven","ror":"https://ror.org/05f950310","country_code":"BE","type":"education","lineage":["https://openalex.org/I99464096"]}],"countries":["BE"],"is_corresponding":false,"raw_author_name":"Man Shi","raw_affiliation_strings":["KU Leuven,MICAS,Belgium"],"affiliations":[{"raw_affiliation_string":"KU Leuven,MICAS,Belgium","institution_ids":["https://openalex.org/I99464096"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5046786058","display_name":"Arne Symons","orcid":"https://orcid.org/0000-0002-4595-9384"},"institutions":[{"id":"https://openalex.org/I99464096","display_name":"KU Leuven","ror":"https://ror.org/05f950310","country_code":"BE","type":"education","lineage":["https://openalex.org/I99464096"]}],"countries":["BE"],"is_corresponding":false,"raw_author_name":"Arne Symons","raw_affiliation_strings":["KU Leuven,MICAS,Belgium"],"affiliations":[{"raw_affiliation_string":"KU Leuven,MICAS,Belgium","institution_ids":["https://openalex.org/I99464096"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066888122","display_name":"Chao Fang","orcid":"https://orcid.org/0000-0003-3430-1189"},"institutions":[{"id":"https://openalex.org/I99464096","display_name":"KU Leuven","ror":"https://ror.org/05f950310","country_code":"BE","type":"education","lineage":["https://openalex.org/I99464096"]}],"countries":["BE"],"is_corresponding":false,"raw_author_name":"Chao Fang","raw_affiliation_strings":["KU Leuven,MICAS,Belgium"],"affiliations":[{"raw_affiliation_string":"KU Leuven,MICAS,Belgium","institution_ids":["https://openalex.org/I99464096"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5012150553","display_name":"Marian Verhelst","orcid":"https://orcid.org/0000-0003-3495-9263"},"institutions":[{"id":"https://openalex.org/I99464096","display_name":"KU Leuven","ror":"https://ror.org/05f950310","country_code":"BE","type":"education","lineage":["https://openalex.org/I99464096"]}],"countries":["BE"],"is_corresponding":false,"raw_author_name":"Marian Verhelst","raw_affiliation_strings":["KU Leuven,MICAS,Belgium"],"affiliations":[{"raw_affiliation_string":"KU Leuven,MICAS,Belgium","institution_ids":["https://openalex.org/I99464096"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5065013150"],"corresponding_institution_ids":["https://openalex.org/I99464096"],"apc_list":null,"apc_paid":null,"fwci":2.8726,"has_fulltext":false,"cited_by_count":8,"citation_normalized_percentile":{"value":0.92112723,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"6"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9373000264167786,"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/T10028","display_name":"Topic Modeling","score":0.9373000264167786,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.9293000102043152,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7556871771812439},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6058816909790039},{"id":"https://openalex.org/keywords/energy","display_name":"Energy (signal processing)","score":0.4724994897842407},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.3478589653968811},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.3425552248954773},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.14885881543159485}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7556871771812439},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6058816909790039},{"id":"https://openalex.org/C186370098","wikidata":"https://www.wikidata.org/wiki/Q442787","display_name":"Energy (signal processing)","level":2,"score":0.4724994897842407},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3478589653968811},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.3425552248954773},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.14885881543159485},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/socc62300.2024.10737844","is_oa":false,"landing_page_url":"https://doi.org/10.1109/socc62300.2024.10737844","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 IEEE 37th International System-on-Chip Conference (SOCC)","raw_type":"proceedings-article"},{"id":"pmh:oai:lirias2repo.kuleuven.be:20.500.12942/751959","is_oa":false,"landing_page_url":"https://lirias.kuleuven.be/handle/20.500.12942/751959","pdf_url":null,"source":{"id":"https://openalex.org/S4306401954","display_name":"Lirias (KU Leuven)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I99464096","host_organization_name":"KU Leuven","host_organization_lineage":["https://openalex.org/I99464096"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"2024 IEEE 37th International System-on-Chip Conference (SOCC), Dresden, Germany, 16-19 September 2024","raw_type":"info:eu-repo/semantics/publishedVersion"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7","score":0.8399999737739563}],"awards":[],"funders":[{"id":"https://openalex.org/F4320322308","display_name":"KU Leuven","ror":"https://ror.org/05f950310"},{"id":"https://openalex.org/F4320334678","display_name":"European Research Council","ror":"https://ror.org/0472cxd90"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":28,"referenced_works":["https://openalex.org/W1999085092","https://openalex.org/W2087417231","https://openalex.org/W2973727699","https://openalex.org/W3132942233","https://openalex.org/W3189877953","https://openalex.org/W4291653336","https://openalex.org/W4295312788","https://openalex.org/W4308083526","https://openalex.org/W4366341968","https://openalex.org/W4380881077","https://openalex.org/W4384918448","https://openalex.org/W4389470964","https://openalex.org/W4391047038","https://openalex.org/W4392222465","https://openalex.org/W4392271821","https://openalex.org/W4393406875","https://openalex.org/W4393407021","https://openalex.org/W4393578753","https://openalex.org/W4396815596","https://openalex.org/W4399794259","https://openalex.org/W4401211619","https://openalex.org/W6767997687","https://openalex.org/W6853048723","https://openalex.org/W6856696905","https://openalex.org/W6860264091","https://openalex.org/W6862033441","https://openalex.org/W6866033557","https://openalex.org/W6869801542"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"The":[0],"rise":[1],"of":[2,97,103,107,111,132,151,224],"Large":[3],"Language":[4],"Models":[5],"(LLMs)":[6],"has":[7],"significantly":[8,142],"escalated":[9],"the":[10,42,73,104,133,177,185,190,199,207,213,220,225],"demand":[11],"for":[12,48,76,184,189],"efficient":[13],"LLM":[14,68,87,123],"inference,":[15,170],"primarily":[16],"fulfilled":[17],"through":[18,100],"cloud-based":[19],"GPU":[20],"computing.":[21],"This":[22],"approach,":[23],"while":[24],"effective,":[25],"is":[26,180,228,242],"associated":[27],"with":[28],"high":[29],"energy":[30,98,166,191,200,222],"consumption":[31],"resulting":[32],"in":[33,168,176,217],"large":[34],"operating":[35],"expenses":[36],"and":[37,63,139,205,237],"considerable":[38],"carbon":[39],"footprints.":[40],"In":[41],"meantime,":[43],"growing":[44],"privacy":[45],"concerns":[46],"advocate":[47],"inference":[49,69,78],"on":[50,89,113],"edge":[51,64,218],"devices,":[52],"which":[53],"are":[54],"constrained":[55],"by":[56,79,202],"a":[57,108,120,148,154],"limited":[58],"battery":[59],"capacity.":[60],"Both":[61],"cloud":[62],"scenarios":[65],"necessitate":[66],"energy-efficient":[67,77],"strategies.This":[70],"paper":[71],"addresses":[72],"urgent":[74],"need":[75],"proposing":[80],"an":[81],"open-source":[82,244],"framework":[83,93,138,241],"designed":[84],"to":[85,125,141,163,212,233],"model":[86],"workloads":[88],"dedicated":[90],"accelerators.":[91],"Our":[92,240],"facilitates":[94],"early":[95],"identification":[96],"bottlenecks":[99,167],"rapid":[101],"modeling":[102],"execution":[105],"efficiency":[106],"wide":[109],"range":[110],"LLMs":[112],"diverse":[114],"hardware":[115,156],"architectures.":[116],"Key":[117],"innovations":[118],"include":[119],"PyTorch-based":[121],"generalized":[122],"template":[124],"easily":[126],"generate":[127],"custom":[128],"workload":[129],"graphs,":[130],"extensions":[131],"ZigZag":[134],"design":[135],"space":[136],"exploration":[137],"techniques":[140],"speed":[143],"up":[144],"simulation":[145],"time":[146],"at":[147,245],"negligible":[149],"loss":[150],"accuracy.":[152],"Using":[153],"representative":[155],"architecture,":[157],"we":[158],"conduct":[159],"three":[160],"case":[161],"studies":[162],"reveal":[164],"critical":[165],"Llama2-7B":[169],"revealing":[171],"that":[172],"1)":[173],"memory-bound":[174],"computing":[175],"decode":[178,238],"stage":[179,227],"detrimental":[181],"not":[182],"only":[183],"latency,":[186],"but":[187],"also":[188],"cost;":[192],"2)":[193],"aggressive":[194],"weight-only":[195],"quantization":[196],"can":[197],"reduce":[198],"cost":[201,223],"4.6":[203],"\u00d7":[204],"shift":[206],"bottleneck":[208],"from":[209],"weight":[210],"fetching":[211],"attention":[214],"mechanism;":[215],"3)":[216],"scenarios,":[219],"relative":[221],"prefill":[226,236],"more":[229],"significant,":[230],"encouraging":[231],"efforts":[232],"optimize":[234],"both":[235],"stage.":[239],"available":[243],"github.com/KULeuven-MICAS/zigzag-llm.":[246]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":5}],"updated_date":"2026-03-16T09:10:04.655348","created_date":"2025-10-10T00:00:00"}