{"id":"https://openalex.org/W4415968362","doi":"https://doi.org/10.1109/lca.2025.3630094","title":"Disaggregated Speculative Decoding for Carbon-Efficient LLM Serving","display_name":"Disaggregated Speculative Decoding for Carbon-Efficient LLM Serving","publication_year":2025,"publication_date":"2025-07-01","ids":{"openalex":"https://openalex.org/W4415968362","doi":"https://doi.org/10.1109/lca.2025.3630094"},"language":null,"primary_location":{"id":"doi:10.1109/lca.2025.3630094","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2025.3630094","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Computer Architecture Letters","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":"https://openalex.org/A5029811967","display_name":"Tongtong Shi","orcid":null},"institutions":[{"id":"https://openalex.org/I219193219","display_name":"Purdue University West Lafayette","ror":"https://ror.org/02dqehb95","country_code":"US","type":"education","lineage":["https://openalex.org/I219193219"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Tianyao Shi","raw_affiliation_strings":["Purdue University, West Lafayette, IN, USA","Purdue University, West Lafayette, USA"],"affiliations":[{"raw_affiliation_string":"Purdue University, West Lafayette, IN, USA","institution_ids":["https://openalex.org/I219193219"]},{"raw_affiliation_string":"Purdue University, West Lafayette, USA","institution_ids":["https://openalex.org/I219193219"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5105683309","display_name":"Yanran Wu","orcid":null},"institutions":[{"id":"https://openalex.org/I219193219","display_name":"Purdue University West Lafayette","ror":"https://ror.org/02dqehb95","country_code":"US","type":"education","lineage":["https://openalex.org/I219193219"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yanran Wu","raw_affiliation_strings":["Purdue University, West Lafayette, IN, USA","Purdue University, West Lafayette, USA"],"affiliations":[{"raw_affiliation_string":"Purdue University, West Lafayette, IN, USA","institution_ids":["https://openalex.org/I219193219"]},{"raw_affiliation_string":"Purdue University, West Lafayette, USA","institution_ids":["https://openalex.org/I219193219"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5038231705","display_name":"Sihang Liu","orcid":"https://orcid.org/0000-0001-9706-6177"},"institutions":[{"id":"https://openalex.org/I151746483","display_name":"University of Waterloo","ror":"https://ror.org/01aff2v68","country_code":"CA","type":"education","lineage":["https://openalex.org/I151746483"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Sihang Liu","raw_affiliation_strings":["University of Waterloo, Waterloo, ON, Canada","University of Waterloo, Waterloo, Canada"],"affiliations":[{"raw_affiliation_string":"University of Waterloo, Waterloo, ON, Canada","institution_ids":["https://openalex.org/I151746483"]},{"raw_affiliation_string":"University of Waterloo, Waterloo, Canada","institution_ids":["https://openalex.org/I151746483"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5043391532","display_name":"Yi Ding","orcid":"https://orcid.org/0000-0003-0127-4487"},"institutions":[{"id":"https://openalex.org/I219193219","display_name":"Purdue University West Lafayette","ror":"https://ror.org/02dqehb95","country_code":"US","type":"education","lineage":["https://openalex.org/I219193219"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yi Ding","raw_affiliation_strings":["Purdue University, West Lafayette, IN, USA","Purdue University, West Lafayette, USA"],"affiliations":[{"raw_affiliation_string":"Purdue University, West Lafayette, IN, USA","institution_ids":["https://openalex.org/I219193219"]},{"raw_affiliation_string":"Purdue University, West Lafayette, USA","institution_ids":["https://openalex.org/I219193219"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5029811967"],"corresponding_institution_ids":["https://openalex.org/I219193219"],"apc_list":null,"apc_paid":null,"fwci":1.2892,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.85704077,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":99},"biblio":{"volume":"24","issue":"2","first_page":"369","last_page":"372"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.47209998965263367,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.47209998965263367,"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"}},{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.11630000174045563,"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":0.06589999794960022,"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/decoding-methods","display_name":"Decoding methods","score":0.8009999990463257},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.692799985408783},{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.5242999792098999},{"id":"https://openalex.org/keywords/workload","display_name":"Workload","score":0.5091000199317932},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.4952999949455261},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.4494999945163727},{"id":"https://openalex.org/keywords/sequential-decoding","display_name":"Sequential decoding","score":0.4174000024795532},{"id":"https://openalex.org/keywords/decoupling","display_name":"Decoupling (probability)","score":0.36970001459121704}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8598999977111816},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.8009999990463257},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.692799985408783},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.5242999792098999},{"id":"https://openalex.org/C2778476105","wikidata":"https://www.wikidata.org/wiki/Q628539","display_name":"Workload","level":2,"score":0.5091000199317932},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.4952999949455261},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.4494999945163727},{"id":"https://openalex.org/C193969084","wikidata":"https://www.wikidata.org/wiki/Q7452500","display_name":"Sequential decoding","level":4,"score":0.4174000024795532},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.4165000021457672},{"id":"https://openalex.org/C205606062","wikidata":"https://www.wikidata.org/wiki/Q5249645","display_name":"Decoupling (probability)","level":2,"score":0.36970001459121704},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.35019999742507935},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.34439998865127563},{"id":"https://openalex.org/C204397858","wikidata":"https://www.wikidata.org/wiki/Q4437907","display_name":"List decoding","level":5,"score":0.3416999876499176},{"id":"https://openalex.org/C46637626","wikidata":"https://www.wikidata.org/wiki/Q6693015","display_name":"Low latency (capital markets)","level":2,"score":0.3409999907016754},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.33889999985694885},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.3325999975204468},{"id":"https://openalex.org/C162262903","wikidata":"https://www.wikidata.org/wiki/Q343527","display_name":"Allocator","level":2,"score":0.30390000343322754},{"id":"https://openalex.org/C141331961","wikidata":"https://www.wikidata.org/wiki/Q2164465","display_name":"Speculative execution","level":2,"score":0.29670000076293945},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.2872999906539917},{"id":"https://openalex.org/C107598950","wikidata":"https://www.wikidata.org/wiki/Q259864","display_name":"Microarchitecture","level":2,"score":0.2791000008583069},{"id":"https://openalex.org/C206891323","wikidata":"https://www.wikidata.org/wiki/Q178655","display_name":"Moore's law","level":2,"score":0.2766999900341034},{"id":"https://openalex.org/C153180980","wikidata":"https://www.wikidata.org/wiki/Q19776675","display_name":"Commit","level":2,"score":0.26660001277923584},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.2614000141620636}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/lca.2025.3630094","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2025.3630094","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Computer Architecture Letters","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":6,"referenced_works":["https://openalex.org/W4281971810","https://openalex.org/W4387321091","https://openalex.org/W4401211704","https://openalex.org/W4409248734","https://openalex.org/W4409537509","https://openalex.org/W4412945375"],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2,68,74],"(LLMs)":[3],"are":[4],"increasingly":[5],"deployed":[6],"in":[7],"practice":[8],"but":[9],"incur":[10],"significant":[11],"computational":[12],"costs":[13],"and":[14,23,34,72,81,94,101,106],"environmental":[15],"impacts.":[16],"Disaggregated":[17],"serving":[18],"techniques,":[19],"particularly":[20],"decoupling":[21],"prefill":[22],"decoding":[24,59,64],"(DPD)":[25],"across":[26],"GPUs,":[27,77],"have":[28],"been":[29],"introduced":[30],"to":[31,46,69,75,122],"improve":[32],"performance":[33],"reduce":[35],"carbon":[36,118],"emissions.":[37],"However,":[38],"DPD":[39,100],"suffers":[40],"from":[41],"high":[42],"bandwidth":[43],"overhead":[44],"due":[45],"frequent":[47],"large":[48],"KV":[49],"cache":[50],"transfers.":[51,84],"To":[52],"address":[53],"this,":[54],"we":[55,89],"present":[56],"disaggregated":[57],"speculative":[58,63],"(DSD),":[60],"which":[61],"leverages":[62],"by":[65,120],"assigning":[66],"draft":[67],"older":[70],"GPUs":[71],"target":[73],"newer":[76],"requiring":[78],"only":[79],"token":[80],"probability":[82],"distribution":[83],"Building":[85],"on":[86],"this":[87],"insight,":[88],"introduce":[90],"GreenLLM,":[91],"an":[92],"SLO-":[93],"bandwidth-":[95],"aware":[96],"framework":[97],"that":[98],"unifies":[99],"DSD,":[102],"profiles":[103],"workload":[104],"characteristics,":[105],"dynamically":[107],"selects":[108],"the":[109],"most":[110],"carbon-efficient":[111],"configuration.":[112],"Across":[113],"diverse":[114],"benchmarks,":[115],"GreenLLM":[116],"reduces":[117],"emissions":[119],"up":[121],"40.6%":[123],"while":[124],"meeting":[125],"latency":[126],"SLOs.":[127]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-03-27T14:29:43.386196","created_date":"2025-11-06T00:00:00"}