{"id":"https://openalex.org/W7167030862","doi":"https://doi.org/10.1145/3797905.3800532","title":"dLLM-Serve: Bridging the Memory Gap in Diffusion Language Model Serving","display_name":"dLLM-Serve: Bridging the Memory Gap in Diffusion Language Model Serving","publication_year":2026,"publication_date":"2026-07-02","ids":{"openalex":"https://openalex.org/W7167030862","doi":"https://doi.org/10.1145/3797905.3800532"},"language":null,"primary_location":{"id":"doi:10.1145/3797905.3800532","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3797905.3800532","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 40th ACM International Conference on Supercomputing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3797905.3800532","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125658018","display_name":"Jiakun Fan","orcid":null},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jiakun Fan","raw_affiliation_strings":["Department of Computer Science, Virginia Tech, Blacksburg, Virginia, USA"],"raw_orcid":"https://orcid.org/0009-0004-4563-8697","affiliations":[{"raw_affiliation_string":"Department of Computer Science, Virginia Tech, Blacksburg, Virginia, USA","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052998095","display_name":"Y S Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I100135526","display_name":"BGI Group (China)","ror":"https://ror.org/045pn2j94","country_code":"CN","type":"company","lineage":["https://openalex.org/I100135526"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yanglin Zhang","raw_affiliation_strings":["None, SHENZHEN, Guangdong, China"],"raw_orcid":"https://orcid.org/0009-0002-6934-3274","affiliations":[{"raw_affiliation_string":"None, SHENZHEN, Guangdong, China","institution_ids":["https://openalex.org/I100135526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069413313","display_name":"Xiangchen Li","orcid":"https://orcid.org/0000-0002-5293-5348"},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xiangchen Li","raw_affiliation_strings":["Department of Computer Science, Virginia Tech, Blacksburg, Virginia, USA"],"raw_orcid":"https://orcid.org/0000-0002-5293-5348","affiliations":[{"raw_affiliation_string":"Department of Computer Science, Virginia Tech, Blacksburg, Virginia, USA","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5005410613","display_name":"Dimitrios S. Nikolopoulos","orcid":"https://orcid.org/0000-0003-0217-8307"},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dimitrios S. Nikolopoulos","raw_affiliation_strings":["Department of Computer Science, Virginia Tech, Blacksburg, Virginia, USA"],"raw_orcid":"https://orcid.org/0000-0003-0217-8307","affiliations":[{"raw_affiliation_string":"Department of Computer Science, Virginia Tech, Blacksburg, Virginia, USA","institution_ids":["https://openalex.org/I859038795"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.94502884,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"515","last_page":"526"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.4268999993801117,"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":0.4268999993801117,"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/T10028","display_name":"Topic Modeling","score":0.04619999974966049,"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/T14347","display_name":"Big Data and Digital Economy","score":0.041600000113248825,"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/bridging","display_name":"Bridging (networking)","score":0.8230000138282776},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.6327999830245972},{"id":"https://openalex.org/keywords/memory-footprint","display_name":"Memory footprint","score":0.5738000273704529},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.5178999900817871},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.474700003862381},{"id":"https://openalex.org/keywords/markov-chain","display_name":"Markov chain","score":0.3700999915599823},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.3443000018596649},{"id":"https://openalex.org/keywords/execution-model","display_name":"Execution model","score":0.3386000096797943},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.3188999891281128}],"concepts":[{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.8230000138282776},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.777899980545044},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6327999830245972},{"id":"https://openalex.org/C74912251","wikidata":"https://www.wikidata.org/wiki/Q6815727","display_name":"Memory footprint","level":2,"score":0.5738000273704529},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5178999900817871},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.474700003862381},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.4361000061035156},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3944999873638153},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.39340001344680786},{"id":"https://openalex.org/C98763669","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov chain","level":2,"score":0.3700999915599823},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.3443000018596649},{"id":"https://openalex.org/C2776834041","wikidata":"https://www.wikidata.org/wiki/Q25346349","display_name":"Execution model","level":2,"score":0.3386000096797943},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.3188999891281128},{"id":"https://openalex.org/C100776233","wikidata":"https://www.wikidata.org/wiki/Q2532492","display_name":"Bridge (graph theory)","level":2,"score":0.31709998846054077},{"id":"https://openalex.org/C2778770139","wikidata":"https://www.wikidata.org/wiki/Q1966904","display_name":"Solver","level":2,"score":0.31610000133514404},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3158999979496002},{"id":"https://openalex.org/C52146309","wikidata":"https://www.wikidata.org/wiki/Q7431116","display_name":"Schema (genetic algorithms)","level":2,"score":0.31189998984336853},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.30970001220703125},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.2985000014305115},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.29429998993873596},{"id":"https://openalex.org/C159886148","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov process","level":2,"score":0.2904999852180481},{"id":"https://openalex.org/C46637626","wikidata":"https://www.wikidata.org/wiki/Q6693015","display_name":"Low latency (capital markets)","level":2,"score":0.2856000065803528},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.27309998869895935},{"id":"https://openalex.org/C96324660","wikidata":"https://www.wikidata.org/wiki/Q205446","display_name":"Dataflow","level":2,"score":0.2700999975204468},{"id":"https://openalex.org/C155281189","wikidata":"https://www.wikidata.org/wiki/Q3518150","display_name":"Tensor (intrinsic definition)","level":2,"score":0.26350000500679016},{"id":"https://openalex.org/C4069607","wikidata":"https://www.wikidata.org/wiki/Q868732","display_name":"Aliasing","level":3,"score":0.26179999113082886},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.26179999113082886},{"id":"https://openalex.org/C2780799671","wikidata":"https://www.wikidata.org/wiki/Q17087362","display_name":"Transient (computer programming)","level":2,"score":0.25589999556541443},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2542000114917755}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3797905.3800532","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3797905.3800532","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 40th ACM International Conference on Supercomputing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3797905.3800532","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3797905.3800532","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 40th ACM International Conference on Supercomputing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":4,"referenced_works":["https://openalex.org/W4387321091","https://openalex.org/W4415797413","https://openalex.org/W4416955988","https://openalex.org/W7133227460"],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"Large":[1],"Language":[2],"Models":[3,13],"(dLLMs)":[4],"have":[5],"emerged":[6],"as":[7],"a":[8,31,48,105],"promising":[9],"alternative":[10],"to":[11,18,54,100,108,117,139],"Autoregressive":[12],"(ARMs),":[14],"utilizing":[15],"parallel":[16],"decoding":[17],"overcome":[19],"sequential":[20],"bottlenecks.":[21],"However,":[22],"existing":[23],"research":[24],"focuses":[25],"primarily":[26],"on":[27,127,151,161],"kernel-level":[28],"optimizations,":[29],"lacking":[30],"holistic":[32],"serving":[33,84],"framework":[34],"that":[35,86],"addresses":[36],"the":[37,62,140,152,162,179],"unique":[38],"memory":[39,88],"dynamics":[40],"of":[41],"diffusion":[42],"processes":[43],"in":[44],"production.":[45],"We":[46,124],"identify":[47],"critical":[49],"\"memory":[50],"footprint":[51],"crisis\"":[52],"specific":[53],"dLLMs,":[55],"driven":[56],"by":[57,146,170],"monolithic":[58],"logit":[59],"tensors":[60],"and":[61,70,92,113,133,156],"severe":[63],"resource":[64],"oscillation":[65],"between":[66],"compute-bound":[67],"\"Refresh\"":[68],"phases":[69],"bandwidth-bound":[71],"\"Reuse\"":[72],"phases.":[73],"To":[74],"bridge":[75],"this":[76],"gap,":[77],"we":[78],"present":[79],"dLLM-Serve,":[80],"an":[81],"efficient":[82],"dLLM":[83,184],"system":[85],"co-optimizes":[87],"footprint,":[89],"computational":[90],"scheduling,":[91],"generation":[93],"quality.":[94],"dLLM-Serve":[95,126,143,177],"introduces":[96],"Logit-Aware":[97],"Activation":[98],"Budgeting":[99],"decompose":[101],"transient":[102],"tensor":[103],"peaks,":[104],"Phase-Multiplexed":[106],"Scheduler":[107],"interleave":[109],"heterogeneous":[110,195],"request":[111],"phases,":[112],"Head-Centric":[114],"Sparse":[115],"Attention":[116],"decouple":[118],"logical":[119],"sparsity":[120,189],"from":[121],"physical":[122],"storage.":[123],"evaluate":[125],"diverse":[128],"workloads":[129],"(LiveBench,":[130],"Burst,":[131],"OSC)":[132],"GPUs":[134],"(RTX":[135],"4090,":[136],"L40S).":[137],"Relative":[138],"state-of-the-art":[141],"baseline,":[142],"improves":[144],"throughput":[145],"1.61":[147],"\u00d7":[148,150,158,160,173],"\u20141.81":[149],"consumer-grade":[153],"RTX":[154],"4090":[155],"1.60":[157],"\u20141.74":[159],"server-grade":[163],"NVIDIA":[164],"L40S,":[165],"while":[166],"reducing":[167],"tail":[168],"latency":[169],"nearly":[171],"4":[172],"under":[174],"heavy":[175],"contention.":[176],"establishes":[178],"first":[180],"blueprint":[181],"for":[182],"scalable":[183],"inference,":[185],"converting":[186],"theoretical":[187],"algorithmic":[188],"into":[190],"tangible":[191],"wall-clock":[192],"acceleration":[193],"across":[194],"hardware.":[196]},"counts_by_year":[],"updated_date":"2026-07-03T06:20:38.788640","created_date":"2026-07-03T00:00:00"}