{"id":"https://openalex.org/W7154246878","doi":"https://doi.org/10.48550/arxiv.2604.10180","title":"Tessera: Unlocking Heterogeneous GPUs through Kernel-Granularity Disaggregation","display_name":"Tessera: Unlocking Heterogeneous GPUs through Kernel-Granularity Disaggregation","publication_year":2026,"publication_date":"2026-04-11","ids":{"openalex":"https://openalex.org/W7154246878","doi":"https://doi.org/10.48550/arxiv.2604.10180"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.10180","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.10180","pdf_url":null,"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.10180","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5133570626","display_name":"Tiancheng Hu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Tiancheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133609063","display_name":"Jin Qin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qin, Jin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133553379","display_name":"Zheng Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133615384","display_name":"Junhao Hu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Junhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133612844","display_name":"Yuzheng Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yuzheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133561291","display_name":"Lei Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Lei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060703788","display_name":"Yizhou Shan","orcid":"https://orcid.org/0009-0007-9519-0546"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shan, Yizhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065660009","display_name":"Mingxing Zhang","orcid":"https://orcid.org/0000-0001-7904-461X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Mingxing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133555237","display_name":"Ting Cao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cao, Ting","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133575998","display_name":"Chunwei Xia","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xia, Chunwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133593402","display_name":"Huimin Cui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cui, Huimin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133593008","display_name":"Tao Xie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xie, Tao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5133577132","display_name":"Chenxi Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Chenxi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.5764999985694885,"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":0.5764999985694885,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.16210000216960907,"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/T12292","display_name":"Graph Theory and Algorithms","score":0.05220000073313713,"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/granularity","display_name":"Granularity","score":0.7930999994277954},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.6790000200271606},{"id":"https://openalex.org/keywords/symmetric-multiprocessor-system","display_name":"Symmetric multiprocessor system","score":0.6104000210762024},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.605400025844574},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.5306000113487244},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.5059999823570251},{"id":"https://openalex.org/keywords/homogeneous","display_name":"Homogeneous","score":0.4918000102043152},{"id":"https://openalex.org/keywords/workload","display_name":"Workload","score":0.47510001063346863},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.3813000023365021}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8561999797821045},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.7930999994277954},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.6790000200271606},{"id":"https://openalex.org/C172430144","wikidata":"https://www.wikidata.org/wiki/Q17111997","display_name":"Symmetric multiprocessor system","level":2,"score":0.6104000210762024},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.605400025844574},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.6014000177383423},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.5306000113487244},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5302000045776367},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.5059999823570251},{"id":"https://openalex.org/C66882249","wikidata":"https://www.wikidata.org/wiki/Q169336","display_name":"Homogeneous","level":2,"score":0.4918000102043152},{"id":"https://openalex.org/C2778476105","wikidata":"https://www.wikidata.org/wiki/Q628539","display_name":"Workload","level":2,"score":0.47510001063346863},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.3813000023365021},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.33660000562667847},{"id":"https://openalex.org/C78766204","wikidata":"https://www.wikidata.org/wiki/Q555032","display_name":"Multi-core processor","level":2,"score":0.33090001344680786},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.3303999900817871},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.3255000114440918},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.30309998989105225},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.29760000109672546},{"id":"https://openalex.org/C11644782","wikidata":"https://www.wikidata.org/wiki/Q15401790","display_name":"Cost efficiency","level":2,"score":0.2948000133037567},{"id":"https://openalex.org/C2778562939","wikidata":"https://www.wikidata.org/wiki/Q1298791","display_name":"Synchronization (alternating current)","level":3,"score":0.2939000129699707},{"id":"https://openalex.org/C184596265","wikidata":"https://www.wikidata.org/wiki/Q2651576","display_name":"Model of computation","level":3,"score":0.2831000089645386},{"id":"https://openalex.org/C162319229","wikidata":"https://www.wikidata.org/wiki/Q175263","display_name":"Data structure","level":2,"score":0.2759999930858612},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.265500009059906},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.2567000091075897},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.25029999017715454}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.10180","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.10180","pdf_url":null,"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.10180","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.10180","pdf_url":null,"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Disaggregation":[0],"maps":[1],"parts":[2],"of":[3,10,184],"an":[4],"AI":[5],"workload":[6],"to":[7,15,33,52,105,137,151,157,163],"different":[8],"types":[9],"GPUs,":[11,139],"offering":[12],"a":[13,26,72,113,173,190],"path":[14],"utilize":[16],"modern":[17],"heterogeneous":[18,59,129,174],"GPU":[19,175],"clusters.":[20],"However,":[21],"existing":[22,158],"solutions":[23],"operate":[24],"at":[25,189],"coarse":[27],"granularity":[28,84],"and":[29,55,117,131,146,153],"are":[30],"tightly":[31],"coupled":[32],"specific":[34],"model":[35,63,133,164],"architectures,":[36,134],"leaving":[37],"much":[38],"room":[39],"for":[40,61,85],"performance":[41,54],"improvement.":[42],"This":[43],"paper":[44],"presents":[45],"Tessera,":[46],"the":[47,81,182],"first":[48],"kernel":[49],"disaggregation":[50,159],"system":[51],"improve":[53],"cost":[56,147],"efficiency":[57,148],"on":[58],"GPUs":[60,130,188],"large":[62],"inference.":[64],"Our":[65],"key":[66],"insight":[67],"is":[68],"that":[69,141],"kernels":[70],"within":[71],"single":[73],"application":[74],"exhibit":[75],"diverse":[76],"resource":[77],"demands,":[78],"making":[79],"them":[80],"most":[82],"suitable":[83],"aligning":[86],"computation":[87,111],"with":[88,95,110,121],"hardware":[89],"capabilities.":[90],"Tessera":[91,142,178],"integrates":[92],"offline":[93],"analysis":[94],"online":[96],"adaptation":[97],"by":[98,149],"extracting":[99],"precise":[100],"inter-kernel":[101],"dependencies":[102],"from":[103],"PTX":[104],"ensure":[106],"correctness,":[107],"overlapping":[108],"communication":[109],"through":[112],"pipelined":[114],"execution":[115],"model,":[116],"employing":[118],"workload-aware":[119],"scheduling":[120],"lightweight":[122],"runtime":[123],"adaptation.":[124],"Extensive":[125],"evaluations":[126],"across":[127],"five":[128],"four":[132],"scaling":[135],"up":[136,150],"16":[138],"show":[140],"improves":[143],"serving":[144],"throughput":[145,183],"2.3x":[152],"1.6x,":[154],"respectively,":[155],"compared":[156],"methods,":[160],"while":[161],"generalizing":[162],"architectures":[165],"where":[166],"prior":[167],"approaches":[168],"do":[169],"not":[170],"apply.":[171],"Surprisingly,":[172],"pair":[176],"under":[177],"can":[179],"even":[180],"exceed":[181],"two":[185],"homogeneous":[186],"high-end":[187],"lower":[191],"cost.":[192]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-15T00:00:00"}