{"id":"https://openalex.org/W4414985109","doi":"https://doi.org/10.1109/ispass69572.2026.00013","title":"Characterizing State Space Model and Hybrid Language Model Performance with Long Context","display_name":"Characterizing State Space Model and Hybrid Language Model Performance with Long Context","publication_year":2026,"publication_date":"2026-04-26","ids":{"openalex":"https://openalex.org/W4414985109","doi":"https://doi.org/10.1109/ispass69572.2026.00013"},"language":"en","primary_location":{"id":"doi:10.1109/ispass69572.2026.00013","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ispass69572.2026.00013","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)","raw_type":"proceedings-article"},"type":"article","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2507.12442","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5078862091","display_name":"Saptarshi Mitra","orcid":"https://orcid.org/0000-0003-2100-2315"},"institutions":[{"id":"https://openalex.org/I204250578","display_name":"University of California, Irvine","ror":"https://ror.org/04gyf1771","country_code":"US","type":"education","lineage":["https://openalex.org/I204250578"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Saptarshi Mitra","raw_affiliation_strings":["University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA","institution_ids":["https://openalex.org/I204250578"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119917225","display_name":"Rachid Karami","orcid":null},"institutions":[{"id":"https://openalex.org/I204250578","display_name":"University of California, Irvine","ror":"https://ror.org/04gyf1771","country_code":"US","type":"education","lineage":["https://openalex.org/I204250578"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Rachid Karami","raw_affiliation_strings":["University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA","institution_ids":["https://openalex.org/I204250578"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101855022","display_name":"Haocheng Xu","orcid":"https://orcid.org/0000-0002-9683-8996"},"institutions":[{"id":"https://openalex.org/I204250578","display_name":"University of California, Irvine","ror":"https://ror.org/04gyf1771","country_code":"US","type":"education","lineage":["https://openalex.org/I204250578"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Haocheng Xu","raw_affiliation_strings":["University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA","institution_ids":["https://openalex.org/I204250578"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050532440","display_name":"Sitao Huang","orcid":"https://orcid.org/0000-0001-7669-1467"},"institutions":[{"id":"https://openalex.org/I204250578","display_name":"University of California, Irvine","ror":"https://ror.org/04gyf1771","country_code":"US","type":"education","lineage":["https://openalex.org/I204250578"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Sitao Huang","raw_affiliation_strings":["University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA","institution_ids":["https://openalex.org/I204250578"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5074768327","display_name":"Hyoukjun Kwon","orcid":"https://orcid.org/0000-0001-9824-1352"},"institutions":[{"id":"https://openalex.org/I204250578","display_name":"University of California, Irvine","ror":"https://ror.org/04gyf1771","country_code":"US","type":"education","lineage":["https://openalex.org/I204250578"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hyoukjun Kwon","raw_affiliation_strings":["University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California, Irvine,Electrical Engineering and Computer Science,Irvine,USA","institution_ids":["https://openalex.org/I204250578"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I204250578"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.00926171,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"15","last_page":"27"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.798799991607666,"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":0.798799991607666,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5582000017166138},{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.4163999855518341},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.4146000146865845},{"id":"https://openalex.org/keywords/applications-of-artificial-intelligence","display_name":"Applications of artificial intelligence","score":0.3662000000476837},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.3603000044822693},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.3555999994277954},{"id":"https://openalex.org/keywords/hybrid-system","display_name":"Hybrid system","score":0.3296000063419342},{"id":"https://openalex.org/keywords/computational-resource","display_name":"Computational resource","score":0.3278000056743622}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7832000255584717},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5582000017166138},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.4163999855518341},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.4146000146865845},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.38440001010894775},{"id":"https://openalex.org/C157170001","wikidata":"https://www.wikidata.org/wiki/Q4781507","display_name":"Applications of artificial intelligence","level":2,"score":0.3662000000476837},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.3603000044822693},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.3555999994277954},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.334199994802475},{"id":"https://openalex.org/C50897621","wikidata":"https://www.wikidata.org/wiki/Q2665508","display_name":"Hybrid system","level":2,"score":0.3296000063419342},{"id":"https://openalex.org/C127964446","wikidata":"https://www.wikidata.org/wiki/Q1092142","display_name":"Computational resource","level":3,"score":0.3278000056743622},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.32120001316070557},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.3181999921798706},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.31209999322891235},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.3052999973297119},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.3018999993801117},{"id":"https://openalex.org/C72434380","wikidata":"https://www.wikidata.org/wiki/Q230930","display_name":"State space","level":2,"score":0.3001999855041504},{"id":"https://openalex.org/C164155591","wikidata":"https://www.wikidata.org/wiki/Q2067766","display_name":"Satisfiability modulo theories","level":2,"score":0.2962999939918518},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2822999954223633},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.2727999985218048},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.2669999897480011},{"id":"https://openalex.org/C183322885","wikidata":"https://www.wikidata.org/wiki/Q17007702","display_name":"Context model","level":3,"score":0.2662999927997589},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.25780001282691956}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.1109/ispass69572.2026.00013","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ispass69572.2026.00013","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2507.12442","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2507.12442","pdf_url":"https://arxiv.org/pdf/2507.12442","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:doi:10.48550/arxiv.2507.12442","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2507.12442","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2507.12442","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":"pmh:oai:arXiv.org:2507.12442","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2507.12442","pdf_url":"https://arxiv.org/pdf/2507.12442","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414985109.pdf","grobid_xml":"https://content.openalex.org/works/W4414985109.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Emerging":[0],"applications":[1,39],"such":[2,85],"as":[3],"AR":[4],"are":[5,102,151,167,258],"driving":[6],"demands":[7],"for":[8,138,153,161,241],"machine":[9],"intelligence":[10],"capable":[11],"of":[12,73,75,95,111,129,245],"processing":[13],"continuous":[14],"and/or":[15],"long-context":[16,139],"inputs":[17],"on":[18,26,141,156,237],"local":[19],"devices.":[20],"However,":[21],"currently":[22],"dominant":[23],"models":[24,136],"based":[25],"Transformer":[27],"architecture":[28],"suffers":[29],"from":[30],"the":[31,115,234],"quadratic":[32],"computational":[33,96,205],"and":[34,59,98,134,143,158,207,263],"memory":[35,213,231],"overhead,":[36],"which":[37,63,107],"hinders":[38],"required":[40],"to":[41,114,169,189,202,229,248],"process":[42],"long":[43,162,195],"contexts.":[44],"This":[45],"has":[46],"spurred":[47],"a":[48,125,183],"paradigm":[49],"shift":[50],"towards":[51],"new":[52],"architectures":[53],"like":[54,223],"State":[55],"Space":[56],"Models":[57],"(SSMs)":[58],"SSM-Transformer":[60],"hybrid":[61,135],"models,":[62],"provide":[64],"near-linear":[65,68],"scaling.":[66],"The":[67],"scaling":[69],"enabled":[70],"efficient":[71],"handling":[72],"millions":[74],"tokens":[76],"while":[77],"delivering":[78],"high":[79],"performance":[80,97,185],"in":[81,93],"recent":[82],"studies.":[83],"Although":[84],"works":[86],"present":[87,124],"promising":[88],"results,":[89],"their":[90,112,203,249],"workload":[91],"characteristics":[92],"terms":[94],"hardware":[99],"resource":[100],"requirements":[101],"not":[103],"yet":[104],"thoroughly":[105],"explored,":[106],"limits":[108],"our":[109,266],"understanding":[110],"implications":[113],"system":[116],"level":[117],"optimizations.":[118],"To":[119,253],"address":[120],"this":[121],"gap,":[122],"we":[123,257],"comprehensive,":[126],"comparative":[127],"benchmarking":[128],"carefully":[130],"selected":[131],"Transformers,":[132],"SSMs,":[133],"specifically":[137],"inference":[140,235],"consumer":[142,157],"embedded":[144,159],"GPUs.":[145],"Our":[146,215],"analysis":[147,217],"shows":[148],"that":[149,219],"SSMs":[150,181],"well-suited":[152],"on-device":[154],"AI":[155],"GPUs":[160],"context":[163],"inferences.":[164],"While":[165],"Transformers":[166],"up":[168,188],"$1.9":[170],"\\times$":[171,191],"faster":[172,192],"at":[173,193,271],"short":[174],"sequences":[175],"($":[176],"\\lt":[177],"8":[178],"\\mathrm{~K}$":[179,199],"tokens),":[180,200],"demonstrate":[182],"dramatic":[184],"inversion,":[186],"becoming":[187],"$4":[190],"very":[194],"contexts":[196],"($\\sim":[197],"57":[198],"thanks":[201],"linear":[204],"complexity":[206],"$\\boldsymbol{\\sim}":[208],"\\mathbf{6":[209],"4":[210],"\\%}$":[211],"reduced":[212],"footprint.":[214],"operator-level":[216],"reveals":[218],"custom":[220],"SSM":[221],"kernels":[222],"selective":[224],"scan":[225],"despite":[226],"being":[227],"hardware-aware":[228],"minimize":[230],"IO,":[232],"dominate":[233],"runtime":[236],"edge":[238],"platforms,":[239],"accounting":[240],"over":[242],"$55":[243],"\\%$":[244],"latency":[246],"due":[247],"sequential,":[250],"element-wise":[251],"nature.":[252],"foster":[254],"further":[255],"research,":[256],"sharing":[259],"CPU/GPU":[260],"profiling":[261],"traces":[262],"have":[264],"made":[265],"characterization":[267],"framework":[268],"SSM-Scope":[269],"open-sourced":[270],"https://github.com/sapmitra/ssm-scope":[272]},"counts_by_year":[],"updated_date":"2026-07-07T14:30:12.667765","created_date":"2025-10-09T00:00:00"}