{"id":"https://openalex.org/W7165557230","doi":"https://doi.org/10.5573/jsts.2026.26.3.228","title":"Nonzero-aware PE Line Partitioning for Improved Bandwidth Utilization in DRAM Bandwidth-scalable SpMV Accelerators","display_name":"Nonzero-aware PE Line Partitioning for Improved Bandwidth Utilization in DRAM Bandwidth-scalable SpMV Accelerators","publication_year":2026,"publication_date":"2026-06-23","ids":{"openalex":"https://openalex.org/W7165557230","doi":"https://doi.org/10.5573/jsts.2026.26.3.228"},"language":"en","primary_location":{"id":"doi:10.5573/jsts.2026.26.3.228","doi":"https://doi.org/10.5573/jsts.2026.26.3.228","is_oa":false,"landing_page_url":"https://doi.org/10.5573/jsts.2026.26.3.228","pdf_url":null,"source":{"id":"https://openalex.org/S2482218150","display_name":"JSTS Journal of Semiconductor Technology and Science","issn_l":"1598-1657","issn":["1598-1657","2233-4866"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","dongbi-c","kci-registered","poland-20","russia-white-list-4"],"host_organization":"https://openalex.org/P4366305964","host_organization_name":"Institute of Electronics Engineers of Korea","host_organization_lineage":["https://openalex.org/P4366305964"],"host_organization_lineage_names":["Institute of Electronics Engineers of Korea"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"JOURNAL OF SEMICONDUCTOR TECHNOLOGY AND SCIENCE","raw_type":"journal-article"},"volume":null,"type":"article","is_traditional":true,"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/A5023184900","display_name":"Hyunji Kim","orcid":"https://orcid.org/0000-0001-9828-3894","observed_orcids":["https://orcid.org/0000-0001-9828-3894"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hyunji Kim","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139095055","display_name":"Ji-Hoon Kim","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ji-Hoon Kim","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.0,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"26","issue":"3","first_page":"228","last_page":"234"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9937000274658203,"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":0.9937000274658203,"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/T10792","display_name":"Matrix Theory and Algorithms","score":0.0013000000035390258,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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/T10904","display_name":"Embedded Systems Design Techniques","score":0.0010000000474974513,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/sparse-matrix-vector-multiplication","display_name":"sparse matrix-vector multiplication","score":0.9980000257492065},{"id":"https://openalex.org/keywords/bandwidth-utilization","display_name":"bandwidth utilization","score":0.8420000076293945},{"id":"https://openalex.org/keywords/memory-bound-kernels","display_name":"memory-bound kernels","score":0.7310000061988831},{"id":"https://openalex.org/keywords/hardware-accelerators","display_name":"hardware accelerators","score":0.3479999899864197}],"concepts":[{"id":"https://openalex.org/C7366592","wikidata":"https://www.wikidata.org/wiki/Q1255620","display_name":"Dram","level":2,"score":0.7348999977111816},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6970999836921692},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.47189998626708984},{"id":"https://openalex.org/C198352243","wikidata":"https://www.wikidata.org/wiki/Q37105","display_name":"Line (geometry)","level":2,"score":0.43059998750686646},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.4092000126838684},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3946000039577484},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.32120001316070557},{"id":"https://openalex.org/C2994168587","wikidata":"https://www.wikidata.org/wiki/Q5295","display_name":"Random access memory","level":2,"score":0.29580000042915344},{"id":"https://openalex.org/C24326235","wikidata":"https://www.wikidata.org/wiki/Q126095","display_name":"Electronic engineering","level":1,"score":0.25440001487731934},{"id":"https://openalex.org/C113843644","wikidata":"https://www.wikidata.org/wiki/Q901882","display_name":"Interface (matter)","level":4,"score":0.2500999867916107}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.5573/jsts.2026.26.3.228","doi":"https://doi.org/10.5573/jsts.2026.26.3.228","is_oa":false,"landing_page_url":"https://doi.org/10.5573/jsts.2026.26.3.228","pdf_url":null,"source":{"id":"https://openalex.org/S2482218150","display_name":"JSTS Journal of Semiconductor Technology and Science","issn_l":"1598-1657","issn":["1598-1657","2233-4866"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","dongbi-c","kci-registered","poland-20","russia-white-list-4"],"host_organization":"https://openalex.org/P4366305964","host_organization_name":"Institute of Electronics Engineers of Korea","host_organization_lineage":["https://openalex.org/P4366305964"],"host_organization_lineage_names":["Institute of Electronics Engineers of Korea"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"JOURNAL OF SEMICONDUCTOR TECHNOLOGY AND SCIENCE","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9","score":0.428600013256073}],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2025890876","https://openalex.org/W4206115986","https://openalex.org/W4310607603","https://openalex.org/W4404688226","https://openalex.org/W7125962849","https://openalex.org/W4389888442","https://openalex.org/W1997955449","https://openalex.org/W2557515111","https://openalex.org/W3158275024","https://openalex.org/W4285266443"],"abstract_inverted_index":{"Sparse":[0],"matrix-vector":[1],"multiplication":[2],"(SpMV)":[3],"is":[4],"a":[5,24,37,135],"memory-bound":[6],"kernel":[7],"widely":[8],"used":[9],"in":[10,161],"scientific":[11],"computing,":[12],"machine":[13],"learning,":[14],"and":[15,60,98,124,133,196,207,212,224],"graph":[16],"analytics,":[17],"where":[18],"bandwidth":[19,232],"utilization":[20],"(BU)":[21],"serves":[22],"as":[23,104],"key":[25],"performance":[26],"metric":[27],"for":[28,209],"hardware":[29],"accelerators.":[30],"In":[31,113],"our":[32,162],"prior":[33,163],"work,":[34],"we":[35,116],"presented":[36],"DRAM":[38,50],"bandwidth-scalable":[39],"SpMV":[40],"accelerator":[41,184],"whose":[42],"processing":[43],"element":[44],"(PE)":[45],"line":[46,82,106,126,215],"count":[47,107],"scales":[48],"with":[49,53,109,218],"bandwidth,":[51],"combined":[52],"offline":[54,159],"pre-processing":[55,160],"to":[56,151,181],"eliminate":[57],"bank":[58],"conflicts":[59],"data":[61,168],"dependencies,":[62],"achieving":[63],"an":[64],"average":[65,201],"BU":[66,202],"of":[67,69,148,204,234],"89%":[68],"the":[70,118,145,158,170,174,182,228,231,235],"theoretical":[71],"maximum.":[72],"However,":[73],"its":[74],"column-index-based":[75],"equal":[76],"partitioning":[77,137,176],"can":[78],"cause":[79],"significant":[80],"PE":[81,95,105,125,155,198,214],"load":[83],"imbalance":[84,127],"when":[85],"nonzero":[86,122,149],"elements":[87,150],"are":[88],"unevenly":[89],"distributed":[90],"across":[91,154,191,221],"columns,":[92],"leaving":[93],"faster":[94],"lines":[96],"idle":[97],"degrading":[99],"BU\u2013an":[100],"effect":[101],"that":[102,139,227],"intensifies":[103],"increases":[108],"higher":[110],"off-chip":[111],"bandwidth.":[112],"this":[114],"paper,":[115],"analyze":[117],"relationship":[119],"between":[120],"column-wise":[121],"distribution":[123],"using":[128],"29":[129,189],"SuiteSparse":[130],"benchmark":[131],"matrices":[132,190,223],"propose":[134],"nonzero-aware":[136],"strategy":[138],"assigns":[140],"column":[141],"ranges":[142],"based":[143],"on":[144,166,188],"actual":[146],"number":[147],"equalize":[152],"workloads":[153],"lines.":[156],"While":[157],"work":[164],"focused":[165],"conflict-free":[167],"rearrangement,":[169],"proposed":[171],"method":[172,229],"targets":[173],"preceding":[175],"stage,":[177],"requiring":[178],"no":[179],"modification":[180],"on-chip":[183],"hardware.":[185],"Cycle-accurate":[186],"simulation":[187],"three":[192],"configurations":[193],"(2,":[194],"4,":[195,211],"8":[197,213],"lines)":[199],"shows":[200],"improvements":[203],"17.2%,":[205],"65.4%,":[206],"113.7%":[208],"2,":[210],"configurations,":[216,225],"respectively,":[217],"zero":[219],"degradation":[220],"all":[222],"confirming":[226],"preserves":[230],"scalability":[233],"baseline":[236],"accelerator.":[237]},"counts_by_year":[],"updated_date":"2026-06-24T06:17:17.354583","created_date":"2026-06-23T00:00:00"}