{"id":"https://openalex.org/W4399487423","doi":"https://doi.org/10.1145/3649476.3658777","title":"Highly Efficient Load-Balanced Dataflow for SpGEMMs on Systolic Arrays","display_name":"Highly Efficient Load-Balanced Dataflow for SpGEMMs on Systolic Arrays","publication_year":2024,"publication_date":"2024-06-10","ids":{"openalex":"https://openalex.org/W4399487423","doi":"https://doi.org/10.1145/3649476.3658777"},"language":"en","primary_location":{"id":"doi:10.1145/3649476.3658777","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649476.3658777","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Great Lakes Symposium on VLSI 2024","raw_type":"proceedings-article"},"type":"conference-paper","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/A5001952974","display_name":"Guocheng Zhao","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guocheng Zhao","raw_affiliation_strings":["School of Computer Science and Engineering, Beihang University, School of Computer Science and Engineering, Beihang University Beijing, China, China"],"raw_orcid":"https://orcid.org/0009-0001-5477-3020","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Beihang University, School of Computer Science and Engineering, Beihang University Beijing, China, China","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5068652901","display_name":"Yongxiang Cao","orcid":"https://orcid.org/0009-0000-2020-5116"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yongxiang Cao","raw_affiliation_strings":["School of Computer Science and Engineering, Beihang University Beijing, China, China"],"raw_orcid":"https://orcid.org/0009-0000-2020-5116","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Beihang University Beijing, China, China","institution_ids":["https://openalex.org/I82880672"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I82880672"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.07608129,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"272","last_page":"276"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11181","display_name":"Advanced Data Storage Technologies","score":0.9984999895095825,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11181","display_name":"Advanced Data Storage Technologies","score":0.9984999895095825,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T10715","display_name":"Distributed and Parallel Computing Systems","score":0.9958999752998352,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11269","display_name":"Algorithms and Data Compression","score":0.9958999752998352,"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/dataflow","display_name":"Dataflow","score":0.8942419290542603},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6878941655158997},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.645852267742157}],"concepts":[{"id":"https://openalex.org/C96324660","wikidata":"https://www.wikidata.org/wiki/Q205446","display_name":"Dataflow","level":2,"score":0.8942419290542603},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6878941655158997},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.645852267742157}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3649476.3658777","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649476.3658777","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Great Lakes Symposium on VLSI 2024","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7","score":0.7599999904632568}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":9,"referenced_works":["https://openalex.org/W2801748224","https://openalex.org/W3016542674","https://openalex.org/W3040024858","https://openalex.org/W3193946904","https://openalex.org/W4309620166","https://openalex.org/W4312929963","https://openalex.org/W4316252394","https://openalex.org/W4320712936","https://openalex.org/W4360831848"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2748952813","https://openalex.org/W2293118914","https://openalex.org/W2998381397","https://openalex.org/W4236419692","https://openalex.org/W3167919718","https://openalex.org/W4251718783","https://openalex.org/W2171015181","https://openalex.org/W4239447582","https://openalex.org/W1484403103"],"abstract_inverted_index":{"To":[0],"enhance":[1],"the":[2,16,50,70,78,125,128,132,137,171],"efficiency":[3],"of":[4,53,65,80,111,131,140,181],"sparse":[5,21,104,157],"neural":[6,35],"network":[7],"models,":[8],"compression":[9],"methods":[10],"are":[11],"commonly":[12],"employed":[13],"to":[14,49,62,94,122,155,166,170],"store":[15],"non-zero":[17,55],"elements":[18],"in":[19,33,77],"a":[20,30,45,151,161],"storage":[22],"format.":[23],"Sparse":[24],"General":[25],"Matrix":[26],"Multiplication":[27],"(SpGEMM)":[28],"is":[29],"critical":[31],"computation":[32],"deep":[34],"networks.":[36],"However,":[37],"when":[38],"utilizing":[39],"systolic":[40,71,133],"arrays":[41],"for":[42],"SpGEMM":[43,173],"computations,":[44],"challenge":[46],"arises":[47],"due":[48],"irregular":[51],"flow":[52],"compressed,":[54],"element":[56],"activation":[57,66,112],"data.":[58],"This":[59,119],"irregularity":[60],"leads":[61],"varying":[63],"lengths":[64],"data":[67,89,113,158],"streams":[68,90,114,159],"entering":[69],"array":[72],"per":[73],"batch,":[74],"potentially":[75],"resulting":[76],"underutilization":[79],"processing":[81],"units.":[82],"Our":[83,144],"research":[84],"focuses":[85],"on":[86],"repackaging":[87],"compressed":[88],"using":[91],"hardware-software":[92],"co-design":[93],"minimize":[95],"software":[96],"pre-processing":[97],"time.":[98],"We":[99],"also":[100],"package":[101],"unevenly":[102],"sized":[103],"matrix":[105],"rows":[106],"post-compression":[107],"into":[108],"multiple":[109],"groups":[110],"with":[115],"approximately":[116],"equal":[117],"lengths.":[118],"approach":[120,177],"aims":[121],"evenly":[123],"distribute":[124],"workload":[126],"across":[127],"fixed":[129],"output":[130],"array,":[134],"thereby":[135],"improving":[136],"utilization":[138],"rate":[139],"Processing":[141],"Elements":[142],"(PE).":[143],"evaluation":[145],"demonstrates":[146],"that":[147],"our":[148,176],"method":[149],"achieves":[150,178],"2.01x":[152,182],"acceleration":[153,164],"compared":[154,169],"uncompressed":[156],"and":[160],"3.63x":[162],"average":[163,180],"relative":[165],"TPUs.":[167],"Furthermore,":[168],"state-of-the-art":[172],"accelerator":[174],"SADD,":[175],"an":[179],"acceleration.":[183]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}