{"id":"https://openalex.org/W7137823278","doi":"https://doi.org/10.1109/tvlsi.2026.3673262","title":"DyTopK: Accelerating Top-K SpMV on Embedded FPGAs via Dynamic Floating-Point Quantization","display_name":"DyTopK: Accelerating Top-K SpMV on Embedded FPGAs via Dynamic Floating-Point Quantization","publication_year":2026,"publication_date":"2026-03-17","ids":{"openalex":"https://openalex.org/W7137823278","doi":"https://doi.org/10.1109/tvlsi.2026.3673262"},"language":null,"primary_location":{"id":"doi:10.1109/tvlsi.2026.3673262","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tvlsi.2026.3673262","pdf_url":null,"source":{"id":"https://openalex.org/S37538908","display_name":"IEEE Transactions on Very Large Scale Integration (VLSI) Systems","issn_l":"1063-8210","issn":["1063-8210","1557-9999"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Very Large Scale Integration (VLSI) Systems","raw_type":"journal-article"},"type":"article","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/A5088092175","display_name":"Shangshang Yao","orcid":"https://orcid.org/0000-0001-7217-1712"},"institutions":[{"id":"https://openalex.org/I170215575","display_name":"National University of Defense Technology","ror":"https://ror.org/05d2yfz11","country_code":"CN","type":"education","lineage":["https://openalex.org/I170215575"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shangshang Yao","raw_affiliation_strings":["College of Computer Science and Technology, National University of Defense Technology, Changsha, China"],"raw_orcid":"https://orcid.org/0000-0001-7217-1712","affiliations":[{"raw_affiliation_string":"College of Computer Science and Technology, National University of Defense Technology, Changsha, China","institution_ids":["https://openalex.org/I170215575"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101761519","display_name":"Jingyu Liu","orcid":"https://orcid.org/0000-0003-0233-4744"},"institutions":[{"id":"https://openalex.org/I170215575","display_name":"National University of Defense Technology","ror":"https://ror.org/05d2yfz11","country_code":"CN","type":"education","lineage":["https://openalex.org/I170215575"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jingyu Liu","raw_affiliation_strings":["College of Computer Science and Technology, National University of Defense Technology, Changsha, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"College of Computer Science and Technology, National University of Defense Technology, Changsha, China","institution_ids":["https://openalex.org/I170215575"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5031114432","display_name":"L. L. Shen","orcid":"https://orcid.org/0000-0002-8204-4466"},"institutions":[{"id":"https://openalex.org/I170215575","display_name":"National University of Defense Technology","ror":"https://ror.org/05d2yfz11","country_code":"CN","type":"education","lineage":["https://openalex.org/I170215575"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Li Shen","raw_affiliation_strings":["College of Computer Science and Technology, National University of Defense Technology, Changsha, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"College of Computer Science and Technology, National University of Defense Technology, Changsha, China","institution_ids":["https://openalex.org/I170215575"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5048593244","display_name":"Z. Y. Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I40246663","display_name":"Chinese Academy of Engineering","ror":"https://ror.org/00z3yke57","country_code":"CN","type":"other","lineage":["https://openalex.org/I40246663"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zuoning Chen","raw_affiliation_strings":["Chinese Academy of Engineering, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chinese Academy of Engineering, Beijing, China","institution_ids":["https://openalex.org/I40246663"]}]}],"institutions":[],"countries_distinct_count":1,"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":0.35424414,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"34","issue":"6","first_page":"1762","last_page":"1771"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.08649999648332596,"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":0.08649999648332596,"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/T10829","display_name":"Interconnection Networks and Systems","score":0.05180000141263008,"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/T12326","display_name":"Network Packet Processing and Optimization","score":0.03779999911785126,"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/quantization","display_name":"Quantization (signal processing)","score":0.48170000314712524},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.4697999954223633},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2903999984264374},{"id":"https://openalex.org/keywords/reconfigurable-computing","display_name":"Reconfigurable computing","score":0.29030001163482666},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.23260000348091125}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6793000102043152},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.48170000314712524},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.4697999954223633},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.35850000381469727},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.2946999967098236},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.2937999963760376},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2903999984264374},{"id":"https://openalex.org/C142962650","wikidata":"https://www.wikidata.org/wiki/Q240838","display_name":"Reconfigurable computing","level":3,"score":0.29030001163482666},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.23260000348091125},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.21359999477863312}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tvlsi.2026.3673262","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tvlsi.2026.3673262","pdf_url":null,"source":{"id":"https://openalex.org/S37538908","display_name":"IEEE Transactions on Very Large Scale Integration (VLSI) Systems","issn_l":"1063-8210","issn":["1063-8210","1557-9999"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Very Large Scale Integration (VLSI) Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.745125949382782,"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7"}],"awards":[{"id":"https://openalex.org/G3185965449","display_name":null,"funder_award_id":"62472436","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G5413123383","display_name":null,"funder_award_id":"WDZC20245250110","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Top-K":[0],"sparse":[1,138],"matrix\u2013vector":[2],"multiplication":[3],"is":[4],"the":[5,22,25,64,115,119,127,149],"computational":[6,213],"backbone":[7],"of":[8,67,121,129,154],"modern":[9],"recommendation":[10],"systems":[11],"and":[12,30,71,204],"graph":[13],"neural":[14],"networks":[15],"(GNNs).":[16],"While":[17],"shifting":[18],"these":[19,159],"workloads":[20],"to":[21,216],"edge":[23,90],"offers":[24],"potential":[26],"for":[27,57,88,175],"ultralow":[28],"latency":[29],"enhanced":[31],"privacy,":[32],"it":[33,221],"presents":[34],"significant":[35,189],"architectural":[36],"challenges.":[37],"Recent":[38],"HBM-based":[39],"field-programmable":[40],"gate":[41],"array":[42],"(FPGA)":[43],"accelerators":[44],"deployed":[45],"in":[46,200,234],"data":[47,151],"centers":[48],"have":[49],"shown":[50],"promise,":[51],"but":[52],"their":[53],"designs":[54],"are":[55],"ill-suited":[56],"power-constrained":[58],"embedded":[59,89],"FPGAs,":[60],"which":[61],"suffer":[62],"from":[63],"limited":[65],"bandwidth":[66,202],"standard":[68,155],"DDR":[69],"interfaces":[70],"restricted":[72,130],"on-chip":[73],"logic":[74,181],"resources.":[75,182],"In":[76],"this":[77],"article,":[78],"we":[79,133,162],"propose":[80,135],"DyTopK,":[81],"a":[82,106,136,164,169],"hardware\u2013software":[83],"codesigned":[84],"accelerator":[85],"tailored":[86],"specifically":[87],"devices.":[91],"Unlike":[92],"prior":[93,223],"approaches":[94],"that":[95,186],"rely":[96],"on":[97,179],"coarse-grained":[98],"blockwise":[99],"quantization":[100],"or":[101],"complex":[102],"filtering,":[103],"DyTopK":[104,187],"introduces":[105],"novel":[107],"Dynamic":[108],"FP4/FP8":[109],"Quantization":[110],"scheme.":[111],"This":[112],"mechanism":[113],"adapts":[114],"numerical":[116],"format":[117,141],"at":[118,247],"granularity":[120],"individual":[122],"nonzero":[123],"elements.":[124],"To":[125],"maximize":[126],"utility":[128],"off-chip":[131],"bandwidth,":[132],"further":[134],"bandwidth-maximal":[137],"matrix":[139],"compression":[140],"named":[142],"packet":[143],"CSR":[144],"(PCSR),":[145],"perfectly":[146],"aligning":[147],"with":[148],"64-bit":[150],"bus":[152],"width":[153],"DDR4":[156],"interfaces.":[157],"Complementing":[158],"data-centric":[160],"optimizations,":[161],"design":[163],"high-efficiency":[165],"hardware":[166],"architecture":[167],"featuring":[168],"scalable":[170],"processing":[171],"element":[172],"array,":[173],"allowing":[174],"flexible":[176],"expansion":[177],"based":[178],"available":[180],"Experimental":[183],"results":[184],"demonstrate":[185],"achieves":[188],"improvements,":[190],"delivering":[191],"<inline-formula":[192,205,227],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[193,206,228],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[194,207,229],"<tex-math":[195,208,230],"notation=\"LaTeX\">$12.4\\times":[196],"$</tex-math>":[197,210,232],"</inline-formula>":[198,211,233],"speedup":[199],"effective":[201],"utilization":[203],"notation=\"LaTeX\">$8.6\\times":[209],"higher":[212],"throughput":[214],"compared":[215],"state-of-the-art":[217],"CPU":[218],"baselines.":[219],"Furthermore,":[220],"outperforms":[222],"FPGA":[224],"implementations":[225],"by":[226],"notation=\"LaTeX\">$2.3\\times":[231],"energy":[235],"efficiency":[236],"while":[237],"maintaining":[238],"negligible":[239],"accuracy":[240],"loss":[241],"(less":[242],"than":[243],"0.5%":[244],"recall":[245],"drop)":[246],"K":[248],"=":[249],"100.":[250]},"counts_by_year":[],"updated_date":"2026-06-03T06:16:58.514037","created_date":"2026-03-18T00:00:00"}
