{"id":"https://openalex.org/W4402426679","doi":"https://doi.org/10.1145/3676536.3676840","title":"Mixed-precision Neural Networks on RISC-V Cores: ISA extensions for Multi-Pumped Soft SIMD Operations","display_name":"Mixed-precision Neural Networks on RISC-V Cores: ISA extensions for Multi-Pumped Soft SIMD Operations","publication_year":2024,"publication_date":"2024-10-27","ids":{"openalex":"https://openalex.org/W4402426679","doi":"https://doi.org/10.1145/3676536.3676840"},"language":"en","primary_location":{"id":"doi:10.1145/3676536.3676840","doi":"https://doi.org/10.1145/3676536.3676840","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3676536.3676840","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676840","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 43rd IEEE/ACM International Conference on Computer-Aided Design","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676840","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5063747079","display_name":"Giorgos Armeniakos","orcid":"https://orcid.org/0000-0001-7184-3740","observed_orcids":["https://orcid.org/0000-0001-7184-3740"]},"institutions":[{"id":"https://openalex.org/I174458059","display_name":"National Technical University of Athens","ror":"https://ror.org/03cx6bg69","country_code":"GR","type":"education","lineage":["https://openalex.org/I174458059"]}],"countries":["GR"],"is_corresponding":false,"raw_author_name":"Giorgos Armeniakos","raw_affiliation_strings":["National Technichal University of Athens, Athens, Greece"],"raw_orcid":"https://orcid.org/0000-0001-7184-3740","affiliations":[{"raw_affiliation_string":"National Technichal University of Athens, Athens, Greece","institution_ids":["https://openalex.org/I174458059"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137905275","display_name":"Alexis Maras","orcid":"https://orcid.org/0009-0001-4508-6384","observed_orcids":["https://orcid.org/0009-0001-4508-6384"]},"institutions":[{"id":"https://openalex.org/I174458059","display_name":"National Technical University of Athens","ror":"https://ror.org/03cx6bg69","country_code":"GR","type":"education","lineage":["https://openalex.org/I174458059"]}],"countries":["GR"],"is_corresponding":false,"raw_author_name":"Alexis Maras","raw_affiliation_strings":["National Technichal University of Athens, Athens, Greece"],"raw_orcid":"https://orcid.org/0009-0001-4508-6384","affiliations":[{"raw_affiliation_string":"National Technichal University of Athens, Athens, Greece","institution_ids":["https://openalex.org/I174458059"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076432415","display_name":"Sotirios Xydis","orcid":"https://orcid.org/0000-0003-3151-2730","observed_orcids":["https://orcid.org/0000-0003-3151-2730"]},"institutions":[{"id":"https://openalex.org/I174458059","display_name":"National Technical University of Athens","ror":"https://ror.org/03cx6bg69","country_code":"GR","type":"education","lineage":["https://openalex.org/I174458059"]}],"countries":["GR"],"is_corresponding":false,"raw_author_name":"Sotirios Xydis","raw_affiliation_strings":["National Technichal University of Athens, Athens, Greece"],"raw_orcid":"https://orcid.org/0000-0003-3151-2730","affiliations":[{"raw_affiliation_string":"National Technichal University of Athens, Athens, Greece","institution_ids":["https://openalex.org/I174458059"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5043131021","display_name":"Dimitrios Soudris","orcid":"https://orcid.org/0000-0002-6930-6847","observed_orcids":["https://orcid.org/0000-0002-6930-6847"]},"institutions":[{"id":"https://openalex.org/I174458059","display_name":"National Technical University of Athens","ror":"https://ror.org/03cx6bg69","country_code":"GR","type":"education","lineage":["https://openalex.org/I174458059"]}],"countries":["GR"],"is_corresponding":false,"raw_author_name":"Dimitrios Soudris","raw_affiliation_strings":["National Technichal University of Athens, Athens, Greece"],"raw_orcid":"https://orcid.org/0000-0002-6930-6847","affiliations":[{"raw_affiliation_string":"National Technichal University of Athens, Athens, Greece","institution_ids":["https://openalex.org/I174458059"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I174458059"],"apc_list":null,"apc_paid":null,"fwci":5.3404,"has_fulltext":true,"cited_by_count":23,"citation_normalized_percentile":{"value":0.97178545,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":88,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"9"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.41670000553131104,"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":0.41670000553131104,"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":0.25270000100135803,"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/T10363","display_name":"Low-power high-performance VLSI design","score":0.19370000064373016,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/risc-v","display_name":"RISC-V","score":0.8140000104904175},{"id":"https://openalex.org/keywords/instruction-set-extension","display_name":"instruction set extension","score":0.9589999914169312},{"id":"https://openalex.org/keywords/mixed-precision-quantization","display_name":"mixed-precision quantization","score":0.8159999847412109},{"id":"https://openalex.org/keywords/hardware-software-co-design","display_name":"hardware-software co-design","score":0.7429999709129333},{"id":"https://openalex.org/keywords/deep-neural-network-inference","display_name":"deep neural network inference","score":0.6119999885559082},{"id":"https://openalex.org/keywords/energy-efficiency","display_name":"energy efficiency","score":0.902999997138977},{"id":"https://openalex.org/keywords/embedded-processors","display_name":"embedded processors","score":0.5139999985694885}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8478413224220276},{"id":"https://openalex.org/C150552126","wikidata":"https://www.wikidata.org/wiki/Q339387","display_name":"SIMD","level":2,"score":0.6602457761764526},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.5894181728363037},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5643739700317383},{"id":"https://openalex.org/C202491316","wikidata":"https://www.wikidata.org/wiki/Q272683","display_name":"Instruction set","level":2,"score":0.5137516260147095},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.49913525581359863},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.47332701086997986},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.42176204919815063},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.4198077619075775},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.4159695506095886},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.35420966148376465},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.35280343890190125},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.23861637711524963},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.13823539018630981},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.10596370697021484},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.0},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3676536.3676840","doi":"https://doi.org/10.1145/3676536.3676840","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3676536.3676840","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676840","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 43rd IEEE/ACM International Conference on Computer-Aided Design","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2407.14274","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2407.14274","pdf_url":"https://arxiv.org/pdf/2407.14274","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2407.14274","doi":"https://doi.org/10.48550/arxiv.2407.14274","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2407.14274","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"article-journal"}],"best_oa_location":{"id":"doi:10.1145/3676536.3676840","doi":"https://doi.org/10.1145/3676536.3676840","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3676536.3676840","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676840","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 43rd IEEE/ACM International Conference on Computer-Aided Design","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4402426679.pdf","grobid_xml":"https://content.openalex.org/works/W4402426679.grobid-xml"},"referenced_works_count":27,"referenced_works":["https://openalex.org/W2152341771","https://openalex.org/W2289252105","https://openalex.org/W2346205343","https://openalex.org/W2588598812","https://openalex.org/W2606722458","https://openalex.org/W2612445135","https://openalex.org/W2741543856","https://openalex.org/W2784372305","https://openalex.org/W2793471971","https://openalex.org/W2895531329","https://openalex.org/W2950027350","https://openalex.org/W3002612001","https://openalex.org/W3005924570","https://openalex.org/W3013186616","https://openalex.org/W3032746325","https://openalex.org/W3036554530","https://openalex.org/W3044604993","https://openalex.org/W3047342117","https://openalex.org/W3049586678","https://openalex.org/W4211085526","https://openalex.org/W4220829081","https://openalex.org/W4312121072","https://openalex.org/W4313483708","https://openalex.org/W4360831846","https://openalex.org/W4389072746","https://openalex.org/W4390306248","https://openalex.org/W4391547488"],"related_works":["https://openalex.org/W4414603543","https://openalex.org/W7165131753","https://openalex.org/W3091872000","https://openalex.org/W4404609150","https://openalex.org/W4386474973","https://openalex.org/W4410810511","https://openalex.org/W3032746325","https://openalex.org/W4306292732","https://openalex.org/W4410738306","https://openalex.org/W3109152438"],"abstract_inverted_index":{"Recent":[0],"advancements":[1],"in":[2,137],"quantization":[3],"and":[4,14,56,78,125,145,178,192,197,216],"mixed-precision":[5,47,64,98,121,142,176],"approaches":[6],"offers":[7],"substantial":[8],"opportunities":[9],"to":[10,35,72,97,148,181],"improve":[11],"the":[12,91,129,134,160,170,182,218],"speed":[13],"energy":[15,208],"efficiency":[16],"of":[17,63],"Neural":[18],"Networks":[19],"(NN).":[20],"Research":[21],"has":[22],"shown":[23],"that":[24,116,200],"individual":[25],"parameters":[26],"with":[27,152],"varying":[28],"low":[29],"precision,":[30],"can":[31,203],"attain":[32],"accuracies":[33],"comparable":[34],"full-precision":[36],"counterparts.":[37],"However,":[38],"modern":[39],"embedded":[40],"microprocessors":[41],"provide":[42],"very":[43],"limited":[44],"support":[45],"for":[46,60,75,90,140,156,174,210],"NNs":[48],"regarding":[49],"both":[50],"Instruction":[51],"Set":[52],"Architecture":[53],"(ISA)":[54],"extensions":[55,95],"their":[57],"hardware":[58,99,119,130],"design":[59],"efficient":[61],"execution":[62,150],"operations,":[65],"i.e.,":[66],"introducing":[67],"several":[68],"performance":[69],"bottlenecks":[70],"due":[71],"numerous":[73],"instructions":[74,168],"data":[76],"packing":[77],"unpacking,":[79],"arithmetic":[80,143],"unit":[81,136],"under-utilizations":[82],"etc.":[83],"In":[84],"this":[85],"work,":[86],"we":[87,132,163],"bring":[88],"together,":[89],"first":[92],"time,":[93],"ISA":[94,123,161],"tailored":[96],"optimizations,":[100],"targeting":[101],"energy-efficient":[102],"DNN":[103],"inference":[104],"on":[105,205],"leading":[106],"RISC-V":[107,171,221],"CPU":[108],"architectures.":[109],"We":[110],"introduce":[111],"a":[112],"hardware-software":[113],"co-design":[114],"framework":[115,202],"supports":[117],"cooperative":[118],"design,":[120],"quantization,":[122],"extensions,":[124],"cycle-accurate":[126],"emulations.":[127],"At":[128,159],"level,":[131,162],"expand":[133],"ALU":[135],"our":[138,201],"micro-architecture":[139],"configurable":[141],"operations":[144],"implement":[146],"multi-pumping":[147],"reduce":[149],"latency,":[151],"soft":[153],"SIMD":[154],"optimization":[155],"2-bit":[157],"operations.":[158],"encode":[164],"three":[165],"distinct":[166],"MAC":[167],"extending":[169],"ISA,":[172],"each":[173],"different":[175],"modes,":[177],"expose":[179],"them":[180],"compiler.":[183],"Our":[184],"extensive":[185],"experimental":[186],"evaluation":[187],"over":[188],"widely":[189],"used":[190],"DNNs":[191],"datasets,":[193],"such":[194],"as":[195],"CIFAR10":[196],"ImageNet,":[198],"demonstrates":[199],"achieve,":[204],"average,":[206],"15\u00d7":[207],"reduction":[209],"less":[211],"than":[212],"1%":[213],"accuracy":[214],"loss":[215],"outperforms":[217],"ISA-agnostic":[219],"state-of-the-art":[220],"cores.":[222]},"counts_by_year":[{"year":2026,"cited_by_count":7},{"year":2025,"cited_by_count":15},{"year":2024,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2024-09-11T00:00:00"}