{"id":"https://openalex.org/W4319917397","doi":"https://doi.org/10.1145/3578178.3578222","title":"A new data conversion method for mixed precision Krylov solvers with FP16/BF16 Jacobi preconditioners","display_name":"A new data conversion method for mixed precision Krylov solvers with FP16/BF16 Jacobi preconditioners","publication_year":2023,"publication_date":"2023-02-10","ids":{"openalex":"https://openalex.org/W4319917397","doi":"https://doi.org/10.1145/3578178.3578222"},"language":"en","primary_location":{"id":"doi:10.1145/3578178.3578222","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3578178.3578222","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the International Conference on High Performance Computing in Asia-Pacific Region","raw_type":"proceedings-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/A5088222852","display_name":"Takuya Ina","orcid":"https://orcid.org/0000-0002-3989-5011"},"institutions":[{"id":"https://openalex.org/I117197279","display_name":"Japan Atomic Energy Agency","ror":"https://ror.org/05nf86y53","country_code":"JP","type":"funder","lineage":["https://openalex.org/I117197279"]}],"countries":["JP"],"is_corresponding":true,"raw_author_name":"Takuya Ina","raw_affiliation_strings":["Center for Computational Science and e-Systems, Japan Atomic Energy Agency, Japan"],"affiliations":[{"raw_affiliation_string":"Center for Computational Science and e-Systems, Japan Atomic Energy Agency, Japan","institution_ids":["https://openalex.org/I117197279"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5017412373","display_name":"Yasuhiro Idomura","orcid":"https://orcid.org/0000-0002-2829-0498"},"institutions":[{"id":"https://openalex.org/I117197279","display_name":"Japan Atomic Energy Agency","ror":"https://ror.org/05nf86y53","country_code":"JP","type":"funder","lineage":["https://openalex.org/I117197279"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Yasuhiro Idomura","raw_affiliation_strings":["Center for Computational Science and e-Systems, Japan Atomic Energy Agency, Japan"],"affiliations":[{"raw_affiliation_string":"Center for Computational Science and e-Systems, Japan Atomic Energy Agency, Japan","institution_ids":["https://openalex.org/I117197279"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086152822","display_name":"Toshiyuki Imamura","orcid":"https://orcid.org/0000-0003-1601-9710"},"institutions":[{"id":"https://openalex.org/I4210129730","display_name":"RIKEN Center for Computational Science","ror":"https://ror.org/03r519674","country_code":"JP","type":"facility","lineage":["https://openalex.org/I4210110652","https://openalex.org/I4210129730"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Toshiyuki Imamura","raw_affiliation_strings":["Center for Computational Science RIKEN, Japan"],"affiliations":[{"raw_affiliation_string":"Center for Computational Science RIKEN, Japan","institution_ids":["https://openalex.org/I4210129730"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5006054555","display_name":"Naoyuki Onodera","orcid":"https://orcid.org/0000-0001-7392-2899"},"institutions":[{"id":"https://openalex.org/I117197279","display_name":"Japan Atomic Energy Agency","ror":"https://ror.org/05nf86y53","country_code":"JP","type":"funder","lineage":["https://openalex.org/I117197279"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Naoyuki Onodera","raw_affiliation_strings":["Center for Computational Science and e-Systems, Japan Atomic Energy Agency, Japan"],"affiliations":[{"raw_affiliation_string":"Center for Computational Science and e-Systems, Japan Atomic Energy Agency, Japan","institution_ids":["https://openalex.org/I117197279"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5088222852"],"corresponding_institution_ids":["https://openalex.org/I117197279"],"apc_list":null,"apc_paid":null,"fwci":0.4029,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.61312878,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":96},"biblio":{"volume":null,"issue":null,"first_page":"29","last_page":"34"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10792","display_name":"Matrix Theory and Algorithms","score":0.9998999834060669,"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"}},"topics":[{"id":"https://openalex.org/T10792","display_name":"Matrix Theory and Algorithms","score":0.9998999834060669,"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/T10339","display_name":"Advanced Numerical Methods in Computational Mathematics","score":0.9965000152587891,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10739","display_name":"Electromagnetic Scattering and Analysis","score":0.9948999881744385,"subfield":{"id":"https://openalex.org/subfields/3107","display_name":"Atomic and Molecular Physics, and Optics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/preconditioner","display_name":"Preconditioner","score":0.9083731174468994},{"id":"https://openalex.org/keywords/conjugate-gradient-method","display_name":"Conjugate gradient method","score":0.6065990328788757},{"id":"https://openalex.org/keywords/jacobi-method","display_name":"Jacobi method","score":0.5437078475952148},{"id":"https://openalex.org/keywords/applied-mathematics","display_name":"Applied mathematics","score":0.5178580284118652},{"id":"https://openalex.org/keywords/convergence","display_name":"Convergence (economics)","score":0.4372546076774597},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.4307935833930969},{"id":"https://openalex.org/keywords/diagonal","display_name":"Diagonal","score":0.41492918133735657},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3638991713523865},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3609810173511505},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.35849452018737793},{"id":"https://openalex.org/keywords/iterative-method","display_name":"Iterative method","score":0.3053302764892578},{"id":"https://openalex.org/keywords/geometry","display_name":"Geometry","score":0.082119882106781}],"concepts":[{"id":"https://openalex.org/C167431342","wikidata":"https://www.wikidata.org/wiki/Q1754327","display_name":"Preconditioner","level":3,"score":0.9083731174468994},{"id":"https://openalex.org/C81184566","wikidata":"https://www.wikidata.org/wiki/Q1191895","display_name":"Conjugate gradient method","level":2,"score":0.6065990328788757},{"id":"https://openalex.org/C67923128","wikidata":"https://www.wikidata.org/wiki/Q1481893","display_name":"Jacobi method","level":2,"score":0.5437078475952148},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.5178580284118652},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.4372546076774597},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4307935833930969},{"id":"https://openalex.org/C130367717","wikidata":"https://www.wikidata.org/wiki/Q189791","display_name":"Diagonal","level":2,"score":0.41492918133735657},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3638991713523865},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3609810173511505},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.35849452018737793},{"id":"https://openalex.org/C159694833","wikidata":"https://www.wikidata.org/wiki/Q2321565","display_name":"Iterative method","level":2,"score":0.3053302764892578},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.082119882106781},{"id":"https://openalex.org/C50522688","wikidata":"https://www.wikidata.org/wiki/Q189833","display_name":"Economic growth","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3578178.3578222","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3578178.3578222","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the International Conference on High Performance Computing in Asia-Pacific Region","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":8,"referenced_works":["https://openalex.org/W1506342804","https://openalex.org/W2599973081","https://openalex.org/W2786704328","https://openalex.org/W2914726413","https://openalex.org/W2921480401","https://openalex.org/W3118941458","https://openalex.org/W3130650642","https://openalex.org/W4234592442"],"related_works":["https://openalex.org/W2159515580","https://openalex.org/W2081820376","https://openalex.org/W2091325992","https://openalex.org/W2059625160","https://openalex.org/W2006453807","https://openalex.org/W2740546554","https://openalex.org/W4285325102","https://openalex.org/W2358739842","https://openalex.org/W2185794209","https://openalex.org/W2138831083"],"abstract_inverted_index":{"Mixed":[0],"precision":[1,21,220],"Krylov":[2],"solvers":[3],"with":[4,97],"the":[5,14,67,78,82,86,92,98,107,115,129,133,136,143,152,155,161,168,173,176,185,190,195,201,204,214,225],"Jacobi":[6,15,100,226],"preconditioner":[7,16,101,227],"often":[8],"show":[9],"significant":[10],"convergence":[11,32,169,191,215],"degradation":[12,33,192],"when":[13],"is":[17,28,34,60,74,111,125,140,147,209,222],"computed":[18],"in":[19,45,122],"low":[20],"such":[22],"as":[23],"FP16":[24],"and":[25,91,119,124,181,187,217],"BF16.":[26],"It":[27],"found":[29],"that":[30],"this":[31,50],"attributed":[35],"to":[36,42,62,211],"loss":[37],"of":[38,66,135,145,167,175,189,198],"diagonal":[39,64,180],"dominance":[40,65],"due":[41],"roundoff":[43,177],"errors":[44,178],"data":[46,56,109,138,157,206],"conversion.":[47],"To":[48],"resolve":[49],"issue,":[51],"we":[52],"propose":[53],"a":[54],"new":[55,108,137],"conversion":[57,110,139,158,207],"method,":[58,85,90],"which":[59],"designed":[61],"keep":[63],"original":[68],"matrix":[69,146],"data.":[70],"The":[71],"proposed":[72,205],"method":[73,96,208],"tested":[75],"by":[76,113,150],"computing":[77,221],"Poisson":[79],"equation":[80],"using":[81],"conjugate":[83],"gradient":[84,94],"general":[87],"minimum":[88],"residual":[89],"biconjugate":[93],"stabilized":[95],"FP16/BF16":[99],"on":[102,160,172,194],"NVIDIA":[103],"V100":[104],"GPUs.":[105],"Here,":[106,184],"implemented":[112],"switching":[114],"round-nearest,":[116],"round-up,":[117],"round-down,":[118],"round-towards-zero":[120],"intrinsics":[121],"CUDA,":[123],"called":[126],"once":[127],"before":[128],"main":[130],"iteration.":[131],"Therefore,":[132],"cost":[134],"negligible.":[141],"When":[142],"coefficients":[144],"continuously":[148],"changed":[149],"scaling":[151],"linear":[153],"system,":[154],"conventional":[156],"based":[159],"round-nearest":[162],"intrinsic":[163],"shows":[164],"periodic":[165],"changes":[166],"property":[170],"depending":[171],"difference":[174],"between":[179],"off-diagonal":[182],"coefficients.":[183],"period":[186],"magnitude":[188],"depend":[193],"bit":[196],"length":[197],"significand.":[199],"On":[200],"other":[202],"hand,":[203],"shown":[210],"fully":[212],"avoid":[213],"degradation,":[216],"robust":[218],"mixed":[219],"enabled":[223],"for":[224],"without":[228],"extra":[229],"overheads.":[230]},"counts_by_year":[{"year":2025,"cited_by_count":2}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}