<?xml version="1.0" encoding="utf-8"?>
<TEI xmlns="http://www.tei-c.org/ns/1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:hal="http://hal.archives-ouvertes.fr/" xmlns:gml="http://www.opengis.net/gml/3.3/" xmlns:gmlce="http://www.opengis.net/gml/3.3/ce" version="1.1" xsi:schemaLocation="http://www.tei-c.org/ns/1.0 http://api.archives-ouvertes.fr/documents/aofr-sword.xsd">
  <teiHeader>
    <fileDesc>
      <titleStmt>
        <title>HAL TEI export of hal-05019106v1</title>
      </titleStmt>
      <publicationStmt>
        <distributor>CCSD</distributor>
        <availability status="restricted">
          <licence target="https://creativecommons.org/publicdomain/zero/1.0/">CC0 1.0 - Universal</licence>
        </availability>
        <date when="2026-05-18T15:31:03+02:00"/>
      </publicationStmt>
      <sourceDesc>
        <p part="N">HAL API Platform</p>
      </sourceDesc>
    </fileDesc>
  </teiHeader>
  <text>
    <body>
      <listBibl>
        <biblFull>
          <titleStmt>
            <title xml:lang="en">BLAS-based Block Memory Accessor with Applications to Mixed PrecisionSparse Direct Solvers</title>
            <author role="aut">
              <persName>
                <forename type="first">Patrick</forename>
                <surname>Amestoy</surname>
              </persName>
              <email type="md5">ecfca3b5bafd2aba872382880b28e15e</email>
              <email type="domain">irit.fr</email>
              <idno type="idhal" notation="numeric">1090241</idno>
              <idno type="halauthorid" notation="string">2145058-1090241</idno>
              <idno type="IDREF">https://www.idref.fr/06167589X</idno>
              <affiliation ref="#struct-1070900"/>
            </author>
            <author role="crp">
              <persName>
                <forename type="first">Antoine</forename>
                <surname>Jego</surname>
              </persName>
              <email type="md5">ac1f56043285e8b248e0d681f194904a</email>
              <email type="domain">toulouse-inp.fr</email>
              <idno type="idhal" notation="string">antoine-jego-2</idno>
              <idno type="idhal" notation="numeric">1565540</idno>
              <idno type="halauthorid" notation="string">2260692-1565540</idno>
              <idno type="IDREF">https://www.idref.fr/275453308</idno>
              <idno type="ORCID">https://orcid.org/0000-0001-8682-3179</idno>
              <affiliation ref="#struct-301329"/>
              <affiliation ref="#struct-541706"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Jean-Yves</forename>
                <surname>L'Excellent</surname>
              </persName>
              <email type="md5">c538dd770d47554763c3e678445576d6</email>
              <email type="domain">ens-lyon.fr</email>
              <idno type="idhal" notation="string">jean-yves-lexcellent</idno>
              <idno type="idhal" notation="numeric">10316</idno>
              <idno type="halauthorid" notation="string">8127-10316</idno>
              <idno type="ORCID">https://orcid.org/0000-0001-5804-993X</idno>
              <idno type="IDREF">https://www.idref.fr/078828031</idno>
              <affiliation ref="#struct-1070900"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Theo</forename>
                <surname>Mary</surname>
              </persName>
              <email type="md5">af1e82da97a221c699f7b8f07bede748</email>
              <email type="domain">lip6.fr</email>
              <idno type="idhal" notation="string">tmary</idno>
              <idno type="idhal" notation="numeric">178018</idno>
              <idno type="halauthorid" notation="string">2568242-178018</idno>
              <idno type="ORCID">https://orcid.org/0000-0001-9949-4634</idno>
              <idno type="IDREF">https://www.idref.fr/230009417</idno>
              <idno type="RESEARCHERID">http://www.researcherid.com/rid/GAO-7221-2022</idno>
              <idno type="RESEARCHERID">http://www.researcherid.com/rid/http://www.researcherid.com/rid/GAO-7221-2022</idno>
              <affiliation ref="#struct-541706"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Grégoire</forename>
                <surname>Pichon</surname>
              </persName>
              <email type="md5">8da7daf0f4566ec6235494df774d51e8</email>
              <email type="domain">inria.fr</email>
              <idno type="idhal" notation="string">gregoire-pichon</idno>
              <idno type="idhal" notation="numeric">2558</idno>
              <idno type="halauthorid" notation="string">23393-2558</idno>
              <affiliation ref="#struct-1096051"/>
              <affiliation ref="#struct-194495"/>
              <affiliation ref="#struct-301329"/>
            </author>
            <editor role="depositor">
              <persName>
                <forename>antoine</forename>
                <surname>jego</surname>
              </persName>
              <email type="md5">61441cb3ac47998796dfeb3a1d8a69e3</email>
              <email type="domain">lip6.fr</email>
            </editor>
            <funder ref="#projanr-109517"/>
            <funder ref="#projanr-98377"/>
            <funder ref="#projanr-51775"/>
          </titleStmt>
          <editionStmt>
            <edition n="v1" type="current">
              <date type="whenSubmitted">2025-04-22 15:18:10</date>
              <date type="whenWritten">2025-04</date>
              <date type="whenModified">2026-05-18 15:30:12</date>
              <date type="whenReleased">2025-04-22 16:19:19</date>
              <date type="whenProduced">2025-04</date>
              <date type="whenEndEmbargoed">2025-04-22</date>
              <ref type="file" target="https://hal.science/hal-05019106v1/document">
                <date notBefore="2025-04-22"/>
              </ref>
              <ref type="file" subtype="author" n="1" target="https://hal.science/hal-05019106v1/file/manuscript.pdf" id="file-5042641-4366247">
                <date notBefore="2025-04-22"/>
              </ref>
            </edition>
            <edition n="v2">
              <date type="whenSubmitted">2026-04-16 18:51:24</date>
            </edition>
            <edition n="v3">
              <date type="whenSubmitted">2026-05-01 15:20:49</date>
            </edition>
            <respStmt>
              <resp>contributor</resp>
              <name key="1033344">
                <persName>
                  <forename>antoine</forename>
                  <surname>jego</surname>
                </persName>
                <email type="md5">61441cb3ac47998796dfeb3a1d8a69e3</email>
                <email type="domain">lip6.fr</email>
              </name>
            </respStmt>
          </editionStmt>
          <publicationStmt>
            <distributor>CCSD</distributor>
            <idno type="halId">hal-05019106</idno>
            <idno type="halUri">https://hal.science/hal-05019106</idno>
            <idno type="halBibtex">amestoy:hal-05019106</idno>
            <idno type="halRefHtml">2025</idno>
            <idno type="halRef">2025</idno>
            <availability status="restricted">
              <licence target="https://about.hal.science/hal-authorisation-v1/">HAL Authorization<ref corresp="#file-5042641-4366247"/></licence>
            </availability>
          </publicationStmt>
          <seriesStmt/>
          <notesStmt>
            <note type="audience" n="2">International</note>
          </notesStmt>
          <sourceDesc>
            <biblStruct>
              <analytic>
                <title xml:lang="en">BLAS-based Block Memory Accessor with Applications to Mixed PrecisionSparse Direct Solvers</title>
                <author role="aut">
                  <persName>
                    <forename type="first">Patrick</forename>
                    <surname>Amestoy</surname>
                  </persName>
                  <email type="md5">ecfca3b5bafd2aba872382880b28e15e</email>
                  <email type="domain">irit.fr</email>
                  <idno type="idhal" notation="numeric">1090241</idno>
                  <idno type="halauthorid" notation="string">2145058-1090241</idno>
                  <idno type="IDREF">https://www.idref.fr/06167589X</idno>
                  <affiliation ref="#struct-1070900"/>
                </author>
                <author role="crp">
                  <persName>
                    <forename type="first">Antoine</forename>
                    <surname>Jego</surname>
                  </persName>
                  <email type="md5">ac1f56043285e8b248e0d681f194904a</email>
                  <email type="domain">toulouse-inp.fr</email>
                  <idno type="idhal" notation="string">antoine-jego-2</idno>
                  <idno type="idhal" notation="numeric">1565540</idno>
                  <idno type="halauthorid" notation="string">2260692-1565540</idno>
                  <idno type="IDREF">https://www.idref.fr/275453308</idno>
                  <idno type="ORCID">https://orcid.org/0000-0001-8682-3179</idno>
                  <affiliation ref="#struct-301329"/>
                  <affiliation ref="#struct-541706"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Jean-Yves</forename>
                    <surname>L'Excellent</surname>
                  </persName>
                  <email type="md5">c538dd770d47554763c3e678445576d6</email>
                  <email type="domain">ens-lyon.fr</email>
                  <idno type="idhal" notation="string">jean-yves-lexcellent</idno>
                  <idno type="idhal" notation="numeric">10316</idno>
                  <idno type="halauthorid" notation="string">8127-10316</idno>
                  <idno type="ORCID">https://orcid.org/0000-0001-5804-993X</idno>
                  <idno type="IDREF">https://www.idref.fr/078828031</idno>
                  <affiliation ref="#struct-1070900"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Theo</forename>
                    <surname>Mary</surname>
                  </persName>
                  <email type="md5">af1e82da97a221c699f7b8f07bede748</email>
                  <email type="domain">lip6.fr</email>
                  <idno type="idhal" notation="string">tmary</idno>
                  <idno type="idhal" notation="numeric">178018</idno>
                  <idno type="halauthorid" notation="string">2568242-178018</idno>
                  <idno type="ORCID">https://orcid.org/0000-0001-9949-4634</idno>
                  <idno type="IDREF">https://www.idref.fr/230009417</idno>
                  <idno type="RESEARCHERID">http://www.researcherid.com/rid/GAO-7221-2022</idno>
                  <idno type="RESEARCHERID">http://www.researcherid.com/rid/http://www.researcherid.com/rid/GAO-7221-2022</idno>
                  <affiliation ref="#struct-541706"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Grégoire</forename>
                    <surname>Pichon</surname>
                  </persName>
                  <email type="md5">8da7daf0f4566ec6235494df774d51e8</email>
                  <email type="domain">inria.fr</email>
                  <idno type="idhal" notation="string">gregoire-pichon</idno>
                  <idno type="idhal" notation="numeric">2558</idno>
                  <idno type="halauthorid" notation="string">23393-2558</idno>
                  <affiliation ref="#struct-1096051"/>
                  <affiliation ref="#struct-194495"/>
                  <affiliation ref="#struct-301329"/>
                </author>
              </analytic>
              <monogr>
                <imprint/>
              </monogr>
            </biblStruct>
          </sourceDesc>
          <profileDesc>
            <langUsage>
              <language ident="en">English</language>
            </langUsage>
            <textClass>
              <keywords scheme="author">
                <term xml:lang="en">Numerica approximation algorithms</term>
                <term xml:lang="en">Single instruction multiple data</term>
                <term xml:lang="en">Solvers</term>
                <term xml:lang="en">Mathematical software performance</term>
              </keywords>
              <classCode scheme="https://dl.acm.org/ccs" n="ACM2012.C.0.1.5"/>
              <classCode scheme="https://dl.acm.org/ccs" n="ACM2012.F.4.1.6"/>
              <classCode scheme="https://dl.acm.org/ccs" n="ACM2012.G.2.0"/>
              <classCode scheme="https://dl.acm.org/ccs" n="ACM2012.G.2.2"/>
              <classCode scheme="halDomain" n="info.info-pf">Computer Science [cs]/Performance [cs.PF]</classCode>
              <classCode scheme="halDomain" n="info.info-ms">Computer Science [cs]/Mathematical Software [cs.MS]</classCode>
              <classCode scheme="halTypology" n="UNDEFINED">Preprints, Working Papers, ...</classCode>
              <classCode scheme="halOldTypology" n="UNDEFINED">Preprints, Working Papers, ...</classCode>
              <classCode scheme="halTreeTypology" n="UNDEFINED">Preprints, Working Papers, ...</classCode>
            </textClass>
            <abstract xml:lang="en">
              <p>Mixed precision “memory accessor” approaches decouple storage and compute precisions (data is stored and accessed in low precision,but computations are kept in higher precision) to reduce data accesses, improve accuracy, and simplify programming. In this work, wedevelop such a memory accessor aimed at accelerating sparse direct solvers and propose several new improvements. In particular,we propose a BLAS-based, block approach that can directly rely on BLAS libraries for efficiency and portability. When consideringBLAS-2 memory-bound operations like triangular solves, we observe that the performance adequately matches the storage cost, inmultiple parallel settings, provided that the conversion from storage to compute precision is efficient, and that the block size is suitablychosen. For the storage precision, we leverage custom floating-point types unsupported by hardware, and we take advantage of therecent AVX512-VBMI instruction set to reach an improved efficiency. We also consider rank-structured matrix representations suchas the block low-rank (BLR) format, and explain how to optimize the memory accessor for such matrices. We present preliminaryperformance experiments using the sparse direct solver MUMPS with adaptive precision BLR approximations. Our results confirm thepotential of these memory accessor approaches to achieve efficiency while optimizing storage.</p>
            </abstract>
          </profileDesc>
        </biblFull>
      </listBibl>
    </body>
    <back>
      <listOrg type="structures">
        <org type="institution" xml:id="struct-1070900" status="VALID">
          <orgName>Mumps Technologies [Lyon]</orgName>
          <desc>
            <address>
              <addrLine>15 parvis René Descartes BP 7000 69342 Lyon Cedex 07</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://mumps.enseeiht.fr/</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-301329" status="VALID">
          <orgName>LIP</orgName>
          <desc>
            <address>
              <country key="FR"/>
            </address>
          </desc>
        </org>
        <org type="researchteam" xml:id="struct-541706" status="VALID">
          <orgName>Performance et Qualité des Algorithmes Numériques</orgName>
          <orgName type="acronym">PEQUAN</orgName>
          <date type="start">2018-01-01</date>
          <desc>
            <address>
              <country key="FR"/>
            </address>
          </desc>
          <listRelation>
            <relation active="#struct-541703" type="direct"/>
            <relation active="#struct-413221" type="indirect"/>
            <relation name="UMR7606" active="#struct-441569" type="indirect"/>
          </listRelation>
        </org>
        <org type="laboratory" xml:id="struct-1096051" status="VALID">
          <idno type="RNSR">202224236C</idno>
          <idno type="ROR">https://ror.org/022gakr41</idno>
          <orgName>Centre Inria de Lyon</orgName>
          <desc>
            <address>
              <addrLine>Inria de Lyon / Bâtiment CEI-2 / Campus La Doua - Villeurbanne / 56, Boulevard Niels Bohr / CS 52132 / 69 603 Villeurbanne</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.inria.fr/fr/centre-inria-de-lyon</ref>
          </desc>
          <listRelation>
            <relation active="#struct-300009" type="direct"/>
          </listRelation>
        </org>
        <org type="institution" xml:id="struct-194495" status="VALID">
          <idno type="IdRef">026402823</idno>
          <idno type="ISNI">0000000121686185</idno>
          <idno type="ROR">https://ror.org/029brtt94</idno>
          <orgName>Université Claude Bernard Lyon 1</orgName>
          <orgName type="acronym">UCBL</orgName>
          <desc>
            <address>
              <addrLine>43, boulevard du 11 novembre 1918, 69622 Villeurbanne cedex</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.univ-lyon1.fr/</ref>
          </desc>
          <listRelation>
            <relation active="#struct-301088" type="direct"/>
          </listRelation>
        </org>
        <org type="laboratory" xml:id="struct-541703" status="VALID">
          <idno type="IdRef">13558292X</idno>
          <idno type="RNSR">199712651U</idno>
          <idno type="ROR">https://ror.org/05krcen59</idno>
          <orgName>LIP6</orgName>
          <date type="start">2018-01-01</date>
          <desc>
            <address>
              <addrLine>4 Place JUSSIEU 75252 PARIS CEDEX 05</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.lip6.fr/</ref>
          </desc>
          <listRelation>
            <relation active="#struct-413221" type="direct"/>
            <relation name="UMR7606" active="#struct-441569" type="direct"/>
          </listRelation>
        </org>
        <org type="regroupinstitution" xml:id="struct-413221" status="VALID">
          <idno type="IdRef">221333754</idno>
          <idno type="ROR">https://ror.org/02en5vm52</idno>
          <orgName>Sorbonne Université</orgName>
          <orgName type="acronym">SU</orgName>
          <date type="start">2018-01-01</date>
          <desc>
            <address>
              <addrLine>21 rue de l’École de médecine - 75006 Paris</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.sorbonne-universite.fr/</ref>
          </desc>
        </org>
        <org type="regroupinstitution" xml:id="struct-441569" status="VALID">
          <idno type="IdRef">02636817X</idno>
          <idno type="ISNI">0000000122597504</idno>
          <idno type="ROR">https://ror.org/02feahw73</idno>
          <orgName>Centre National de la Recherche Scientifique</orgName>
          <orgName type="acronym">CNRS</orgName>
          <date type="start">1939-10-19</date>
          <desc>
            <address>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.cnrs.fr/</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-300009" status="VALID">
          <idno type="ROR">https://ror.org/02kvxyf05</idno>
          <orgName>Institut National de Recherche en Informatique et en Automatique</orgName>
          <orgName type="acronym">Inria</orgName>
          <desc>
            <address>
              <addrLine>Domaine de VoluceauRocquencourt - BP 10578153 Le Chesnay Cedex</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.inria.fr/en/</ref>
          </desc>
        </org>
        <org type="regroupinstitution" xml:id="struct-301088" status="VALID">
          <idno type="ROR">https://ror.org/01rk35k63</idno>
          <orgName>Université de Lyon</orgName>
          <desc>
            <address>
              <addrLine>92 rue Pasteur - CS 30122, 69361 Lyon Cedex 07</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.universite-lyon.fr/</ref>
          </desc>
        </org>
      </listOrg>
      <listOrg type="projects">
        <org type="anrProject" xml:id="projanr-109517" status="VALID">
          <idno type="anr">ANR-23-CE46-0005</idno>
          <orgName>MixHPC</orgName>
          <desc>Algorithmes en précision mixte pour le calcul haute performance</desc>
          <date type="start">2023</date>
        </org>
        <org type="anrProject" xml:id="projanr-98377" status="VALID">
          <idno type="anr">ANR-22-EXNU-0002</idno>
          <orgName>Exa-MA</orgName>
          <desc>Methods and Algorithms for Exascale</desc>
          <date type="start">2022</date>
        </org>
        <org type="anrProject" xml:id="projanr-51775" status="VALID">
          <idno type="anr">ANR-20-CE46-0009</idno>
          <orgName>INTERFLOP</orgName>
          <desc>Plateforme d'analyse pour l'arithmétique flottante</desc>
          <date type="start">2020</date>
        </org>
      </listOrg>
    </back>
  </text>
</TEI>