On the Arithmetic Intensity of Distributed-Memory Dense Matrix Multiplication Involving a Symmetric Input Matrix (SYMM)
hal.structure.identifier | COmposabilité Numerique and parallèle pour le CAlcul haute performanCE [CONCACE] | |
dc.contributor.author | AGULLO, Emmanuel | |
hal.structure.identifier | Algorithmes Parallèles et Optimisation [IRIT-APO] | |
hal.structure.identifier | Centre National de la Recherche Scientifique [CNRS] | |
dc.contributor.author | BUTTARI, Alfredo | |
hal.structure.identifier | COmposabilité Numerique and parallèle pour le CAlcul haute performanCE [CONCACE] | |
dc.contributor.author | COULAUD, Olivier | |
hal.structure.identifier | Laboratoire Bordelais de Recherche en Informatique [LaBRI] | |
hal.structure.identifier | Outils et Optimisations pour le Calcul Haute Performance et l'Apprentissage [TOPAL] | |
dc.contributor.author | EYRAUD-DUBOIS, Lionel | |
hal.structure.identifier | Outils et Optimisations pour le Calcul Haute Performance et l'Apprentissage [TOPAL] | |
dc.contributor.author | FAVERGE, Mathieu | |
hal.structure.identifier | Biodiversité, Gènes & Communautés [BioGeCo] | |
hal.structure.identifier | Pleiade, from patterns to models in computational biodiversity and biotechnology [PLEIADE] | |
dc.contributor.author | FRANC, Alain | |
hal.structure.identifier | Outils et Optimisations pour le Calcul Haute Performance et l'Apprentissage [TOPAL] | |
hal.structure.identifier | Laboratoire Bordelais de Recherche en Informatique [LaBRI] | |
dc.contributor.author | GUERMOUCHE, Abdou | |
hal.structure.identifier | Algorithmes Parallèles et Optimisation [IRIT-APO] | |
hal.structure.identifier | Institut National Polytechnique (Toulouse) [Toulouse INP] | |
dc.contributor.author | JEGO, Antoine | |
hal.structure.identifier | COmposabilité Numerique and parallèle pour le CAlcul haute performanCE [CONCACE] | |
dc.contributor.author | PERESSONI, Romain | |
hal.structure.identifier | Service Expérimentation et Développement [Bordeaux] [SED] | |
dc.contributor.author | PRUVOST, Florent | |
dc.contributor.editor | IEEE | |
dc.date.created | 2023 | |
dc.date.issued | 2023-06 | |
dc.date.conference | 2023-05-15 | |
dc.description.abstractEn | Dense matrix multiplication involving a symmetric input matrix (SYMM) is implemented in reference distributed-memory codes with the same data distribution as its general analogue (GEMM). We show that, when the symmetric matrix is dominant, such a 2D block-cyclic (2D BC) scheme leads to a lower arithmetic intensity (AI) of SYMM than that of GEMM by a factor of 2. We propose alternative data distributions preserving the memory benefit of SYMM of storing only half of the matrix while achieving up to the same AI as GEMM. We also show that, in the case we can afford the same memory footprint as GEMM, SYMM can achieve a higher AI. We propose a task-based design of SYMM independent of the data distribution. This design allows for scalable A-stationary SYMM with which all discussed data distributions, may they be very irregular, can be easily assessed. We have integrated the resulting code in a reduction dimension algorithm involving a randomized singular value decomposition dominated by SYMM. An experimental study shows a compelling impact on performance. | |
dc.description.sponsorship | Solveurs pour architectures hétérogènes utilisant des supports d'exécution, objectif scalabilité - ANR-19-CE46-0009 | |
dc.language.iso | en | |
dc.rights.uri | http://creativecommons.org/licenses/by/ | |
dc.source.title | International Parallel and Distributed Processing Symposium | |
dc.subject.en | Matrix multiplication | |
dc.subject.en | SYMM | |
dc.subject.en | GEMM | |
dc.subject.en | 2DBC | |
dc.subject.en | task-based programming | |
dc.subject.en | Symmetric | |
dc.subject.en | SBC | |
dc.subject.en | TBC | |
dc.subject.en | 3D | |
dc.subject.en | 2.5D | |
dc.title.en | On the Arithmetic Intensity of Distributed-Memory Dense Matrix Multiplication Involving a Symmetric Input Matrix (SYMM) | |
dc.type | Communication dans un congrès | |
dc.subject.hal | Informatique [cs]/Calcul parallèle, distribué et partagé [cs.DC] | |
dc.subject.hal | Informatique [cs]/Bio-informatique [q-bio.QM] | |
bordeaux.page | 357-367 | |
bordeaux.conference.title | IPDPS 2023 - 37th International Parallel and Distributed Processing Symposium | |
bordeaux.country | US | |
bordeaux.title.proceeding | International Parallel and Distributed Processing Symposium | |
bordeaux.conference.city | St. Petersburg, FL | |
bordeaux.peerReviewed | oui | |
hal.identifier | hal-04093162 | |
hal.version | 1 | |
hal.invited | non | |
hal.proceedings | oui | |
hal.conference.organizer | IEEE | |
hal.conference.end | 2023-05-19 | |
hal.popular | non | |
hal.audience | Internationale | |
hal.origin.link | https://hal.archives-ouvertes.fr//hal-04093162v1 | |
bordeaux.COinS | ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.btitle=International%20Parallel%20and%20Distributed%20Processing%20Symposium&rft.date=2023-06&rft.spage=357-367&rft.epage=357-367&rft.au=AGULLO,%20Emmanuel&BUTTARI,%20Alfredo&COULAUD,%20Olivier&EYRAUD-DUBOIS,%20Lionel&FAVERGE,%20Mathieu&rft.genre=unknown |
Fichier(s) constituant ce document
Fichiers | Taille | Format | Vue |
---|---|---|---|
Il n'y a pas de fichiers associés à ce document. |