Tech. Sci. Informatiques最新文献

{"title":"Reconvergence de contrôle implicite pour les architectures SIMT","authors":"Nicolas Brunie, Caroline Collange","doi":"10.3166/tsi.32.153-178","DOIUrl":"https://doi.org/10.3166/tsi.32.153-178","url":null,"abstract":"Les architectures paralleles qui obeissent au modele SIMT telles que les GPU tirent parti de la regularite des applications en executant plusieurs threads concurrents sur des unites SIMD de maniere synchrone. Lorsque les threads empruntent des chemins divergents dans le graphe de flot de controle, leur execution est sequentialisee jusqu'au prochain point de convergence. La reconvergence doit etre effectuee au plus tot de maniere a maximiser l'occupation des unites SIMD. Nous proposons dans cet article deux techniques permettant de traiter la divergence de controle en SIMT et d'identifier dynamiquement les points de reconvergence, dont une qui opere en espace constant et gere les sauts indirects et la recursivite. Nous evaluons une realisation materielle consistant a partager le materiel existant de l'unite de gestion de la divergence memoire. En termes de performances, cette solution est au moins aussi efficace que les techniques de l'etat de l'art employes par les GPU actuels.","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127238403","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Méthodologie dédiée aux applications parallèles sur plateforme reconfigurable dynamiquement","authors":"C. Foucher, F. Muller, A. Giulieri","doi":"10.3166/tsi.32.253-280","DOIUrl":"https://doi.org/10.3166/tsi.32.253-280","url":null,"abstract":"L'integration croissante de ressources materielles reconfigurables dans les systemes de calcul ouvre de nombreuses voies en permettant d'implementer des accelerateurs materiels dynamiquement. Neanmoins, l'utilisation de ces circuits reste complexe avec les outils actuels, notamment dans le cas de la reconfiguration partielle. C'est pourquoi nous proposons dans cet article une methodologie destinee a faciliter la creation d'applications faisant usage de materiel reconfigurable, et particulierement des applications paralleles. Nous introduisons une couche de virtualisation destinee a isoler l'application de la plateforme d'execution, qui permet notamment d'assurer la portabilite des applications sur divers supports. Nous detaillons egalement la structure modulaire que nous proposons pour les applications, qui facilite la reutilisation d'IP deja existants lors de la conception d'une application. Par ailleurs, nous presentons plusieurs implementations de plateformes destinees a l'execution de telles applications.","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"113 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124328240","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Une approche pour la simulation de comportements émotionnels fondée sur la théorie de la conservation des ressources","authors":"Sabrina Campano, Etienne de Sevin, Vincent Corruble, N. Sabouret","doi":"10.3166/tsi.32.113-138","DOIUrl":"https://doi.org/10.3166/tsi.32.113-138","url":null,"abstract":"The simulation of believable behaviours for virtual agents requires to take human factors such as emotions into account. Most computational models dealing with this subject in- clude emotion categories in their architecture. However, determining the number of categories to use and their influence on behaviour is a difficult task. In this paper, we present a model without emotion categories, centered around the theory of Conservation of Resources (Hobfoll, 1989). Each agent can acquire or protect resources, and behaviour choice depends on resources state, as well as agent's needs and preferences. We also define an evaluation protocol based on verbal reports made by human observers, in order to assess if the produced behaviours are recognized as emotional and believable. Our first results tend to indicate that the proposed model is able to produce believable and emotional behaviours. RESUME. La simulation de comportements credibles chez les agents virtuels demande de prendre en compte des facteurs humains tels que les emotions. La plupart des travaux traitant ce sujet proposent des architectures incluant des categories d'emotions. Cependant, determiner le nom- bre d'emotions a utiliser et leur influence sur le comportement d'un agent est une tâche difficile. Dans cet article, nous presentons un modele pour la simulation de comportements emotionnels fonde sur la theorie de la Conservation des Ressources (Hobfoll, 1989) qui n'utilise pas de categories d'emotions. Un agent peut acquerir et proteger des ressources, et le choix de son comportement depend de l'etat des ressources, de ses besoins, et de ses preferences. Nous presentons egalement un protocole destine a evaluer si les comportements des agents sont percus comme credibles et emotionnels par des observateurs humains. Les resultats obtenus","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121655728","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Extraction de relations complexes. Application à des résultats expérimentaux en physiologie rénale","authors":"Annie Minard, B. Grau, Anne-Laure Ligozat, S. R. Thomas","doi":"10.3166/tsi.32.77-111","DOIUrl":"https://doi.org/10.3166/tsi.32.77-111","url":null,"abstract":"Dans cet article, nous presentons un modele d'extraction d'information permettant d'alimenter semi-automatiquement une base de donnees avec des resultats d'experimentations, avec comme cadre d'application la physiologie renale quantitative. L'extraction d'information est guidee par une ressource termino-ontologique qui exploite des patrons d'extraction et un lexique. Elle a ete integree dans un assistant guidant l'expert pour remplir la base de donnees. Nous avons evalue le systeme d'extraction d'information intrinsequement et extrinsequement, et montrons que le peuplement de la base de donnees par les experts est rendu plus rapide et plus complet grâce a l'outil propose.","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133577575","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Évaluation quantitative de la sécurité. Approche basée sur les vulnérabilités","authors":"Géraldine Vache Marconato, Vincent Nicomette, M. Kaâniche","doi":"10.3166/tsi.32.41-75","DOIUrl":"https://doi.org/10.3166/tsi.32.41-75","url":null,"abstract":"Les travaux presentes dans cet article sont consacres a la securisation des systemes d'information en se basant sur des mesures quantitatives. Les mesures ont pour but d'aider a la prevision des risques et de fournir les informations necessaires permettant d'assurer le niveau de securite du systeme en operation. Dans cette approche, nous prenons en consideration des facteurs externes ayant un impact significatif sur la securite du systeme. Nous avons identifie trois facteurs, lies au processus d'exploitation d'une vulnerabilite : le cycle de vie de la vul-nerabilite, le comportement des attaquants et le comportement de l'administrateur du systeme. Nous avons etudie les dependances entre ces differents facteurs et comment leur evolution pou-vait impacter la securite du systeme. A partir de cette etude, nous avons defini des mesures quantitatives prenant en compte ces facteurs externes et avons modelise, en utilisant le forma-lisme des SAN (Stochastic Activity Networks), le processus d'exploitation de la vulnerabilite qui conduit a la compromission du systeme. Nous avons distingue deux scenarios selon que la vulnerabilite est decouverte par une personne malveillante ou non malveillante. En analysant une base de donnees de vulnerabilites, nous avons caracterise la probabilite d'occurrence de plusieurs evenements du cycle de vie de la vulnerabilite, que nous utilisons pour l'obtention de nos mesures. ABSTRACT. The objective of this work is the evaluation of information systems security using quantitative measures. These measures aim at forecasting risks and providing information to monitor the security level of the system in operation. In our approach, we take into account some environmental factors that have a significant impact on the security of the system. We have identified three such factors that are related to the vulnerability exploitation process: the vulnerability life cycle, the behavior of the attackers and the behavior of the system administrator. We have studied the dependencies between these factors and how the evolution of these factors could impact the system security. From this study, we have defined quantitative security measures taking into account these environmental factors and we have developed a model 1 re soumission a Technique et science informatiques, le 22/03/2011","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117353494","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Du traçage de l'activité informatique à sa redocumentation en texte. Une approche narrative et un outil auteur","authors":"L. Yahiaoui, Yannick Prié, Zizette Boufaïda","doi":"10.3166/tsi.32.7-40","DOIUrl":"https://doi.org/10.3166/tsi.32.7-40","url":null,"abstract":"Dans ce travail, nous presentons une approche generique pour le processus de redocumentation de l’activite mediee informatiquement. Cette approche combine le contenu de traces de l’activite et les connaissances des utilisateurs dans des formes narratives textuelles, facilement echangeables, decrivant l’activite de facon personnalisee pour les utilisateurs. Elle utilise des modeles formels pour les traces et les documents produits, ainsi que les principes de la theorie de la structure rhetorique(RST). Elle soutient la redocumentation a travers deux phases : une phase automatique pour generer un document texte initial et une phase interactive pour personnaliser ce document. Nous presentons notre outil de redocumentation ActRedoc, une implementation de notre approche, ainsi que les resultats de sa premiere evaluation.","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129558155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Politiques de gestion de protections pour l'implémentation de sections critiques","authors":"Emmanuel Sifakis, L. Mounier","doi":"10.3166/tsi.31.1153-1181","DOIUrl":"https://doi.org/10.3166/tsi.31.1153-1181","url":null,"abstract":"L’exploitation des architectures multicœurs repose sur une bonne gestion du parallélisme à l’intérieur des applications. Une telle gestion nécessite notamment d’éviter les conflits d’accès (ou race conditions) aux variables partagées entre plusieurs processus. Ceci peut se faire par l’introduction de sections critiques destinées à protéger les accès à ces variables. L’exécution atomique de ces sections est alors garantie soit par des mémoires transactionnelles, soit par des primitives de synchronisation de type mutex. Il apparait toutefois que, dans ce deuxième cas, ces verrous de synchronisation ne sont pas toujours utilisés de manière optimale par les applications. Nous proposons dans ce travail différentes politiques de gestion des verrous qui ont pour objectif d’améliorer les temps d’exécution. Nous présentons également une bibliothèque de gestion de verrous dédiée, qui permet l’acquisition simultanée de plusieurs objets de synchronisation, et qui offre de meilleures performances que des implémentations bien établies comme les pthreads. Ces résultats sont validés expérimentalement.","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125572452","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Les riches heures de l'ordonnancement","authors":"D. Trystram","doi":"10.3166/tsi.31.1021-1047","DOIUrl":"https://doi.org/10.3166/tsi.31.1021-1047","url":null,"abstract":"Parallelizing means executing simultaneously multiple operations. The efficient management of the assignment of tasks on available computing resources is a challenging problem. Whatever the considered execution support, this problem is crucial at any level from basic instructions, library components to full applications. Moreover, it can be studied from various points of view (users, system administrators). Scheduling is the main part for efficient resource management. It consists in determining where and when to execute the tasks while optimizing one or several objectives under hardware constraints. This is an old problem which arised with the first vector machines in the early seventies. Today, it is still a challenging problem since many new constraints should be considered. The purpose of this paper is to review the main constraints and to show how to take them into account in the scheduling policies. We defend a position of well-founded approaches for designing scheduling algorithms that allow to obtain guaranteed performances at a reasonable cost.","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126728433","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Traitement d'images sur architectures parallèles et hétérogènes","authors":"Sidi Ahmed Mahmoudi, Pierre Manneback, Cédric Augonnet, Samuel Thibault","doi":"10.3166/tsi.31.1183-1203","DOIUrl":"https://doi.org/10.3166/tsi.31.1183-1203","url":null,"abstract":"ABSTRACT. Image processing algorithms present a necessary tool for various domains relatedto computer vision. These algorithms are hampered by their high consumption of computingtimes when processing large sets of high resolution images. In this work, we propose a deve-lopment scheme enabling an efficient exploitation of parallel (GPU) and heterogeneous (Multi-CPU/Multi-GPU) platforms, in order to improve performance of image processing algorithms.The proposed scheme enables an efficient scheduling of hybrid tasks and an effective manage-ment of heterogeneous memories. We present also parallel and hybrid implementations of edgeand corner detection methods. Experimental results showed a global speedup ranging from 5to 25, when processing different sets of images, by comparison with CPU implementations. MOTS-CLES : calcul heterogene, GPU, traitement d’images, detection des coins et contours. KEYWORDS: heterogeneous computing, GPU, image processing, corner and edge detection. DOI:10.3166/TSI.31.1183-1203 c 2012 Lavoisier","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128517981","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"HydEE, vers un protocole de recouvrement arrière hiérarchique pour les machines exascales. De l'exploitation du déterminisme des émissions dans les protocoles de recouvrement arrière","authors":"Amina Guermouche, Thomas Ropars, F. Cappello","doi":"10.3166/tsi.31.1049-1078","DOIUrl":"https://doi.org/10.3166/tsi.31.1049-1078","url":null,"abstract":"The move towards exascale super-computers requires new fault tolerance solutions. Regarding parallel message passing applications, existing rollback-recovery protocols are not suited. To be able to deal with very large scale applications and high failure rate, a protoTechnique et science informatiques 2 Technique et science informatiques col should be able to confine failures consequences to a small subset of the processes, while providing good failure free performance, and logging a limited amount of data, especially in memory. To fulfill these needs, we propose HydEE, a hierarchical rollback-recovery protocol that combines coordinated checkpointing and message logging. HydEE leverages the senddeterminism of scienfitic parallel applications to tolerate multiple failures without relying on a stable storage. Our experiments show that for most applications, saving less than 15% of the messages payload in memory is enough to limit the rollbacks after a failure to less than 15% of the processes. MOTS-CLÉS : Calcul haute performances, applications parallèles, MPI, tolérance aux fautes, recouvrement arrière, protocoles hiérarchiques","PeriodicalId":109795,"journal":{"name":"Tech. Sci. Informatiques","volume":"110 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126063489","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}