A possible solution to the reliability challenges of new fabrication technologies is self-test and self-reconfiguration with no or limited external control. This necessitates the inclusion of mechanisms for detection, recovery and reconfiguration, providing the chip with self-healing capability. While appropriate mechanisms for system recovery and reconfiguration exist, detection remains to be a challenge in realization of selfhealing and graceful degradable many-core processors. In this paper we propose a scalable test architecture to distribute test stimuli among homogeneous processing cores in a many-core processor. We have incorporated test infrastructure in the architecture that periodically suspends normal operation of processing cores and applies test to them. This procedure is performed in an online fashion without any downtime visible to the end user. Our proposed fault detection mechanism can be accompanied with appropriate recovery and reconfiguration techniques in order to realize a many-core processor with selfhealing capability.
Due to worsening aging effects and incomplete testing and verification processes, systems being built in new fabrication technologies have encountered serious reliability challenges. A promising solution to these reliability challenges is self-test and self-reconfiguration with no or limited external control. In this paper, a scalable self-test mechanism for online testing of many-core processors has been proposed. Several hardware components are incorporated in the many-core architecture that distribute software test routines among the processing cores, monitor behavior of the processing cores during test routine execution, and detect faulty cores. A merit-based probabilistic test generation (MPTG) method to generate test for register-transfer level components considering the limitations imposed by the neighboring components has been proposed. In addition, a test generation approach utilizing MPTG has been proposed for software test routine generation in this environment. Experimental results show that the proposed test generation method results in good stuck-at fault coverage in a limited number of test cycles. In addition, the proposed test mechanism is extensively scalable in terms of hardware and timing overhead making it applicable to many-cores with a large number of processing cores. Résumé-En raison de l'aggravation des effets de détérioration des processus de test et de vérification incomplets, les systèmes en cours de construction avec les nouvelles technologies de fabrication sont exposés à des problèmes de fiabilité importants. Une solution prometteuse à ces défis de fiabilité est l'autotest et l'autoreconfiguration avec pas ou peu de contrôle externe. Dans cet article, un mécanisme d'autotest extensible pour les tests en ligne de processeurs Multi-Coeurs a été proposé. Plusieurs composants matériels sont incorporés dans l'architecture Multi-Coeurs à la fois distribuant des routines de test de logicielparmi les noyaux de traitement, surveillant le comportement des noyaux de traitement pendant l'essai d'exécution de routine, et détectant les noyaux défectueux. Une méthode fondée sur le mérite de génération de tests de probabilités (MGtP) afin de générer un test pour enregistrer et transférer des composants de niveau prenant en compte les limitations imposées par les composants voisins a été proposée. En outre, une approche de génération de tests utilisant MGtP a été proposée pour le test de logiciel génération de routine dans cet environnement. Les résultats expérimentaux montrent que la méthode proposée de génération de tests offre une couverture de défaut à un nombre limité de cycles d'essai. En outre, le mécanisme d'essai proposé est largement évolutif en terme de matériel et en terme du calendrier des frais généraux, le rendant ainsi applicable à des multi-coeurs dont un grand nombre destinés pour le traitement des données.
No abstract
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.