Analyzing the Error Propagation in a Parameterizable Network-on-Chip Router - LIRMM - Laboratoire d’Informatique, de Robotique et de Microélectronique de Montpellier Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Analyzing the Error Propagation in a Parameterizable Network-on-Chip Router

Résumé

The constant reduction in the components size in integrated circuits and the increase of the operating frequency make Systems-on-Chip (SoCs) more vulnerable to noise and other interference phenomena. Such phenomena can lead to faults, which generate errors that may result in a system crash. SoCs with dozens of cores use Networks-on-Chip as their interconnection architecture. In this context, this work presents an analysis of the error propagation in a parameterizable router architecture that allows different combinations of input and output controllers, data width and buffers depth. The router has been described focusing on design flexibility and low logical resource occupation. We elaborated different combinations of the router architecture and evaluated the error propagation by means of Single Event Upset fault injections. The synthesis results presented a reasonable increase in term of logical elements when applying wider data representations, and combinations using Mealy finite state machines in the routing component had the lowest error propagation rate at a price of a small degradation in the maximum operating frequency.
Fichier principal
Vignette du fichier
2019_LATS___Analyzing_the_Error_Propagation_in_a_Parameterizable_Network_on_Chip_Router___HAL_Version.pdf (312.03 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

lirmm-02008414 , version 1 (29-09-2021)

Identifiants

Citer

Douglas Rossi de Melo, Cesar Zeferino, Luigi Dilillo, Eduardo Augusto Bezerra. Analyzing the Error Propagation in a Parameterizable Network-on-Chip Router. LATS 2019 - 20th IEEE Latin American Test Symposium, Mar 2019, Santiago, Chile. ⟨10.1109/LATW.2019.8704580⟩. ⟨lirmm-02008414⟩
137 Consultations
64 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More