Software-Based Mitigation for Memory Address Decoder Aging
D.H.P. Kraak (TU Delft - Computer Engineering)
C.C. Gursoy (Tallinn University of Technology)
Innocent Agbo (TU Delft - Computer Engineering)
Mottagiallah Taouil (TU Delft - Computer Engineering)
Maksim Jenihhin (Tallinn University of Technology)
J. Raik (Tallinn University of Technology)
S. Hamdioui (TU Delft - Quantum & Computer Engineering)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Integrated circuits typically contain design margins to compensate for aging. As aging impact increases with technology scaling, bigger margins are necessary to achieve the desired reliability. However, these increased margins lead to a reduced performance and lower yield. Alternatively, mitigation schemes can be deployed to reduce the aging. This paper proposes a software-based method to mitigate the aging of the memory's address decoder logic due to Bias Temperature Instability. The method is based on periodically applying a rejuvenation application on top of a user application. The goal of the rejuvenation application is to recover aged transistors of the critical paths of the address decoder. The experimental results show that the proposed method significantly reduces aging in cases when applications consist of memory access patterns that result in an unbalanced stress in the address decoder logic. In particular, it reduces the degradation of the address decoder's setup delay by up to 43% with an execution overhead of only 1%.