Abstract
The reliability of embedded processors fabricated using nanoscale technology nodes is threatened by accelerated transistor aging particularly, Bias Temperature Instability (BTI). In embedded memories such as instruction caches, BTI degrades the Static Noise Margin (SNM) of the memory cell, which in turn affects the stability of the stored value. Various bit flipping based solutions have been proposed to address BTI-induced aging of memory components. Nevertheless, the state-of-the-art techniques have considerable area and power overheads. In this paper, we propose an aging-aware instruction encoding technique to mitigate BTI-induced aging of instruction caches. Opcode, register and function code fields of an instruction are re-encoded so that the BTI-induced aging of the instruction cache is minimized. Simulation results show that the proposed technique achieves up to 40% SNM degradation improvement (equivalent to 47% MTTF improvement) with a negligible power overhead (0.1%).
Original language | English |
---|---|
Title of host publication | Proceedings of the 17th International Symposium on Quality Electronic Design, ISQED 2016 |
Editors | Peter Wright, Saibal Mukhopadhyay, Brian Cline |
Place of Publication | Piscataway, NJ |
Publisher | IEEE |
Pages | 325-330 |
Number of pages | 6 |
ISBN (Print) | 978-1-5090-1213-8 |
DOIs | |
Publication status | Published - 2016 |
Event | 17th International Symposium on Quality Electronic Design, ISQED 2016 - Santa Clara, CA, United States Duration: 14 Mar 2016 → 16 Mar 2016 |
Conference
Conference | 17th International Symposium on Quality Electronic Design, ISQED 2016 |
---|---|
Abbreviated title | ISQED 2016 |
Country/Territory | United States |
City | Santa Clara, CA |
Period | 14/03/16 → 16/03/16 |
Keywords
- Static Noise Margin
- Transistor Aging
- BTI
- SRAM