UC Irvine

With technology advancement to the nanoscale level, 3D stacking of Integrated Circuits (ICs) provides significant advantages in saving device footprints, improving power management, and continuing performance enhancement which aim to increase energy efficiency and scalability particularly for many-core and NoCs systems. However, the benefits of these systems can be jeopardized because they became more subjected to elevated temperature induced by thermal management challenges as well as delay degradation (i.e. aging) caused by load imbalance.

On one hand, the elevated temperature in 3D systems can be solved by Thermal Through Silicon Vias (TTSVs). However, past research has either overestimated or underestimated the effects of TTSVs as a consequence of the lack of detailed 3D IC models or system-level simulations. In this dissertation, we propose a simulation flow to accurately simulate TTSV effects on 3D ICs. Furthermore, we present a hierarchical approach to optimize the floorplan of a 3D Nehalem-based multicore processor via Simulated Annealing (SA) with respect to the area, temperature, and wirelength. By using detailed 3D thermal model along with full system and validated thermal simulators, our results show accurate thermal analysis of 3D ICs. In addition, we found that the peak temperature of 3D IC such as 3D Nehalem is reduced with a minimal area overhead.

On the other hand, previous mitigation techniques to reduce aging phenomena effects on NoC system either ignore the runtime operating conditions or impose significant overheads to the system. Hence, this dissertation also presents an online monitoring method through a Centralized Aging Table (CAT) for routers in NoCs. Our methodology populates CAT by values that represent aging degradation for each different pairs of stress and temperature ranges during a given period of time. Moreover, utilizing CAT, we propose an online adaptive aging-aware routing algorithm in order to avoid highly aged routers in 2D NoC which eventually leads to age and load balancing between routers. We also extend this idea to 3D NoC by proposing AROMa, which is an aging-aware deadlock-free adaptive routing algorithm integrated with a novel online aging monitoring system for 3D NoCs. The monitoring system in AROMa exploits Distributed-Centralized-Aging-Table (D-CAT) to obtain routers' aging rates for each layer of 3D NoCs. Consequently, AROMa swaps between different k-best source-destination shortest paths periodically to avoid highly aged routers, force them in the recovery phase, and accordingly balance aging in the network. Our results demonstrate that our online routing algorithm and monitoring methodology improves delay degradation of maximum aged router and aging imbalance while ensuring that the impact of our proposed methodology on network latency, Energy-Delay-Product (EDP) and link utilization is negligible.