Through-silicon vias (TSVs) in 3D ICs show a significant power consumption, which can be reduced using coding techniques. This work presents an approach which reduces the TSV power consumption by a signal-aware bit assignment which includes inversions to exploit the MOS effect. The approach causes no overhead and results in a guaranteed reduction of the overall power consumption. An analysis of our technique shows a reduction in the TSV power consumption by up to 48 % for real correlated data streams (e.g. image sensor), and 11 % for low-power encoded random data streams.
1 Introduction
3D integration is a promising solution to overcome the challenges that arise with the limit of Moore’s law. To connect the dies of a 3D system on chip (3D SoC), through-silicon via (TSV) arrays are typically used as they yield to a short delay and a high reliability [1]. Previous work shows that shifting from 2D to 3D integration, employing TSVs, allows for a significant reduction in the circuit footprint and delay, but often increases the power consumption [2]. The system power consumption is significantly affected by TSVs as they suffer from capacitive coupling which additionally impairs the signal integrity [3]. In TSV arrays, the coupling capacitances are large due to the relatively large TSV dimensions and the conductive substrate [4]. Additionally, the high number of aggressors in 3D further increases the coupling. Thus, coupling is a critical design concern for 3D integrated circuits (3D ICs) and consequently caught the attention of academia and industry (e.g. [4–15]).
8 Conclusion
This work presents an approach to reduce the TSV power consumption by an intelligent, physical-effect-aware, local bit-to-TSV assignment, which exploits the stochastic bit properties of the transmitted data. Analyses for a large set of real and synthetic data streams underline the importance and efficiency of our low-power approach which is able to reduce the power consumption of modern TSVs by over 40 %, without inducing noticeable overhead costs.
