Abstract

This paper presents a new methodology that implements a low swing clock tree. For low power IC design, low swing clock trees are one of the known techniques to lower the overall power dissipation through decreasing the power consumption of the clock network, while trading off the clock skew, local timing (slack) and the variation-tolerance (due to decreased noise margin). In this paper, an iterative skew minimization scheme for low swing clock trees is proposed via in-place buffer sizing considering multiple process corners. The proposed approach can preserve the power savings of the low swing clock tree implementation across multiple process corners. The effect of the decreased clock swing on the local timing is analyzed: The degradation in the timing slack is shown to be insignificant due to bounded clock slew eliminating most of the timing degradation on the clock network or the logic paths induced by decreased clock swing. The experimental results show that the proposed methodology can achieve an average of up to 11% power savings, with a skew degradation of less than 5% compared to the original full-swing clock tree, satisfying a practical skew budget. The proposed scheme is highly practical as it only performs in-place buffer sizing on the original clock tree.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call