A fast symbolic computation approach to statistical analysis of mesh networks with multiple sources

Hao, Dan; Shi, Jianrong

doi:10.1109/aspdac.2010.5419852

Cited by 3 publications

(1 citation statement)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It can also be noted that many of above methods either require first pass DC simulation or typically have problems in handling non-tree topologies. Recursive moment based method, used in this work, along with the concepts of obtaining Thevenin equivalent of meshes allow its application to such circuits [23].…”

Section: Prior Artmentioning

confidence: 99%

Accurate Current Estimation for Interconnect Reliability Analysis

Jain

2012

IEEE Trans. VLSI Syst.

View full text Add to dashboard Cite

An improved and efficient method for static estimation of average and root-mean-squared currents used for electromigration (EM) reliability analysis is presented in this work. Significantly different from state-of-the-art, the proposed method gives closed-form expressions for average and RMS currents in one complete cycle. The proposed method can be readily configured to work with different combinations of ramp and exponential waveforms. Subsequently, the inadequacies of using conventional EM-severity metrics: either the net's lumped capacitance or the net's effective capacitance, along with the regular timing slew, for EM analysis are outlined. As a correction, and, application of proposed method, we provide formulations for deriving the effective "EM" slew, which can be used with conventional approaches to accurately compute the currents. Further, unlike traditional wisdom, we note that not just the RMS current, but even the total charge transfer can depend on the waveform type, and propose formulations to that regard. Additionally, for the first time, we present a method for incorporating the driver's dynamic IR drop while computing RMS currents. Alongside, we lay recommendations for ensuring the standard-cell EM safety at chip level. Finally, we share model-validation results from a production 40 nm design, enabling a 40% higher performance closure.

show abstract

Section: Prior Artmentioning

confidence: 99%