An efficient current-based logic cell model for crosstalk delay analysis

Authors:
Debasish Das;William Scott;Shahin Nazarian;Hai Zhou
Affiliations:
EECS, Northwestern University, Evanston, IL 60208, USA;Magma Design Automation, San Jose, CA 95110, USA;Magma Design Automation, San Jose, CA 95110, USA;EECS, Northwestern University, Evanston, IL 60208, USA
Venue:
ISQED '09 Proceedings of the 2009 10th International Symposium on Quality of Electronic Design
Year:
2009

Citing 0
Cited 3

Pessimism reduction in coupling-aware static timing analysis using timing and logic filtering

IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
FA-STAC: An algorithmic framework for fast and accurate coupling aware static timing analysis

IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Computational performance optimisation for statistical analysis of the effect of nano-CMOS variability on integrated circuits

VLSI Design

Quantified Score

Hi-index	0.00

Visualization

Abstract

Logic Cell modeling is an important component in the analysis and design of CMOS integrated circuits, mostly due to nonlinear behavior of CMOS cells with respect to the voltage signal at their input and output pins. A current-based model for CMOS logic cells is presented which can be used for effective crosstalk noise and delta delay analysis in CMOS VLSI circuits. Existing current source models are expensive and need a new set of Spice-based characterization which is not compatible with typical EDA tools. In this paper we present Imodel, a simple nonlinear logic cell model that can be derived from the typical cell libraries such as NLDM, with accuracy much higher than NLDM-based cell delay models. In fact, our experiments show an average error of 3% compared to Spice. This level of accuracy comes with a maximum runtime penalty of 19% compared to NLDM-based cell delay models on medium sized industrial designs.