| A. Dharchoudhury, S.M. Kang, "Worst-Case Analysis and Optimization of VLSI Circuit Performances", IEEE Trans. on Computer-Aided Design, vol. 14, no. 4, pp 481-192, 1993 |
....to the growing influence of process fluctuation and changes in the operating conditions, design centering is necessary in addition to nominal design in order to ensure a high production yield. Many approaches to design centering, based on statistical, e.g. 2, 11] and deterministic methods, e.g. [1, 6, 12], were presented. Usually design centering algorithms are computationally very expensive. Hence the design centering process should be started from a good nominal design in order to keep the computational cost small. This can be achieved by introducing worst case parameter sets for process and ....
....algorithms are computationally very expensive. Hence the design centering process should be started from a good nominal design in order to keep the computational cost small. This can be achieved by introducing worst case parameter sets for process and operating conditions into nominal design [4, 6]. In the digital domain, process fluctuations are being considered by means of slow and fast worst case parameter sets. These are calculated for a given process and typical circuit performance like delay, but independent from a specific circuit. For digital cell libraries, these parameter sets ....
[Article contains additional citation context not shown here]
A. Dharchoudhury and S. M. Kang. Worst-case analysis and optimization of VLSI circuit performances. IEEE TCAD, 14(4), 1995.
....problems by multiple criteria optimization (MCO) on a set of robustness objectives instead of optimizing the yield directly [10 12] But circuit performances are often considerably correlated, which is difficult to account for in MCO. Algorithms relying on precalculated worst case parameter sets [9] face the problem that the variances of the statistical parameters depend on the design parameters when local process variations and mismatch are to be considered. In this case, the worst case parameter set is known to strongly depend on the design parameters and will hence change during the ....
A. Dharchoudhury, S. M. Kang, "Worst-case analysis and optimization of VLSI circuit performances", IEEE TCAD, 1995.
.... approximation [5] or the ellipsoidal technique [1] More recently, deterministic methods for design centering have been presented that use multiple robustness objectives (MROs) for individual performances, e.g. linearized performance penalties [9] worst case distances [2] performance scores [4]. These approaches have the advantage of a multiple criteria problem formulation analoguous to the nominal design task [10] which enables the application of the same powerful algorithms. Unfortunately, the scalarized formulation of the optimization problem, e.g. with a sum of exponential ....
.... Delta fi i (d) ff 0 (6) The positive factor ff in this equation is a scaling factor for the worst case distances. The method presented in this report is not limited to the optimization of the worst case distances. Generally, any robustness objective (e.g. LPP [9] or performance score [4]) and any convex objective function that transforms these multiple robustness objectives into one scalar value may be used. Additionally, constraints must be respected during the sizing process, in order to guarantee properly functioning circuits. These so called structural constraints [14, 18] ....
A. Dharchoudhury and S. M. Kang. Worst-Case Analysis and Optimization of VLSI Circuit Performances. In IEEE Trans. Circuits and Systems (CAS), volume 14, pages 481--492, Apr. 1995.
No context found.
A. Dharchoudhury, S.M. Kang, "Worst-Case Analysis and Optimization of VLSI Circuit Performances", IEEE Trans. on Computer-Aided Design, vol. 14, no. 4, pp 481-192, 1993
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC