US 7,555,740 B2
Method and system for evaluating statistical sensitivity credit in path-based hybrid multi-corner static timing analysis
Nathan C. Buck, Underhill, Vt. (US); John P. Dubuque, Jericho, Vt. (US); Eric A. Foreman, Fairfax, Vt. (US); Peter A. Habitz, Hinesburg, Vt. (US); Kerim Kalafala, Rhinebeck, N.Y. (US); Peihua Qi, Wappingers Falls, N.Y. (US); Chandramouli Visweswariah, Croton-on-Hudson, N.Y. (US); and Xiaoyue Wang, Kanata (Canada)
Assigned to International Business Machines Corporation, Armonk, N.Y. (US)
Filed on Feb. 27, 2007, as Appl. No. 11/679,251.
Prior Publication US 2008/0209373 A1, Aug. 28, 2008
Int. Cl. G06F 17/50 (2006.01)
U.S. Cl. 716—12  [716/4; 716/6; 716/18] 22 Claims
OG exemplary drawing
 
1. A method for analyzing a timing design of an integrated circuit and operable by a computing processor, the method comprising:
providing an initial static timing analysis of the integrated circuit;
selecting a static timing test with respect to a static timing test point based on the initial static timing analysis;
selecting a timing path leading to the static timing test point for the static timing test;
determining an integrated slack path variability by a computer processor for the timing path based on a joint probability distribution of at least one statistically independent parameter
analyzing the timing design based on the integrated slack path variability; and determining a sensitivity of the timing path to each of the at least one statistically independent parameter, the sensitivity being determined based on a probability distribution of the statistically independent parameter wherein the sensitivity to the statistically independent parameter is determined approximately using formula:
S=(Pb−Pw)/(σb−σw),
where S indicates the sensitivity, Pb indicates a slack under a best corner of the statistically independent parameter, and Pw indicates a slack under a worst corner of the statistically independent parameter, and σb and σw indicate numbers of sigma points of distribution limits at the best and worst corners of the statistically independent parameter, respectively.