- Patent Number:
7,243,320
- Appl. No:
11/301999
- Application Filed:
December 12, 2005
- Abstract:
An Integrated Circuit Design tool incorporating a Stochastic Analysis Process (“SAP”) is described. The SAP can be applied on many levels of circuit components including transistor devices, logic gate devices, and System-on-Chip or chip designs. The SAP replaces a large number of traditional Monte Carlo simulations with operations using a small number of sampling points or corners. The SAP is a hierarchical approach using a model fitting process to generate a model that can be used with any number of performance memos to generate performance variation predictions along with corresponding statistical information (e.g., mean, three-sigma probability, etc.). The SAP provides an efficient way of modeling the circuit or system variation due to global parameters such as device dimensions, interconnect wiring variations, economic variations, and manufacturing variations.
- Inventors:
Chiu, Hsien-Yen (San Jose, CA, US); Wang, Meiling (Tucson, AZ, US); Li, Jun (San Jose, CA, US)
- Assignees:
Anova Solutions, Inc. (Santa Clara, CA, US)
- Claim:
1. A method comprising: receiving an input distribution of a plurality of parameters of a pre-defined system and mutual correlations between the parameters; normalizing parameter distributions and decomposing the mutual correlations to generate standardized independent parameter sets; generating a specific set of input values for sampling of the pre-defined system based on the standardized independent parameter sets; performing pre-defmed system sampling on the specific set of input values to generate pre-defined system output values; performing orthogonal polynomial fitting on the specific set of input values, the pre-defined system output values, and the standardized independent parameter sets to generate a stochastic analysis process (SAP) model; and using the SAP model to generate an output distribution of the pre-defined system.
- Claim:
2. The method of claim 1 wherein the output distribution is a statistical representation of the output distribution of the pre-defined system based on variations of input parameters, wherein the statistical representation comprises one of graphical PDF, CDF, mean and sigma value sets.
- Claim:
3. The method of claim 2 wherein the specific set of input values for sampling is determined through steps of: defining a number of fitting order of orthogonal polynomials; obtaining a number of the parameters; selecting higher probability roots of the orthogonal polynomials; acid converting the higher probability roots to the specific set of input values for sampling of the pine-defined system.
- Claim:
4. The method of claim 3 wherein the pre-defined system is one of a transistor device, a logic gate device, and system-on-chip circuit.
- Claim:
5. The method of claim 4 wherein the SAP model simulates performance of a hierarchically more abstract version of the pre-defined system that includes one or more blocks of the pre-defined system.
- Claim:
6. The method of claim 5 wherein the pre-defined system is a logic gate, and wherein the hierarchically more abstract version comprises a system-on-chip device.
- Claim:
7. The method of claim 1 further comprising: defining the SAP model as containing multiple independent standardized variables; rebuilding one or more additional SAP models from a dominant vector of a linear combination of the parameters; and optimizing vector selection by minimizing a difference between the SAP model and the additional SAP models, the additional SAP models representing approximate behavior characteristics of the SAP model using fewer independent variables than the SAP model.
- Claim:
8. A method comprising: defining a hierarchical relationship between a lowest level system and a highest level system; pre-generating an analytic model comprising one or more stochastic analysis process (SAP) models to predict system behavior for the lowest level system; and recursively using the one or more SAP models for the lowest level system to build the highest level system SAP model.
- Claim:
9. The method of claim 8 wherein the at least one of the one or more SAP models is generated by: receiving an input distribution of a plurality of parameters of the lowest level system and mutual correlations between the parameters; normalizing parameter distributions and decomposing the mutual correlations to generate standardized independent parameter sets; generating a specific set of input values for sampling of the lowest level system based on the standardized independent parameter sets; performing lowest level system sampling on the specific set of input values to generate lowest level system output values; and performing orthogonal polynomial fitting on the specific set of input values, the lowest level system output values, and the standardized independent parameter sets to generate the SAP model.
- Claim:
10. The method of claim 9 further comprising using the SAP model to generate an output distribution of the lowest level system.
- Claim:
11. The method of claim 10 wherein the output distribution is a statistical representation of the output distribution of the lowest level system based on variations of input parameters.
- Claim:
12. The method of claim 11 wherein the specific set of input values for sampling is determined through steps of defining a number of fitting order of orthogonal polynomials; obtaining a number of the parameters; selecting higher probability roots of the orthogonal polynomials; and converting the higher probability roots to the specific set of input values for sampling of the lowest level system.
- Claim:
13. The method of claim 12 wherein the lowest level system is one of a transistor device, a logic gate device, and system-on-chip circuit.
- Claim:
14. The method of claim 13 wherein the lowest level system is a transistor device, and wherein the hierarchically more abstract version comprises a system-on-chip device.
- Claim:
15. The method of claim 9 further comprising: defining a lowest level SAP model as containing multiple independent standardized variables; rebuilding one or more additional SAP models from a dominant vector of a linear combination of the parameters; and optimizing vector selection by minimizing a difference between the SAP model and the additional SAP models, the additional SAP models representing approximate behavior characteristics of the SAP model using fewer independent variables than the SAP model.
- Claim:
16. A system comprising: means for receiving an input distribution of a plurality of parameters of a pre-defined circuit and mutual correlations between the parameters; means for normalizing the parameter distributions and decomposing the mutual correlations to generate standardized independent parameter sets; means for generating a specific set of input values for sampling of the pre-defined circuit based on the standardized independent parameter sets; means for performing pre-defined circuit sampling on the specific set of input values to generate pre-defined circuit output values; means for performing orthogonal polynomial fitting on the specific set of input values, the pre-defined circuit output values, and the standardized independent parameter sets to generate a stochastic analysis process (SAP) model; and means for using the SAP model to generate an output distribution of the pre-defined circuit.
- Claim:
17. The system of claim 16 wherein the output distribution is a statistical representation at the output distribution of the pre-defined circuit based on variations of input parameters.
- Claim:
18. The system 17 further comprising: means for defining a number of fitting order of orthogonal polynomials; means for obtaining a number of the parameters; means for selecting higher probability roots of the orthogonal polynomials; and means for converting the higher probability roots to the specific set of input values for sampling of the pre-defined circuit.
- Claim:
19. The system of claim 18 wherein the pre-defined circuit is one of a transistor device, a logic gate device, and system-on-chip circuit, and wherein the SAP model simulates performance of a hierarchically more abstract version of the pre-defined circuit that includes one or more blocks of the pre-defined circuit.
- Claim:
20. The system of claim 19 further comprising: means for defining the SAP model as containing multiple independent standardized variables; means for rebuilding one or more additional SAP models from a dominant vector of a linear combination of the parameters; and means for optimizing vector selection by minimizing a difference between the SAP model and the additional SAP models, the additional SAP models representing approximate behavior characteristics of the SAP model using fewer independent variables than the SAP model.
- Current U.S. Class:
716/4
- Patent References Cited:
2002/0073394 June 2002 Milor et al.
2005/0235232 October 2005 Papanikolaou et al.
- Other References:
Hatono et al.,“Modeling and Online Scheduling of Flexible Manufacturing Systems Using Stochastic Petri Nets” ,Feb. 1991, IEEE Transactions on Software Engineering, vol. 17, iss. 2, pp. 126-132. cited by examiner
Sethi et al.,“Hierarchical Production and Setup Scheduling in Stochastic Manufacturing Systems”, Dec. 1994, Proceedings of the 33rd IEEE Conference on Decision and Control, vol. 2, pp. 1571-1576. cited by examiner
Shi et al.,“Design and Optimization of Complex Real-Time Dependable Systems”, Feb. 1996, Proceedings of WORDS' 96., Second Workshop on Object-Oriented Real-Time Dependable Systems, pp. 218-224. cited by examiner
Soner et al.,“An Asymptotic Analysis of Hierarchical Control of Manufacturing Systems”, Dec. 1988, Proceddings of the 27th IEEE conference on Decision and Control, vol. 3, pp. 1856-1857. cited by examiner
- Primary Examiner:
Lin, Sun James
- Attorney, Agent or Firm:
Courtney Staniford & Gregory LLP
- Accession Number:
edspgr.07243320
Processing Request
No Comments.