Pattern Signature Generation for IC Design Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current IC design verification methods, such as DRC and lithography simulations, are inefficient and costly, especially at 65 nm and below, due to the high complexity and memory requirements of pattern matching and comparison, which necessitates a more effective way to identify and match patterns without extensive sorting and simulation.
Innovation Solution
Generating a pattern signature by transforming two-dimensional data into one-dimensional mathematical functions, compressing these into a single variable function, and calculating moment values to create a compact representation for quick pattern matching and comparison, reducing the need for vertex sorting and simulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vertex sorting and comparison approach is used for pattern matching, then pattern identification accuracy is improved, but computational expense and memory requirements increase significantly
Solution Approach 1:
The patent extracts the essential geometric characteristics of patterns by computing signatures based on polygon properties (area, perimeter, centroid, moments) rather than using all vertex data. This extraction reduces the amount of data needed for pattern matching while preserving identification accuracy.
Solution Approach 2:
The patent transforms pattern representation from detailed geometric data (vertices) to summarized parameters (signatures). By changing the representation parameters from raw vertex coordinates to derived geometric properties, the system achieves efficient pattern matching without requiring complex sorting and comparison of all vertices.
2Manufacturing precision
If lithography simulations are performed for each pattern, then manufacturing precision is improved, but time consumption and cost increase
Solution Approach 1:
The patent performs preliminary pattern analysis and signature computation before lithography simulations. By pre-computing geometric signatures and storing them in a database, the system can quickly identify similar patterns without performing time-consuming simulations for each unique pattern.
Solution Approach 2:
The patent creates a database of pattern signatures as a simplified representation model. Instead of performing full lithography simulations for every pattern, the system uses these signature copies to quickly match and retrieve simulation results for similar patterns, reducing the need for repeated simulations.
3Adaptability or versatility
If all vertices are stored for large patterns, then pattern comparison capability is improved, but memory requirements increase significantly
Solution Approach 1:
The patent extracts only the essential geometric information needed for pattern comparison by computing signatures from polygon properties. This extraction eliminates the need to store all vertex data while maintaining the ability to compare and identify patterns effectively.
Solution Approach 2:
The patent segments the pattern comparison task into two parts: computing geometric signatures from polygon data and using these signatures for matching. This segmentation allows the system to work with compact signature data rather than full vertex data, significantly reducing memory requirements.
Data Source
AI summary
A method, system, and computer program product are disclosed for generating a pattern signature to represent a pattern in an integrated circuit design. In one approach, the method, system and computer program product transform pattern data, two dimensional data for the pattern, into a set of one dimensional mathematical functions, compress the set of one dimensional mathematical functions into a single variable function, compress the single variable function by calculating a set of values for the single variable function, and generate a pattern signature for the pattern from the set of values.