Sigma Margin Timing Optimization for IC Yield
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing approaches to determining timing criticality in electronic design optimization using slack metrics are inaccurate, as they do not account for the probability of slack variations, leading to inefficient prioritization and resource allocation in fixing timing violations.
Innovation Solution
The use of sigma margin (SM) values, calculated as the mean of slack divided by its standard deviation, to assess the criticality of timing paths and prioritize path fixing based on the probability of negative slack, allowing for more accurate identification and correction of critical paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If slack-based prioritization is used to identify critical paths, then the process is simple to implement, but the accuracy of timing criticality assessment deteriorates due to not accounting for slack variations
Solution Approach 1:
The patent transforms the timing criticality assessment by changing from using raw slack values to using sigma margin (SM) values. The SM value is calculated as the mean slack divided by the standard deviation of slack, which changes the parameter representation to account for variability. This transformation maintains computational simplicity while significantly improving the accuracy of critical path identification by incorporating probabilistic information about slack variations.
2Ease of operation
If paths with larger negative slack are prioritized for fixing, then the approach is intuitive and easy to understand, but resource allocation becomes inefficient due to not considering probability of slack violations
Solution Approach 1:
The patent replaces the intuitive but inefficient mechanical approach of prioritizing by raw slack magnitude with a statistical approach using sigma margin values. The SM value incorporates both the mean slack and its standard deviation, providing a normalized measure that accounts for the probability of violations. This substitution maintains ease of understanding while dramatically improving resource allocation efficiency by prioritizing paths based on their actual risk of timing failures rather than just the magnitude of their slack violation.
3Reliability
If all timing paths are fixed to meet timing requirements, then timing yield is maximized, but design complexity and cost increase due to over-engineering non-critical paths
Solution Approach 1:
The patent segments timing paths into different priority groups based on their sigma margin values. By calculating SM for each path and sorting them, the system identifies a critical set of paths that are most likely to cause timing failures. This segmentation allows the optimization process to focus resources on fixing only the most critical paths first, rather than treating all paths equally. The segmentation approach maintains high timing yield for critical paths while reducing unnecessary complexity in fixing non-critical paths.
Data Source
AI summary
A method for timing optimization is disclosed. The method includes obtaining information on timing of paths in a chip, the information including a mean of slacks and a sigma of slacks of each of the paths; determining a sigma margin (SM) value each of the paths, the SM value being obtained by dividing the mean by the sigma; dividing the paths into groups based on SM values, an SM value of a path in one of the groups being different from that of a path in another one of the groups; and determining a yield requirement that indicates the maximum number of paths allowable in each group in order to achieve a predetermined yield.