Statistical Leakage Power Analysis for IC Yield
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Shrinking process node sizes in integrated circuit manufacturing lead to significant process variations, causing increased leakage power issues and reduced manufacturing yield due to excessive leakage currents, which existing approaches fail to accurately manage effectively.
Innovation Solution
The implementation of statistical leakage power analysis and characterization with consideration of state correlation, using a system and method that generates accurate leakage library models to encapsulate process variations and their effects on leakage power, including global, local, random, and environmental variations, and employing sensitivity values to compute the impact of these variations on overall design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of moving object
If process node sizes are shrunk to improve integration density, then manufacturing precision deteriorates due to significant process variations causing excessive leakage currents
Solution Approach 1:
The patent transforms fixed worst-case corner values into variable statistical parameters (mean and standard deviation) that can be adjusted based on actual process conditions. This allows the design to adapt to different process nodes and variations, improving leakage power estimation accuracy while maintaining compatibility with scaled technologies.
Solution Approach 2:
The patent implements a feedback mechanism where statistical leakage power analysis results are used to refine yield estimates and guide design optimizations. The system continuously iterates between leakage analysis, variation characterization, and design modification to achieve target yield levels despite process variations.
2Ease of operation
If static maximum and minimum worst-case corner values are used to manage leakage power, then ease of operation improves, but measurement precision deteriorates due to overly pessimistic and inaccurate leakage power estimation
Solution Approach 1:
The patent replaces the traditional mechanical approach of using fixed corner values with a statistical computing model that processes variation data through mathematical analysis. This substitution enables more accurate leakage estimation by accounting for the probabilistic nature of process variations rather than relying on conservative worst-case assumptions.
Solution Approach 2:
The patent changes the fundamental parameters from static corner values to dynamic statistical distributions. By representing leakage power as a statistical variable with mean and standard deviation rather than fixed max/min values, the system achieves both operational simplicity through standardized analysis procedures and improved precision through accurate variation modeling.
3Reliability
If range between max and min worst-case corner values is increased to account for process variations, then reliability improves, but manufacturing precision deteriorates due to overly large ranges relative to designed feature sizes
Solution Approach 1:
The patent transforms the wide static range between max and min corner values into a more precise statistical representation using mean and standard deviation parameters. This transformation maintains reliability by capturing the full distribution of possible outcomes while improving manufacturing precision by quantifying variations in terms that can be directly compared to feature sizes.
Solution Approach 2:
The patent adds a statistical dimension to leakage power analysis by introducing probability distributions and standard deviations. This additional dimension allows the system to characterize process variations more accurately without simply expanding the range, instead providing a nuanced understanding of variation magnitude and likelihood that improves both reliability assessment and manufacturing precision.
Data Source
AI summary
A method, system, and computer program product are disclosed for performing statistical leakage power characterization to estimate yield of a circuit in terms of leakage power. According to some approaches, this is performed with consideration of state correlation.