IC Layout Manufacturability Modeling for Gradient Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current integrated circuit design methods rely on binary Boolean design rules, which are insufficient for optimizing manufacturability and yield due to their inability to handle continuous and differentiable manufacturability models.
Innovation Solution
The development of machine learning-based systems that generate differentiable manufacturability parameters, combining physics-based models and machine learning models to provide continuous and differentiable representations of manufacturability, enabling holistic optimizations of integrated circuit layouts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If binary Boolean design rules are used for manufacturability assessment, then design rule checking can be automated, but gradient based optimization methods become impractical
Solution Approach 1:
The patent transforms the binary Boolean manufacturability output into a continuous differentiable parameter space. By representing manufacturability as a continuous value rather than a discrete Boolean, the system enables gradient-based optimization methods to be applied, resolving the contradiction between automation and optimization capability.
Solution Approach 2:
The patent introduces a differentiable manufacturability model as an intermediary between the binary design rule checking system and the optimization algorithms. This intermediary layer translates discrete design rule violations into continuous gradients, allowing optimization methods to operate effectively while maintaining compatibility with existing automated design rule checking infrastructure.
2Measurement precision
If detailed manufacturability data is retained by foundries, then accurate manufacturability assessment is possible, but design optimization cannot proceed without continuous differentiable models
Solution Approach 1:
The patent changes the parameter representation from discrete Boolean outcomes to continuous differentiable values. This transformation maintains the precision of manufacturability assessment by preserving the underlying physical mechanisms while making the data suitable for gradient-based optimization, thereby resolving the contradiction between measurement precision and optimization feasibility.
3Device complexity
If Boolean design rules with binary outcomes are used, then manufacturability constraints can be encoded as logic, but physical mechanisms underlying manufacturability cannot be represented continuously
Solution Approach 1:
The patent introduces a differentiable manufacturability model as an intermediary that bridges the simple Boolean encoding and the complex physical mechanisms. This intermediary preserves the encoding simplicity while recovering the lost physical mechanism information by representing manufacturability as a continuous differentiable function that captures the underlying physics of the manufacturing process.
Data Source
AI summary
Systems, computer-implemented methods, and instructions encoded in machine-accessible storage media are provided for determining manufacturability of an integrated circuit layout. A computer-implemented method includes receiving a layout describing the integrated circuit to be manufactured by a semiconductor manufacturing process. The method also includes generating a differentiable manufacturability parameter as an output of a machine learning model using the layout, the machine learning model being trained to generate the differentiable manufacturability parameter. The differentiable manufacturability parameter describes the manufacturability of the integrated circuit by the semiconductor manufacturing process.