Deep Learning Routing Prediction for EDA PPA Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In electronic design automation, there is a conflict between power, performance, and area (PPA) and design turnaround time (TAT) at advanced nodes, particularly with structures like Fin Field Effect Transistors, where existing tools often require trade-offs and struggle to accurately predict post-route timing and design-rule-check (DRC) during the detail routing phase.
Innovation Solution
A computer-implemented method using a deep neural network to transform and predict detail route data, including wire length and via number, based on global route data, addressing the challenge by training the network with both global and detail route data to improve prediction accuracy for RC parasitic parameters and congestion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional routing tools are used to perform detailed routing calculations, then routing completion is achieved, but design turnaround time increases significantly
Solution Approach 1:
The patent performs preliminary routing calculations using a neural network model before detailed routing is executed. The system predicts routing metrics such as wire length, via count, and congestion levels in advance, allowing the detailed routing process to proceed more efficiently with pre-computed guidance, thereby reducing overall design turnaround time while maintaining routing accuracy
Solution Approach 2:
The patent replaces traditional mechanical/computational routing calculation methods with a neural network-based predictive system. Instead of relying on conventional algorithms that require extensive computational resources and time, the system uses machine learning models trained on historical routing data to predict outcomes, substituting the mechanical calculation process with an intelligent prediction approach that is both faster and equally accurate
2Manufacturing precision
If detailed routing calculations are performed to ensure accurate PPA optimization, then routing precision is improved, but computational cost and time increase
Solution Approach 1:
The system performs preliminary predictions of power, performance, and area metrics using the neural network model before executing detailed routing. This advance prediction allows PPA optimization to be guided by pre-computed data, reducing the need for iterative detailed calculations while maintaining optimization accuracy, thus improving computational efficiency
Solution Approach 2:
The patent uses a neural network model that has been trained on historical routing data and PPA outcomes. The model creates a computational copy or approximation of the complex PPA optimization process, allowing rapid prediction of results without performing the full detailed routing calculation each time, thereby maintaining accuracy while dramatically improving calculation speed
Data Source
AI summary
The present disclosure relates to a computer-implemented method for routing in an electronic design. Embodiments may include receiving, using at least one processor, global route data associated with an electronic design as an input and generating detail route data, based upon, at least in part, the global route data. Embodiments may further include transforming one or more of the detail route data and the global route data into at least one input feature and at least one output result of a deep neural network. Embodiments may also include training the deep neural network with the global route data and the detail route data and predicting an output associated with a detail route based upon, at least in part, a trained deep neural network model.