Learning-Based Analog Circuit Placement With Well Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing analog placement methods struggle to efficiently minimize total chip area and interconnecting wire-length while adhering to critical constraints in analog layout synthesis, impacting circuit performance.
Innovation Solution
A learning-based auto placement method using a reinforcement learning model with an actor-critic framework to determine optimal placement plans for analog circuits, considering bounded-sliceline grid units and constraints like well-island, proximity, and symmetry-island, through a training process involving an actor model and critic model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If conventional analog placement methods are used, then placement constraints can be satisfied, but chip area and wire-length cannot be efficiently minimized
Solution Approach 1:
The patent replaces conventional mechanical/optical measurement systems with an AI-based simulation system. The AI model predicts analog layout quality metrics (chip area, wire-length) by simulating electrical signal propagation and interference effects, eliminating the need for actual physical measurements while achieving accurate optimization guidance.
Solution Approach 2:
The patent transforms the placement optimization problem by changing parameters from direct physical measurements to AI-predicted quality metrics. The system uses learned relationships between device geometry, material properties, and electrical performance to optimize placement based on predicted outcomes rather than trial-and-error physical testing.
2Measurement precision
If physical measurement methods are used to evaluate analog layout quality, then accurate circuit performance assessment is possible, but the process is time-consuming and inefficient
Solution Approach 1:
The patent creates a virtual copy of the physical measurement and evaluation process through AI simulation. The system replicates electrical signal behavior, interference patterns, and performance characteristics in a computational environment, providing accurate layout quality assessment without requiring actual physical testing of each placement configuration.
Solution Approach 2:
The patent substitutes physical measurement apparatus and procedures with an AI-based computational system. The AI model learns from training data to predict circuit performance metrics directly from placement configurations, replacing time-consuming physical measurements with instant computational predictions.
3Length of stationary object
If traditional placement optimization is performed, then some constraints are met, but interconnecting wire-length cannot be effectively minimized
Solution Approach 1:
The patent applies local quality optimization by considering specific local constraints (such as well-island, proximity, and symmetry-island constraints) individually while optimizing overall wire-length. The AI model learns to balance local placement requirements with global wire-length minimization, adjusting device positions to satisfy local constraints while reducing total interconnection length.
Solution Approach 2:
The patent transforms the complex constrained optimization problem by changing the approach from direct constraint satisfaction to AI-guided optimization. The system learns from training data the optimal balance between satisfying multiple constraints and minimizing wire-length, using predicted performance metrics to guide placement decisions that simultaneously consider all factors.
Data Source
AI summary
A method for learning-based auto placement of analog circuit is proposed. The method includes acquiring a netlist file of an analog circuit including plural devices, obtaining a total number of plural wells of the analog circuit and the number of devices sharing a first well which has the greatest number of devices sharing the same well among the wells according to the netlist file, determining the number of plural bounded-sliceline grid (BSG) units according to the total number of the wells, determining a size of each BSG unit according to the number of devices sharing the first well, and performing a training process by using a reinforcement learning model comprising an actor model and a critic model, to determine an optimum placement plan.