Static Voltage Drop Prediction Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing semiconductor device design process is time-consuming and lacks early prevention for static voltage drop (SIR) issues, with traditional methods requiring lengthy iterations and being inefficient in finding optimal bump locations during the floorplan layout revising stage, leading to SIR divergency risks.
Innovation Solution
A static voltage drop prediction system utilizing machine learning techniques to predict SIR violations in floorplan layouts before placement, with automatic bump assignment optimization to reduce SIR results, implemented in an electronic design automation (EDA) system that includes a SIR prediction platform and bump assignment adjustment tool.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional SIR analysis is performed after routing using trial and error method, then SIR violations can be detected, but the design process becomes time-consuming and requires multiple iterations
Solution Approach 1:
The patent performs SIR prediction at the floorplan stage before routing is completed. By using machine learning models to predict SIR violations early in the design process based on floorplan characteristics, the system identifies potential problems before they require time-consuming routing iterations and manual adjustments.
Solution Approach 2:
The patent replaces the manual trial-and-error mechanical adjustment process with an automated machine learning-based prediction system. The ML model automatically analyzes floorplan data and predicts SIR violations, eliminating the need for repeated manual routing adjustments and iterations.
2Manufacturing precision
If manual trial and error method is used to find optimal bump locations, then bump assignment can be optimized, but the process lacks efficiency and may miss optimal solutions within limited trials
Solution Approach 1:
The patent replaces manual trial-and-error bump location optimization with an automated machine learning system that efficiently searches and identifies optimal bump assignments. The ML model processes multiple configurations rapidly and identifies optimal solutions without being constrained by limited manual trial attempts.
Solution Approach 2:
The system performs self-optimization of bump locations using the trained machine learning model. The model automatically analyzes the floorplan and determines optimal bump assignments without requiring manual intervention or iterative adjustments by designers.
3Reliability
If floorplan layout is revised after routing stage to fix SIR issues, then SIR violations can be corrected, but long fixing iterations are induced and SIR divergency risk increases
Solution Approach 1:
The patent performs SIR prediction and optimization at the floorplan stage before routing is finalized. By identifying and addressing potential SIR violations early in the design process, the system prevents the need for complex post-routing revisions and multiple fixing iterations.
Solution Approach 2:
The machine learning model provides early feedback on potential SIR violations based on floorplan characteristics. This feedback allows designers to adjust the floorplan proactively before routing, avoiding the complexity of post-routing SIR fixes and reducing the risk of SIR divergency.
Data Source
AI summary
A method is provided, including following operations: receiving, by a static voltage drop (SIR) prediction circuitry, floorplan data of a floorplan layout of a semiconductor device; generating a first SIR result by a machine learning model based on the floorplan data; generating a first similarity value based on a comparison of the floorplan data with a plurality of training data; generating a second SIR result based on the first SIR result and a first compensation value, corresponding to the first similarity value, in a mapping table; and generating a bump assignment data to update the floorplan data based on a comparison between the second SIR result with a plurality of predetermined SIR values.