Tie Net Data Preprocessing for Router Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing router programs face difficulties in efficiently routing tie net data due to its massively parallel nature, leading to sub-optimal solutions and prolonged processing times in large-scale integrated circuit designs.
Innovation Solution
A preprocessing method organizes tie net data into simpler, optimized connections between pins and the power grid, transforming the routing problem into a more manageable form for the router, allowing for efficient routing and subsequent restoration of the original design state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If unedited tie net data is used by the router, then the original design state is preserved, but the router experiences difficulty in routing due to the massively parallel nature of the problem, resulting in sub-optimal solutions and prolonged processing times
Solution Approach 1:
The preprocessing program segments the massively parallel tie net routing problem by organizing unused elements into groups based on their connectivity patterns and power grid connections. This segmentation transforms the overwhelming parallel routing task into manageable subsets that the router can process efficiently, directly improving router processing speed without sacrificing design integrity
Solution Approach 2:
The preprocessing program performs preliminary organization of tie net data before routing, creating an optimized data structure that groups unused elements by their power grid connections and routing characteristics. This preliminary action prepares the data in advance to eliminate the massively parallel complexity during actual routing, enabling faster processing while maintaining optimal routing solutions
2Manufacturing precision
If unedited tie net data is routed directly, then no additional processing steps are required, but the router produces sub-optimal routing solutions due to the complexity of the massively parallel problem
Solution Approach 1:
By segmenting tie net data into organized groups based on power grid connections and routing patterns, the preprocessing program enables the router to find optimal paths for each group separately. This segmentation improves routing solution quality by allowing the router to focus on smaller, manageable subsets rather than being overwhelmed by the full massively parallel problem
Solution Approach 2:
The preprocessing program acts as an intermediary between the original design data and the router, transforming unedited tie net data into an optimized format with hierarchical organization. This intermediary processing step improves routing solution quality by presenting the router with pre-organized data structures that highlight optimal routing opportunities while maintaining the original design intent
Data Source
AI summary
A method for preprocessing tie net routing data organizes the data into a plurality of tie nets each based on an optimal connection path between a pin or set of pins and the power grid. The router then routs the data embodying the thusly-simplified plurality of tie nets. Once the routing is complete, post processor takes the routed design and returns it to its original net list state while keeping the routing solution.