Probabilistic Interconnect Planning for IC Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current electronic design automation tools face challenges in efficiently planning interconnect routing due to long run times and tool dependency, leading to inconsistent results across different routing tools, especially as interconnect delay becomes a dominant factor in complex integrated circuit designs.
Innovation Solution
The implementation of probabilistic interconnect planning methodologies, which analyze each net probabilistically, considering all possible routing patterns rather than specific ones, to reduce run time and provide tool-independent results by assigning probabilities to interconnect models and generating maps indicating routing characteristics such as wire congestion and power densities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If global routing is used for interconnect planning, then routing patterns can be determined, but run time becomes excessively long
Solution Approach 1:
The patent segments the interconnect planning process by dividing the routing problem into discrete nets and evaluating each net independently through probabilistic analysis. Instead of performing global routing that considers all nets simultaneously (which is computationally expensive), the method breaks down the problem into smaller, manageable units that can be processed quickly and in parallel, thereby reducing overall run time while still determining appropriate routing patterns.
Solution Approach 2:
The patent changes the fundamental parameter of the routing analysis from deterministic (finding specific routing patterns) to probabilistic (evaluating probabilities of different routing patterns). This parameter change allows the system to quickly assess routing feasibility and congestion without exhaustively searching for specific routes, significantly reducing computation time while maintaining sufficient accuracy for interconnect planning.
2Manufacturing precision
If a specific routing tool is used for global routing, then routing patterns can be generated, but results become tool-dependent and inconsistent
Solution Approach 1:
The patent creates a probabilistic model that copies and evaluates multiple possible routing patterns for each net rather than relying on a single routing tool's specific algorithm. By analyzing the probabilities of different routing patterns and considering all feasible options, the method produces results that are independent of any particular routing tool's implementation details, ensuring consistency across different tools while still generating accurate routing patterns.
Solution Approach 2:
The patent develops a universal probabilistic analysis framework that can evaluate routing patterns regardless of which specific routing tool is used. This multi-functional approach allows the same analysis method to work with different routing tools and algorithms, making the interconnect planning process adaptable and versatile while producing consistent, tool-independent results.
Data Source
AI summary
Systems and methods for interconnect planning which utilize probabilistic methodologies. One embodiment comprises a method for planning interconnect models in an integrated circuit design. Nets and a set of interconnect models that can be used to connect the pins of each net are first defined. For each net, the probability that each interconnect model will be used to connect the pins of the net is evaluated. Tiles in the integrated circuit design are then assigned probabilities indicating the likelihood that each of the interconnect models will traverse the tiles. A map is then generated to indicate probabilistic routing characteristics (e.g., probabilities of wire congestion, interconnect component congestion, power densities, interconnect model usage) based on the probabilities assigned to each of the tiles in the integrated circuit design. The map may then be output (e.g., printed or otherwise displayed) to a user or stored for later use.