Timing-Driven Cloning Using Voronoi Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cloning techniques for integrated circuit design are inefficient in achieving optimal timing closure due to their inability to consider interconnect delays and physical layout information, leading to suboptimal gate placement and increased computational complexity, especially in designs with many sinks.
Innovation Solution
A timing-driven cloning method that iteratively partitions sinks into clusters, computes a figure of merit for optimal cloned gate placement based on linear delay models, and uses Voronoi polygons to determine valid partitions, allowing for efficient computation and improved timing optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If existing cloning techniques are used without considering interconnect delays and physical layout, then the cloning process is simpler and faster, but timing closure is suboptimal and placement is incorrect
Solution Approach 1:
The patent segments the cloning process into distinct phases: (1) identifying critical nets requiring cloning, (2) partitioning sinks into clusters using Voronoi polygons, (3) computing optimal gate locations based on linear delay models, and (4) validating timing closure. This segmentation allows complex timing optimization to be broken down into manageable steps that can be systematically executed.
Solution Approach 2:
The patent performs preliminary actions by pre-computing Voronoi polygons for sink partitioning and pre-establishing linear delay models before the actual cloning occurs. This allows the complex timing optimization to be prepared in advance, enabling faster execution during the cloning process while maintaining optimality.
2Manufacturing precision
If all possible partitions of sinks are enumerated to find optimal cloned gate location, then timing optimization is complete and optimal, but computational time becomes exponential and intractable
Solution Approach 1:
The patent segments the sink set into disjoint clusters using Voronoi polygon boundaries, which divides the complex optimization problem into smaller sub-problems. Instead of enumerating all possible partitions of all sinks, the algorithm only needs to consider partitions created by Voronoi boundaries, significantly reducing the search space from exponential to polynomial time.
Solution Approach 2:
The patent introduces Voronoi polygons as an intermediary structure that mediates between the complete set of sinks and the cloned gate location. The Voronoi polygons provide a geometric framework that automatically creates valid partitions based on proximity, eliminating the need to manually enumerate all possible partitions while ensuring optimality.
3Reliability
If cloned gates are placed without considering linear delay models and interconnect delays, then placement is faster and simpler, but timing requirements are not met
Solution Approach 1:
The patent changes the delay model parameter from simple gate delays to linear delay models that incorporate interconnect delays proportional to distance. This parameter change allows the placement algorithm to account for physical layout and interconnect characteristics, achieving timing closure while maintaining polynomial time complexity through the linear (rather than exponential) nature of the delay calculations.
Data Source
AI summary
A timing-driven cloning method iteratively partitions sinks of the net into different sets of clusters and for each set computes a figure of merit for a cloned gate location which optimizes timing based on linear delay, that is, a delay proportional to the distance between the cloned gate location and the sinks. The set having the highest figure of merit is selected as the best solution. The original gate may also be moved to a timing-optimized location. The sinks are advantageously partitioned using boundaries of Voronoi polygons defined by a diamond region surrounding the original gate, or vice versa. The figure of merit may be for example worst slack, a sum of slacks at the sinks in the second cluster, or a linear combination of worst slack and sum of the slacks.