Static Timing Analysis Signal Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current static timing analysis (STA) methods, particularly graph-based static timing analysis (GBA) and path-based static timing analysis (PBA), face challenges in accuracy and efficiency due to overestimation of delays and exponential complexity, leading to pessimistic results and high runtime/memory overhead in large integrated circuit designs.
Innovation Solution
The method involves propagating multiple timing signals in integrated circuit designs, merging subsets of signals at shared nodes based on similarity and impact on arrival times, and generating representative merged signals to reduce accuracy loss and runtime, using a proximity-based clustering algorithm to manage memory and runtime overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If graph-based static timing analysis (GBA) is used to determine slack values with linear complexity, then runtime efficiency is improved, but timing accuracy deteriorates due to overestimation of delays and pessimistic slack
Solution Approach 1:
The patent segments the set of timing signals into multiple subsets based on their convergence characteristics at shared nodes. Instead of treating all timing signals uniformly as in GBA, the method divides them into groups where signals within each subset are merged and signals from different subsets are discriminated. This segmentation allows selective merging that maintains timing accuracy while managing complexity, resolving the contradiction between runtime efficiency and timing accuracy.
2Measurement precision
If path-based static timing analysis (PBA) is used to analyze each timing path in isolation, then timing accuracy is improved, but runtime and memory overhead increase exponentially with circuit size
Solution Approach 1:
The patent merges timing signals from the same subset at shared nodes to reduce the number of signals that need to be propagated through the circuit. By combining signals that have similar characteristics and convergence behavior, the method maintains the accuracy benefits of PBA while achieving the efficiency of GBA. The merging operation reduces exponential complexity to manageable levels while preserving timing precision for discriminated signal subsets.
3Productivity
If all timing signals converging at a shared node are merged together, then runtime overhead is reduced, but timing accuracy is lost due to over-generalization
Solution Approach 1:
The patent applies local quality by treating different timing signals differently based on their specific characteristics and convergence behavior. Instead of uniform merging or non-merging, the method selectively merges signals within subsets while discriminating between subsets. This local differentiation ensures that signals with similar properties are combined for efficiency, while signals with different properties are kept separate to maintain accuracy in arrival time computation.
Data Source
AI summary
A method for performing static timing analysis of an integrated circuit design, wherein at least two timing paths share a shared node comprises propagating along the at least two timing paths a plurality of timing signals characterized by a set of timing parameters and determining respective values of the timing parameters at the shared node. Subsets of timing signals are defined based on relations between the determined parameter values of different timing signals. For each of the subsets representative parameter values are identified and a merged timing signal is propagated from the shared node at least partially along the at least two timing paths. Therein the merged timing signal has at the shared node the representative parameter values of the subset. The method also comprises generating timing data based on the merged timing signals and storing the timing data.