Statistical Static Timing Analysis Proxy Slack Debugging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional block-based statistical static timing analysis (SSTA) faces challenges in accurately identifying and debugging timing constraint violations due to misleading slack distributions and difficulties in tracing failing tests, leading to confusion and inefficiencies in designer and optimization tool operations.
Innovation Solution
The method involves building a timing graph, forward and backward propagating probabilistic distributions to calculate arrival and required arrival times, and identifying a statistically worst slack, which can be replaced by a proxy worst slack for edges, allowing for improved reporting and debugging by focusing on specific timing issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If block-based statistical static timing analysis is used to propagate timing quantities as statistical distributions, then the computational efficiency is improved and the coverage of process variations is enhanced, but the ability to accurately identify and debug timing constraint violations deteriorates due to misleading slack distributions
Solution Approach 1:
The patent segments the statistical slack distribution into multiple discrete slack values, each associated with a specific probability. This segmentation transforms the continuous statistical distribution into discrete, traceable units that can be individually tracked through the timing analysis process, enabling accurate identification of failing tests while maintaining the computational efficiency of statistical methods.
Solution Approach 2:
The patent introduces an intermediary mechanism that tracks the origin and propagation of each discrete slack value through the timing graph. This intermediary tracking system bridges the gap between the statistical distribution approach and the need for precise debugging, allowing designers to trace failing slacks back to their source without sacrificing computational efficiency.
2Adaptability or versatility
If statistical distributions are propagated through the timing graph, then the entire space of process variations is covered in a single run, but the tracing of failing tests becomes difficult due to the aggregated nature of statistical results
Solution Approach 1:
The patent segments the aggregated statistical distribution into discrete slack values, each carrying information about its origin and propagation path. This segmentation preserves test tracing information that would otherwise be lost in the aggregated statistical results, while still maintaining the ability to cover the entire process variation space.
Solution Approach 2:
The patent applies a metaphorical 'coloring' or tagging mechanism to discrete slack values, where each slack value carries metadata identifying its source and propagation history. This tagging system enables easy tracing of failing tests through the timing graph while maintaining the comprehensive process variation coverage provided by statistical distributions.
3Measurement precision
If multiple deterministic STA timing runs are executed to cover different corners, then the extreme performance bounds are accurately captured, but the computational overhead increases significantly
Solution Approach 1:
The patent merges the advantages of deterministic corner analysis with statistical distribution propagation by executing a single statistical timing run that simultaneously covers all process variations. This combining approach captures extreme performance bounds accurately while avoiding the computational overhead of multiple deterministic runs.
Solution Approach 2:
The patent changes the parameter representation from discrete corner values to continuous statistical distributions. This parameter transformation enables a single timing run to cover the entire process variation space, including extreme performance bounds, without requiring multiple separate deterministic analyses.
Data Source
AI summary
Methods for analyzing timing of an integrated circuit using block-based static statistical timing analysis and for practical worst test definition and debug. The method includes building a timing graph, determining a slack for each of the nodes in the timing graph, and identifying a statistically worst slack for at least one of the nodes. The method further includes replacing this statistically worst slack with a proxy worst slack.