Gate-Level Simulation X-Pessimism Reduction via Reconvergent Path Correlation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional gate-level simulators and verification techniques exhibit X-pessimism due to their inability to account for correlations between reconvergent paths, leading to inaccurate propagation of indeterminate values and increased uncertainty in signal values, especially during power-up modeling of complex low-power circuits.

Innovation Solution

The system identifies combinational blocks expected to exhibit X-pessimism and modifies the gate-level design by adding correcting blocks or replicating combinational blocks, assigning concrete values to indeterminate inputs, and using unique free input variables to represent sources of indeterminate values, thereby reducing X-pessimism during simulation and verification.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional gate-level simulators propagate indeterminate values through reconvergent paths, then simulation speed is maintained, but measurement precision of signal values deteriorates due to X-pessimism

Engineering Contradiction:
Improveaccuracy of indeterminate value propagationVSAvoidcomplexity of simulation system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces a correlation tracking mechanism as an intermediary between the conventional simulator and the indeterminate value propagation. This mechanism tracks correlations between reconvergent paths and uses this information to adjust propagation behavior, thereby improving measurement precision without fundamentally changing the simulator architecture

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent dynamically changes the propagation parameter (whether to propagate X values) based on detected correlations in reconvergent paths. When correlations are detected, the system switches from standard propagation to a modified propagation mode that accounts for the correlations, thus improving accuracy adaptively

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If gate-level simulation accounts for correlations between reconvergent paths, then measurement precision improves, but device complexity increases due to additional tracking mechanisms

Engineering Contradiction:
Improveaccuracy of signal value propagationVSAvoidcomplexity of correlation tracking
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the correlation tracking into modular components: correlation detection modules that identify reconvergent paths, correlation storage structures that record relationships, and propagation adjustment modules that apply the tracking information. This segmentation reduces overall system complexity by making each component independent and manageable

Inventive Principle:
Principle #1Segmentation

3Reliability

If conventional verification techniques are used, then verification speed is maintained, but reliability of verification results deteriorates due to X-pessimism

Engineering Contradiction:
Improveaccuracy of verification resultsVSAvoidverification speed
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent performs preliminary correlation analysis before the main verification process. By pre-identifying and storing correlation information in reconvergent paths, the system prepares the data structures needed for accurate verification in advance, allowing the main verification to proceed at normal speed while benefiting from the pre-computed correlation information

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS8650513B2Reducing x-pessimism in gate-level simulation and verification
Publication Date: 2014.02.11 SYNOPSYS INC
  • US8650513B2 patent drawing
  • US8650513B2 patent drawing
  • US8650513B2 patent drawing

AI summary

Methods and apparatuses are described for reducing or eliminating X-pessimism in gate-level simulation and/or formal verification. A system can identify a set of reconvergent inputs of a combinational block in a gate-level design. Next, the system can determine whether or not the combinational block is expected to exhibit X-pessimism during gate-level simulation. If the combinational block is expected to exhibit X-pessimism during gate-level simulation, the system can modify the gate-level design to reduce X-pessimism during gate-level simulation. In some embodiments, the system can build a model for the gate-level design by using unique free input variables to represent sources of indeterminate values. The system can then use the model to perform formal verification.