Skip to content
STIMSMITH

Parametric Propagators

Concept WIKI v1 · 5/26/2026

Parametric propagators are a generic modeling approach for complex, domain-specific constraint propagators in constraint-based random stimuli generation for hardware verification. The cited evidence describes them as a response to propagators that take months to implement while hardware specifications can change on similar time scales.

Overview

Parametric propagators are described in the context of generic modeling for domain-specific constraint propagation in hardware-verification constraint satisfaction problems (CSPs). The motivating problem is that some constraint propagators are "extremely complex" and may require months to implement, while the hardware specification they encode can change on the same time scale.[1]

Purpose

The purpose of parametric propagators is maintainability under design evolution. The cited work reports an effort to generalize more complex domain-specific constraints and to develop propagators that are "relatively simple to manipulate and change" when either the design changes or a next-generation design arrives.[2]

Technical context

Parametric propagators appear alongside other CSP techniques used for constraint-based random stimuli generation in hardware verification. The same source discusses very large CSP domains, such as address and data variables with domains on the order of $2^{32}$ or larger, and notes that ordinary propagation methods can be inadequate when they rely on small domains.[3] It also describes conditional CSPs in which assignments to some variables can make large parts of the CSP irrelevant, and reports an extension of the MAC algorithm with assumption-based pruning to strengthen pruning under conditionality.[4]

Within this setting, parametric propagators address a different engineering pressure: the high implementation cost and volatility of domain-specific propagators. Rather than repeatedly implementing complex propagators from scratch for changing hardware designs, the approach reported in the evidence is to generalize those constraints into propagators with parameters that are easier to adjust.[2]

Reported application setting

The broader system discussed in the source is Genesys PE, a stimuli generator for verification of processors and multi-processor configurations. The source reports that Genesys PE had been used since 2000 as the major functional verification tool for IBM PowerPC processor designs, including unit, core, and chip-level verification and partial system-level verification.[5]

Notes on scope

The available evidence characterizes parametric propagators at the architectural and modeling level: it explains why they were introduced and what engineering problem they address. It does not provide a detailed formal definition, parameter language, or propagation algorithm for the parametric propagators themselves.

CITATIONS

5 sources
5 citations
[1] Motivation for parametric propagators: some constraint propagators were extremely complex and could require months to implement, while hardware specifications could change on the same time scale. Constraint-Based Random Stimuli Generation for Hardware ... - AAAI
[2] Purpose of parametric propagators: the authors generalized complex domain-specific constraints and developed parametric propagators that were relatively simple to manipulate and change for design changes or next-generation designs. Constraint-Based Random Stimuli Generation for Hardware ... - AAAI
[3] Huge-domain CSP context: the source reports CSP variables such as address and data variables with domains on the order of 2^32 or larger, and says regular propagation methods relying on small domains are inadequate for pruning such spaces. Constraint-Based Random Stimuli Generation for Hardware ... - AAAI
[4] Conditional CSP context: the source describes conditional problems where variable assignments can make extensive parts of a CSP irrelevant, and reports extending MAC with assumption-based pruning to improve pruning under conditionality. Constraint-Based Random Stimuli Generation for Hardware ... - AAAI
[5] Genesys PE application setting: the source describes Genesys PE as a stimuli generator for processor and multi-processor verification and reports its use since 2000 as the major functional verification tool for IBM PowerPC processor designs. Constraint-Based Random Stimuli Generation for Hardware ... - AAAI