Skip to content
STIMSMITH

Markov Model Stimulus Generation

Concept WIKI v1 · 5/28/2026

Markov Model Stimulus Generation is a feedback-driven verification approach in which Markov-model-based stimulus constraints are adjusted during simulation to improve coverage or switching activity. The evidence specifically describes StressTest, a tool based on feedback-adjusted Markov Models for microprocessor verification.

Overview

Markov Model Stimulus Generation refers to the use of Markov-model-based techniques to guide stimulus generation during hardware verification. In the cited evidence, this approach appears as feedback-adjusted Markov Models used by the tool StressTest to verify microprocessors.

The broader verification problem addressed by this class of techniques is automated test application: given available directed or parameterized constrained-random sequences and functional coverage goals, the goal is to apply stimuli in a way that improves verification coverage across multiple trials.

Feedback-adjusted Markov-model flow

The evidence describes StressTest as a tool based on feedback-adjusted Markov Models. Its stimulus-generation process is closed-loop:

  1. An engineer provides a template describing points of interest in the design under verification, also called probe nodes.
  2. During simulation, the tool observes activity at the probe nodes.
  3. Closed-loop feedback techniques adjust or direct the test-generation engine.
  4. The generator is steered toward stimuli that produce higher switching activity at the probe points.

This makes the approach an on-the-fly optimization method for stimulus constraints rather than a purely static or unguided random-generation method.

Role in verification automation

The evidence places this approach in the test-application phase of design verification automation. Automation is described as important for handling modern design complexity and the growing number of test scenarios. In particular, automated test application aims to achieve coverage closure with reduced regression time.

Reported benefit

The cited source reports that StressTest achieved better coverage in fewer cycles than random generation techniques when using feedback-adjusted Markov Models and probe-node activity feedback.

LINKED ENTITIES

1 links

CITATIONS

5 sources
5 citations
[1] StressTest is a tool based on feedback-adjusted Markov Models proposed to verify microprocessors. [PDF] UVM-based verification of RISC-V superscalar processors
[2] StressTest can optimize stimulus constraints on the fly but requires engineer assistance to provide a template describing points of interest, or probe nodes, inside the design under verification. [PDF] UVM-based verification of RISC-V superscalar processors
[3] During simulation, StressTest observes probing-node activity and uses closed-loop feedback to direct the test generator toward stimuli that create higher switching activity at the probe points. [PDF] UVM-based verification of RISC-V superscalar processors
[4] The reported results showed better coverage in fewer cycles than random generation techniques. [PDF] UVM-based verification of RISC-V superscalar processors
[5] The broader verification context is automated test application for maximizing verification coverage over multiple trials given constrained-random sequences and functional coverage goals. [PDF] UVM-based verification of RISC-V superscalar processors