Skip to content
STIMSMITH

Coverage Points

Concept WIKI v1 · 6/2/2026

Coverage points are verification targets used to assess whether a design has exercised desired behaviors, combinations, or states during testing. In the cited hardware design verification paper, hitting all possible design coverage points is presented as a major verification bottleneck for complex designs, especially when random or constrained-random stimulus struggles to reach hard-to-hit scenarios.

Coverage Points

In the provided evidence, coverage points appear in the context of hardware design verification (DV). They refer to the targets that a verification process aims to hit so that the design's functionality is adequately exercised during testing.[1]

Role in design verification

The cited paper describes a verification workflow in which random and constrained-random stimulus are used to explore a design's behavior. While random stimulus can theoretically exercise all combinations given enough time, the paper states that, in practice, complex designs make this difficult within realistic schedules.[2]

Because of that, verification teams often need to steer the DV environment toward hard-to-hit combinations. In this framing, coverage points are the concrete objectives that the verification campaign tries to reach, and the time required to hit them can become a dominant schedule constraint.[3]

Why coverage points matter

The evidence specifically states that, as designs grow more complex, the effort required to guide verification toward all possible design coverage points becomes more challenging and time consuming.[3] This makes coverage points important both as:

  • a measure of verification progress, and
  • a practical driver of verification cost and schedule risk.[1][3]

Machine learning and coverage points

The cited paper proposes augmenting existing constrained-random DV flows with supervised learning and reinforcement learning. Its reported goal is to achieve DV objectives of full design coverage more quickly and with fewer resources.[4]

In the paper's examples, the machine-learning-based approach is reported to perform better than random or constrained-random methods on functional coverage and on reaching complex hard-to-hit states.[5] This positions coverage points as a central optimization target for automated verification strategies.

Scope of this concept in the evidence

Based on the provided source, the term coverage points is supported specifically in the context of integrated-circuit / hardware design verification. The evidence does not provide a formal universal definition beyond that usage, so this article reflects the term as used in that verification setting.[1]

[1]: The paper discusses design verification schedules in which hitting all possible design coverage points is a central objective. [2]: The paper states that purely random stimulus has difficulty exercising all possible combinations in a timely fashion for highly complex designs. [3]: The paper states that verification time to hit all possible design coverage points can become the dominant schedule limitation. [4]: The paper says its machine-learning-based approach aims to achieve DV objectives of full design coverage on an accelerated timescale and with fewer resources. [5]: The paper reports better results on functional coverage and reaching complex hard-to-hit states than random or constrained-random approaches.

LINKED ENTITIES

1 links
Optimizing Design Verification using Machine Learning: Doing better than Random MENTIONS

CITATIONS

5 sources
5 citations
[1] Coverage points are used in the evidence as verification targets in hardware design verification. Optimizing Design Verification using Machine Learning: Doing better than Random
[2] Purely random stimulus can theoretically exercise all combinations, but in practice it struggles to do so in a timely fashion for highly complex designs. Optimizing Design Verification using Machine Learning: Doing better than Random
[3] The verification time required to hit all possible design coverage points is described as a dominant schedule limitation as designs become more complex. Optimizing Design Verification using Machine Learning: Doing better than Random
[4] The paper proposes supervised learning and reinforcement learning to enhance constrained-random DV and help achieve full design coverage more quickly and with fewer resources. Optimizing Design Verification using Machine Learning: Doing better than Random
[5] The paper reports improved functional coverage and better reachability of complex hard-to-hit states relative to random or constrained-random approaches. Optimizing Design Verification using Machine Learning: Doing better than Random