One Coverage Database Without One Misleading Percentage
Statement, branch, toggle, FSM, functional and assertion coverage do not answer the same question. Why reduce them to one number?
A single coverage percentage looks simple—but simplicity can hide verification gaps.
If multiple coverage families are averaged together, easy-to-hit metrics may compensate for critical areas that remain untested. Even worse, a metric that was never collected may disappear from the final result.
vWIZ-Coverage keeps every metric family in one database while preserving each as a separate verification question.
One database, separate metrics
The coverage database can bring together:
- Statement coverage
- Branch coverage
- Toggle coverage
- FSM coverage
- Expression coverage
- Covergroups and crosses
- Assertion coverage
Each family retains its own numerator, denominator and collection status.
One database feeds the dashboard, but every coverage family remains independently visible.
This prevents a high toggle score, for example, from hiding an uncovered FSM transition or an untested functional combination.
Missing is not zero—or complete
There are three very different conditions:
0% coverage → Measured, but nothing was covered
100% coverage → Measured, and everything was covered
Not collected → No valid measurement exists
An unfed metric should not be silently treated as zero, full or irrelevant.
vWIZ-Coverage reports it as Not Collected, preserving the distinction between a verified result and missing evidence.
Connect coverage to the source
Summary percentages identify that a gap exists. Engineers still need to find it.
The code-coverage workspace connects statement and branch results to the source lines that produced them.
Covered and uncovered lines remain visible together in their source context.
This also keeps statement and branch coverage separate. Executing a line does not prove that every outcome of its decision was exercised.
Why this matters for signoff
A credible coverage workflow should answer four questions:
What was measured?
What was covered?
What remains open?
What was never collected?
Merging all coverage into a single percentage weakens those answers.
Keeping the families separate makes gaps visible, evidence traceable and signoff decisions easier to defend.
The takeaway
One database should create one place to investigate coverage—not one artificial number.
A green dashboard is valuable only when engineers can see exactly what contributed to it and what did not.
👉 Explore the complete vWIZ-Coverage workflow and technical evidence on WIOWIZ
#verification #semiconductor #vlsi #testing
Top comments (0)