Can LabVIEW simulation mode establish integration-test validity without hardware?

The goal is to define which parts of a LabVIEW VI integration test can be trusted when Simulation mode substitutes software instruments for real VISA devices, without physical hardware or production credentials. This question assumes LabVIEW 2020 or later; LabVIEW 2020 introduced improved serial VISA simulation fidelity, while earlier releases need separate review.

Simulation mode is not a complete model of instrument timing, timeout handling, error propagation, or property-node state. A simulated property value may be a default rather than the configured device value, while timing-sensitive branches may behave differently from a real session. The instrument model, transport, and exact LabVIEW version therefore affect what a passing simulation run proves.

The unresolved decision is where to place the boundary between simulation-only evidence and hardware-required evidence. A useful comparison must account for returned values, error data, and elapsed-time-dependent behavior rather than treating the final status alone as sufficient.

Which VI behaviors can be accepted as evidence from simulation mode alone? Which outputs require comparison with a hardware run? What test metadata is needed to distinguish simulated coverage from hardware-validated coverage?