In FortiOS 7.6, when a Performance SLA probe mode is set to Prefer Passive, FortiGate attempts to measure link performance using passive monitoring first, based on real user traffic. Only when passive monitoring is not possible does FortiGate temporarily fall back to active probing.

With Prefer Passive, FortiGate passively monitors TCP traffic flowing through the SD-WAN member to calculate SLA metrics such as latency, jitter, and packet loss. This behavior directly matches option A.

During passive monitoring, FortiGate relies on observed traffic to infer link health. Because no synthetic probes are sent, a completely dead link (with no traffic passing) cannot be detected by the SLA during passive mode. As a result, dead members may not be immediately detected, which makes option D correct.

Option B is incorrect because there is no fixed 3-minute timer defined in FortiOS 7.6 that forces a return from active probing back to passive monitoring.

Option C is incorrect because passive SLA monitoring is based on TCP traffic, not ICMP traffic. ICMP is used for active probing, not passive monitoring.

Option E is incorrect because traffic subject to passive SLA monitoring cannot be offloaded to hardware. Passive SLA measurement requires software inspection of packets, which prevents NPU offloading.

Therefore, the two correct observable impacts of configuring the probe mode as Prefer Passive are A and D.