The windowed watchdog part gives a higher level of fault coverage so is
normally preferred in high reliability or critical systems. The watchdog pulse
has to occur within a defined time window to avoid a reset. An example of a
fault that this captures is where the processor is stuck in a loop and
providing WDI pulses continuously at the clock frequency. This would be
detected as a fast watchdog fault by a windowed watchdog but would not be
detected as a fault by a normal watchdog.