I see you are a smart persom
You understand what you are doing
PWM VID
The NCP81174 receives the PWMVID signal from the
upstream controller for the Vcore regulation. The signal is
decoded internally and passed to the VID buffer output
(VIDBUF), where the duty cycle is converted to a
corresponding signal between 0 V and 2 V. The VIDBUF
high level is derived from a precise 2.0 V reference voltage.
It is PWM so on pin 2 you should see this PWM on the scope. By the multimeter you measure True RMS of this PWM which gives you an integrated value of 0.75V. Check on the scope if you have it correct and measure it's duty cycle precisely.
The question is why does the GPU ask for 0.75V? Maybe a wrong or corrupted bios? Should it be 0.75V?
I don't have such card at the moment but I would like to measure it on a working 1060. All similar cards I have are already dead.
There is a normal CPU inside a GPU which executes code an ordinary microcontroller. It has PWM outputs like cheap Atmels, PICs, ARMs. So the bios may be bad. We must have a hope.
The NCP81174 can decode both 1.8 V and 3.3 V
PWMVID output levels. If the EN voltage input is more than
2.5 V, a 3.3 V PWMVID I/O is identified. If EN voltage is
above 1.1 V (Overall chip enable threshold) but less than
2.5 V, a 1.8 V PWMVID I/O is assumed.
Get a scope and check it... How many volts peak to peak is this PWM and what about EN level?
---
If you check the levels you could supply external PWM and test what will happen. If you have a scope with arbitrary generator... It will be interesting.