Firstly, many thanks to all those involved - and sorry for the delay in posting an update.
I think I finally have stability. I've had about 5+ hours in both FH and WZ without crashing.
The short answer is the v511.09 SD (studio driver) appear to yield stability on my system.
The longer ansewr and methodology.
Background:
The OC scan in ASUS GPU Tweak was used to retrieve the driver "profile" for the card when running the ASUS v472.12 driver downloaded from the ASUS website. This profile shows an expected voltage and frequency chart. The assumption is that these are the "last known good" values for voltage and frequency.
AIDA64 was used to grab logs while running the FH5 benchmark tool. These logs were analysed to find voltages and frequencies hit during the benchmark run. These benchmark runs were repeated many times using the drivers mentioned below.
This is what has been learnt:
1. Using data from my 3080, the more recent v497.xx nVidia GR (game ready) drivers are pushing too high frequency on too low voltage in comparison to the v472.12 drivers supplied by ASUS.
2. Using data from Nikos 3070, we can see the same pattern. I need access to the "profile" for the 3070 using v472.12 to confirm what the expected values should be.
3. Just like a CPU overclock, this "too high frequency on too little voltage" is introducing the instability that yields a CTD (crash to desktop) and/or BSOD (Windows Blue Screen Of Death). This "too high frequency" is the "maximum boost" that is being attempted by settings within the driver.
4. When using these more recent GR drivers it appears to be possible to regain stability by forcing the cards to NOT boost too high. A decrease of 30 Mhz and increase of "Power and Temperature target" to 110% brought my voltages and frequencies in line with "last known good" values.
5. The new v511.09 SD (studio drivers) from nVidia use a voltage range that is simillar to the v472.12 drivers from ASUS and this driver appears to be stable.
More testing will be required when the next GR driver is released. For now, I am sticking with the SD driver with no overclock or underclock in place and I will feedback later.
Other things I would like to test:
@Ribellu: ROG-STRIX-RTX3080TI-O12G-GAMING and the 120 / 144 Hz issue.
"I have this error when I play in 4K by setting my screen to native 144Hz."
Can you try the SD driver? If you still get the BSOD / CTD crashes, can you generate the logs for 120 and 144 Hz for investigation?
