I have been playing The Division and almost every time i get a crash at some point - the dreaded black screen / fans go to 100% / reboot.
I also get quite frequently the 'Default Radeon Wattman settings have been restored due to unexpected system failure".
I have a 16:10 (2560 x 1600) monitor with a fixed 60Hz refresh, so I have been using a combination of FRTC (60fps) combined with the Auto Undervolt GPU feature (using Radeon Settings 19.3.1). I don't get as many crashes when using just FRTC on its own. Perhaps FRTC and Auto Undervolt GPU don't play nicely together?
I have not yet tracked down the reason behind the game crashing, but have read that the Auto Undervolt GPU feature might be the culprit - but I have read opinions claiming that it is either too conservative or too aggressive compared with trial and error manual undervolting. I have also read that crashes might be caused by an uneven distribution of contact between the die and the copper base of the heatsink, causing localised heating seemingly beyond the capacity of the reference cooling solution to manage.
Can anyone give me some tips/tricks to a logical process to try to identify the cause. For example, how do I record the Junction Temp in real time while gaming as the standard Radeon Performance Monitoring Data recording does not record Junction Temp as far as I can tell, and I can't tell if GPU-Z does not record Junction Temp either. Am I missing something obvious in not knowing how to record Junction Temp to a file that I can look back after recovering from a crash?
I really like Vega 20 in all its 7nm and 16GB of HBM2 glory, but it is a very expensive card and having crashes while gaming is quite frustrating. I am really hopeful to work towards understanding what the root cause is (heatsink vs Auto Undervolting vs driver vs Win10), but would be keen to hear of other's experience with this card and any if there were areas I should focus on.