So I copied the settings of that bios DeathAngel74 and found out a few interesting things. One, which I already knew from messing about, is that certain power limits will cause an nvidia kernel BSOD when MSI afterburner starts putting you in a boot loop unless you are fast enough to flash another bios before it BSODs, in which case the adapter disables which stops it even after it re-enables, so there's no risk of a brick.
Just to be safe since I've had the Maxwell fan settings cause a BSOD as well I changed everything back to my bios but left the voltage like yours. I am very happy to report that 1250mv actually runs with the same or within a few % of the stability 1313mv provides. At least at 1493 | 3900. As soon as temps get to about 82c the risk of errors is there, and by the time it's at 84c I almost always have an error. 87c and there are too many errors to call it stable, hundreds quickly instead of 10 every minute or so. So temperature is a big factor after you have "enough" voltage that more won't help but possibly hurt. I'm also pleased to report peak wattage was 484w instead of 505w during the OCCT stress test! Idle temps are at 37c instead of 47c.
Laithan, considering my card is "crap" it might be worth a look changing/lowering the voltage steps because even mind can handle it. You could smooth out that rough voltage transition. It also really wouldn't hurt to have the max voltage on clock 74 AND 73 just in case..
Lastly, why don't you run with the TDP Entry, Base Entry, and Boost Entry all on State #0? It throttles when it's set to #1, #2, #0, even if you change the clocks to the same thing.
I'm going to try to modify the fans and possible test up and down in voltage while I'm still awake. I've been awake for almost 30 hours... (I am really bad at this whole" keep a sleep schedule" thing)... Edited by cookieboyeli - 10/27/15 at 6:37pm