Originally Posted by
mdzcpa
I think I have it sorted!
I lowered the VCCSA as suggested but the problem persisted...and seemed to worsen a little....BSODs 2 times in a row when booting when I usually would only get the one. So I went through the VTT and VCCSA settings at automatic and took note of all the preset voltages at the various ram speeds (1600, 1833, 2133, 2400). Although I was downclocking the ram from spec to 2133 and to 1833 (from 2400) the XMP setting was auto adjusting the VTT and VCCSA down pretty dramatically. I just went in and set the VTT and VCCSA voltages manually to be one step higher. In other words I set the VCCSA and VTT for the auto 2400 settings with the memory running at 2133. Boot problem gone.
It would seem so basic to have tried these voltage increases earlier, but the system was so stable under stress tests with less votlage I just didn't consider needing more voltage for a boot issue.
I didn't want to over volt more than I needed, so I brought the voltages down a little with some testing. I ended up with VCCSA and VTT of 1.3v for running the ram at 2133 and the CPU anywhere between 4.7 and 5ghz. Those voltages are somewhere between the 2133 (1.25v) and 2400 (1.35v) auto presets. I'd prefer to run those voltages lower as I know I am bench stable at 1.25v with anything I can throw at it. Even the 1.2v you suggested is LinX, Prime, and 3D stable. I'm not sure if there is something else I should be adjusting instead of these voltages in order to resolve the boot issue, but for now I'm rock solid it seems. I assume under good water cooling 1.44v Vcore and 1.3v VTT and VCCSA will be fine long term. In the meantime i will tinker with other settings over the next few days and see if i can resolve the boot issue another way and lower the VCCSA and VTT voltages.
I appreciate your help in this. I'm pretty happy with a 24/7 5Ghz SB-E machine. In fact, with the boot issue under control, there may be a little more under the hood still.