Originally Posted by Jpmboy
yeah, those guides are wrong period (and I know who's guides you mean). Disabled the VIN Tracker (tweaker menu) and lower VCCIN a lot. I run my 10980XE/Omega at 4.8 with 1.2V manual override VCCIN is at 1.7V LLC5 with droop under load down to 1.584V (under x265 4K 12x override) and INT. Blender bench droops to 1.596V. My 7080XE at at 1.7V VCCIN also.
On x299, Vcore kills are rare... VCCIN kills are not (for the few kills there have been).
VCCIN @ 2V is waaay too high.
ooh wow that's crazy but the good kind!
I'll give those tweaks a try tonight if the system hasn't frozen mid workday. So far we're going on about 10 hours after rebooting with a lowered VCCIN (dropped it from 2.100 to 1.87 last night before going to bed).
On a plus side - loop temps overnight are looking nice... CPU loop settled at 5c over ambient, graphics card loop 9c over ambient. Would be fun to see how close to ambient I can get the CPU loop [IMG class=inlineimg]/forum/images/smilies/smile.gif[/IMG]
Haha spoke too soon... Just froze again.
Going to do a little bit of hunting today for a 10980xe but looks like tonight will be upgrade ordering time.
Had to raise VCCIN to 1.82 to get it through a CBR20 run, insta-crashed with 1.80 and below. SA is set to manual 1.160 and VCCIO is set to 1.110 but both report higher in SIV here. Anywho, I think at this point something is definitely wrong with either the CPU's memory controller, this RAM, or maybe the mb. I could just run the RAM slower but that's counter to the whole reason this machine exists.
Also - I tried booting it up with the 4000 mhz DIMMS I have in the 9900ks rig and it won't even POST with their XMP profile which seems to also point to a memory controller weakness, no?