When I go to memory section of my X299 Dark’s BIOS, pick XMP profile for my memory that has 1.35V as voltage, and reboot back into BIOS I get BIOS reporting that voltage is 1.365V/1.365V, a 0.015V increase over XMP’s setting.
 
If I try to manually set voltage to 1.4V line changes to yellow which I interpret as warning. If I set it to 1.399V it goes back to white.
 
If I with voltage set to 1.399V boot back into BIOS now it shows voltage as 1.412V/1.412V, a 0.013V increase over my manual setting.
 
If I then boot to Windows with 1.399V manual setting, fire up HWInfo and take a look at memory voltages over a period of time (with readout once every 1000ms) I get following:
 
Under X299 Dark Fintek F71889A/879A section:
1.388V min, 1.399V max, 1.399V avg for DIMM 0/1
1.399V min, 1.411V max, 1.400V avg for DIMM 2/3
 
Under X299 Dark (EVGA) section:
0.874V min, 1.415V max, 1.412V avg for DIMM 0/1
0.874V min, 1.414V max, 1.412V avg for DIMM 2/3
 
So first why I see such intermittent huge drops under X299 Dark (EVGA) section of sensors that I don’t see under Fintek one? If I really had such drops wouldn’t system be crashing and not able to pass many hours of memory tests like it did?
 
Second, with X299 Dark (EVGA) section average values constantly overshooting desired values by approximately 0.013V does that, in combination with previous question and BIOS info imply that:
 
a) BIOS is using “EVGA” sensors, not Fintek ones, to display current values
b) HWInfo’s readout is in line with BIOS readout so it is not question of HWInfo reading sensor incorrectly
c) “EVGA” sensor readout can’t be relied on while Fintek one can?
 
Third, I expect some variation in voltages between DIMM 0/1 and 2/3 over period of time. Obviously my DIMM 0/1 can undershoot desired target as much as 0.011V while 2/3 can overshoot it as much as 0.012V, creating a maximum possible differential of 0.023V. Is this amount of variation within normal range?
 
Last, what is exact official reason why EVGA’s BIOS displays memory voltage entries in yellow moment you enter 1.4V?

PINKTULIP
It has been like that from day one 

You too have seen similar behavior? If yes which one of things I have seen you too have?

PINKTULIP

ZoranC

PINKTULIP
It has been like that from day one 

You too have seen similar behavior? If yes which one of things I have seen you too have?

Yes, all of them

... and it doesn't seem to be only DIMM voltage sensor, I have seen 12V readouts in [EVGA] section also show intermittent dips to so low values that they would be impossible if real without system having crashes. 

Edgesunito
have you tried different ram?

 
Different memory makes no difference.

Exnetic
Same voltage readings here. But never had a problem with OC.

... which seems to further confirm "EVGA's sensors" are off and can't be relied on.

arestavo
You could never rely on software read sensors to be 100% accurate. They're close, but not precision readings - you'll need some nice probes and voltmeters/ammeters if you want precision.

I'm not relying on software sensors to be 100% accurate, however I am expecting them to be close enough. DIMM sensor showing intermittent drops to 0.874V when it should be 1.4V or 12V rail sensor showing huge drops is not even close, it is so far to be completely unreliable.

wmmills
Being that the mem controller is on chip and connected with the SA, it probably drops down because of intels speed step/c1e to match the cpu speed.

 
Except that all of my cores are at same fixed speed and my C states are disabled. Besides, if voltage was really dropping wouldn’t that drop show in Fintek section too, not just EVGA one?

Change ram

metamana
Change ram

This has nothing to do with RAM and changing it doesn't do anything (already checked). Besides, it it had to do with RAM and not sensor itself voltage on 12V rail wouldn't show intermittent huge drops too plus whole system would be crashing left and right.