Question Does the CPU set memory controller voltage based on the XMP profile?

[carbide]

Hi everyone,

I'm having issues with some vengeance 1866 ddr3 part number: CMZ16GX3M2A1866C9
9-10-9-27 2T 1.5v

Basically, it was fine and stable, now it's not. I can't run it memtest stable at it's XMP profile (which has been fine since 2014) though 1333 defaults is ok.

I looked in aida64 and it found, in the xmp section, the text "xmp voltage 1.50v, memory controller: 1.30v"

by memory controller do we mean, cache voltage (Vring) of the cpu, or system agent VCCSA?

Either way, when I've manually set these voltages, I'd never dare go over a max of maybe... 1.25v cache, and +0.1/0.2 offset on VCCSA (my board only has system agent offset voltages)

Maybe time to try 1.3v but, I'm a little nervous this would degrade the CPU?!

Thoughts?

 

[chris11d7]

Personally, I've had an Intel 4930K run at 1.4V just fine since mid-year 2014, and I leave the thing on all the time (40+ hours a week); From my research, 1.5 is the absolute maximum for Ivy Bridge E without causing long-term damage, although I do not know how different Devil's Canyon played with these higher voltages; I know the power efficiency in that generation sucked compared to newer systems, as this was the first generation that the CPU handled most of the duties of the north bridge (PLX, QPI, and some other controllers).
Before adjusting voltages, I would try to run a single stick of memory to see if you get the same issues at 1866MHz (looking at your manual for Z97i Plus, it doesn't matter which slot you use for single stick). I'd try each of the two individually, that would tell you if one of them can't handle the higher X.M.P anymore.
Let us know what you find! Thank you,

Chris Nelson

 

[carbide]

so, I think from further investigation, VCCIO digital is the memory controller voltage. I can only, again, set an offset voltage, and i can't see the actual voltage listed in the bios - I assume this is due to product model not being ROG/high end ATX despite being able to see full details of the mosfets current in/out and power in/out, VRM temp etc in hwinfo64.

VCCIO/VTT in hwinfo always shows 1.016v and when I've manually set a +0.035 offset, I've seen, an expected 1.050v in hwinfo64.

After resetting the bios, restarting, then setting the XMP profile (I've tried with and without MCE) I never see VCCIO reach 1.3v, (which I assume the XMP profile would lead the bios to do?) it's always 1.016v

Maybe I've just been lucky in the past at being able to run the sticks with mem controller so low. I will increase it, but from everything I've ever read on the web, it's very easy to degrade the memory controller...

I'll try running one stick, then the other, but I'm starting to wonder if these sticks are quite old - I.e. mem controller spec to 1.3v was for sandy/ivy ?!