Originally Posted by Nikado7
Here's what I mean, now that I'm home and can test it. I used 2 multimeters just so I knew there was no error. Then compared it to HWiNFO. Here's what I got. Now you see what I mean. To me if the magic limit is 1.35 or 1.4, it should be the cap using a meter and under a load. Obviously HWiNFO isn't showing us the truth at all. Which would further question people's vcores they brag about, when they could be at an entirely different voltage. Most are at level 1 and look how far off it is.. Is this normal? Is everyone actually at way different voltages than they think they are? I'm beginning to wonder.
You measured with a meter from the onboard read points, right? Or did you measure from behind the socket?
In order to get an accurate reading, you need to measure directly from the VRM.
If you do not, your measurement will show a voltage rise due to 'voltage drop' across the power plane, also known as power plane impedance. This "voltage drop" causes what you MEASURE to be higher than what is really there.
It's basically the same thing as this: (notice the voltage read points are affected):
In order to avoid this, you need to measure as close to the VRM as possible. You want the "on-die sense" reading.
Even if you measure from the caps behind the CPU socket, you're going to run into the impedance issue.
tl;dr: HWinfo64 is accurate on the Maximus XI boards.
I quote from Elmor:
When measuring voltages on a motherboard and large currents are involved, it's important to use accurate measurement points due to resistance/impedance present on the board. The power plane for example can be modeled as a resistor connected in series between the VRM output and the target device, together with Ohm's law, U = R*I. The larger the current, the larger the voltage drop. If you measure the voltage at the VRM output, for example at the inductor, you'll see a large difference compared to the voltage measured at the CPU socket MLCC. What we're interested in is what voltage the CPU die is actually getting after passing through the output filter, power plane, socket and package.
Something to look out for is when you're seeing a voltage reported during load which is much higher than what you've set. It would require a negative load-line which is just not supported on any controller as far as I know. At 0 mOhm load-line, you get exactly what you set (Level 8 on M11).