cancel
Showing results for 
Search instead for 
Did you mean: 

CPU core voltage mismatch between motherboard censor and CPU censor

sunhang36
Level 7
Hi, first time post here.
My C6H got a weird issue. I got different readings from motherboard and CPU. As you can see in HWiNFO64, my 1700 run 3.9Ghz @1.419v (auto) but I got insane 1.504v from motherboard censors. I even got almost 1.4v @ 3.0ghz. The same thing happened to my ram. I set 1.35v in BIOS but I got 1.417v.65376

I've tried 0083, 1201 and 1401. Nothing changed. I suppose it is an issue for the motherboard chips? 6537465375
8,663 Views
10 REPLIES 10

alexjames
Level 7
I believe what you're seeing here is 'Vdroop', where CPU voltage drops when it's under very high load. My 1700 does the exact same thing.

The motherboard has to consume extra power/voltage in order to compensate for the reduction voltage being received by the CPU.

I could be completely wrong here, but that's the conclusion I've come to after a few weeks of testing. My 4670K did the same thing to a lesser extent.

alexjames wrote:
I believe what you're seeing here is 'Vdroop', where CPU voltage drops when it's under very high load. My 1700 does the exact same thing.

The motherboard has to consume extra power/voltage in order to compensate for the reduction voltage being received by the CPU.

I could be completely wrong here, but that's the conclusion I've come to after a few weeks of testing. My 4670K did the same thing to a lesser extent.



Nope, the CPU is not under high load. It's definitely not vdroop. I just get incorrect reading from motherboard. 1.4v @ 3.0Ghz and 1.5v@ 3.9Ghz

alexjames
Level 7
My bad, I should have noticed that given the low temps.

entropic-remnan
Level 9
I would add that neither "measurement" is accurate at all. On my mobo, the highest of those readings is off by as much as 0.070 volts. Usually about 0.050 volts or so. I might add that the mobo shows the voltage as being LOWER than it actually is. This actually surprised and frightened me when I measured it with a big overclock.

You should obtain a voltmeter if you don't have one (a so-called "multimeter") and read the voltage points provided near the mobo power connector. The results might surprise you.
Tired of trolls and mods that act like this platform has no problems and it's the users fault. Later.

HisEvilness
Level 7
Use CPUID HWMonitor and also remember load line cal is set to extreme that gives more volts then your offset on the bios.

JustinThyme
Level 13
Dont read too much into it. No BIOS is going to change it and even a multimeters will vary from one to the next, cheap meters are horrendous and are pretty much only good for detecting presence or absence of voltage, certainly not looking at something in the mV range and being anywhere near accurate. In my field I am required to have all my instruments calibrated at least once a year and even with those (my cheapo fluke 87 runs $400) the only one I trust for accuracy in the mV range is a Techtronics Oscope ($7500). None of this is going to matter as I cant get on the CPU pins to measure the Vcore and even a high impedance meter will pull it down. What is measured on the PSU rails has nothing to do with Vcore other than its the source that is provided to feed mosfets for regulation which is a positive feedback loop.

What does all this mean? Precisely squat as you cant measure at the CPU pins with an accurate meter. Watch your temps (these will also vary from point to point) and keep them reasonable. I dont even like seeing 80C much less anything higher. All the monitoring points are meant to be read more of as a change from one setting to the next than bawlz on accurate. Two identical systems with identical components and identical settings will 100% of the time not have identical readings. Components vary from one to the next even when made in the same batch. Good example that everyone knows? Silicon lottery.



“Two things are infinite: the universe and human stupidity, I'm not sure about the former” ~ Albert Einstein

entropic-remnan
Level 9
Ditto. Measured with a Fluke 173 and the reading disparity is real. Took out my Harbor Freight "always in the car" meter and it reads within 10 mv of the Fluke.

ASUS provides a Vcore measuring point and recommends it for accurate measurements. Nobody is reading the PSU, we are reading the test points ASUS specifically provided to fully verify Vcore -- nice post and I agree with most of it but it's irrelevant for what I was talking about.
Tired of trolls and mods that act like this platform has no problems and it's the users fault. Later.

If thats the case, then how can we be sure we're not pushing too much voltage to the CPU to cause degradation then? Its made much worse with the 1600X as it seems overclocking on those only relies on offset voltages. I can run 4.0GHz @ 1.375 (VID?) + 0.00500 with LLC 3 and that passes stability tests or what I run now which is 3.95GHz @ 1.375 - 0.0050 with LLC auto. Max temps were about 77C @ 4GHz and 68C @ 3.95GHz.

Also in that same vein, can someone explain to me the difference or the mismatch between Vcore and VID? Mine seems to be stuck at 0.988V (bios 1201) or 1.000V (bios 1401)

replace with a new board, now the readings are the same.