What can cause Vcore to be higher than stock setting?

[mormakil]

I'm having this problem some time ago but now things are getting worse.

One day, after a stupid mistake from my part, the mb auto shutdown because of CPU overheating (110ºC). Then MBM5 started to show 1.85volts Vcore instead of 1.65 . I turned it on and off several times and with doing nothing else it went back to 1.65. It worked normally for some weeks, with F@H 24 hours, but some days ago, I went to check CPU temperature and it was 70ºC at full load, because Vcore was again 1.85.

I modded my Antec TruePower some months ago and managed to damage the temperature/noise controller , so I suspected it was PSU's fault. I tried another one and worked at 1.65, but when I tried again the Antec it was 1.65V too. Wishing it really was PSU fault, I bought and Aopen PSU (a FSP-300-60PN) and installed it today.

But, only 1 hour after leaving it running F@H Vcore raised suddenly to 1.87V, and some minutes later it was 1.95V . That raised an alarm because of temperature over 70ºC and my brother shutdown the PC. While I'm writting this Vcore is 1.95V , I'm using CPUidle in order to keep this thing cool.

So, what do you think is causing this behaviour?, MB, CPU?. How does the CPU tells the motherboard it's stock voltage. I can't set Vcore in BIOS (well I can set a 5%,7.5% and 10% overvolt). My system's spec are bellow.

[dukla2000]

How does the CPU tells the motherboard it's stock voltage.

AMD have a set of bridges on the CPU that flag (via CPU pins) to the mobo the VCore spec.

I'd blame the mobo - can you RMA it? To prove things you need to measure the VCore on the mobo but is really difficult and 1 slip of a multimeter probe can fry things you didn't mean to! This would eliminate MBM etc as a source of error, but as you know your temps are climbing seems the VCore genuinely is also climbing, which has to be the mobo (IMHO).

[lm]

Which mobo it is?

[mormakil]

Gigabyte 7VAX rev 1.1

The worse part is that I replaced northbridge heatsink

Maybe I'll try with another CPU on this Mobo, just to check it behaves the same way with another one.

[dukla2000]

Maybe I'll try with another CPU on this Mobo, just to check it behaves the same way with another one.

Conceivably it could be the CPU, e.g. if you made a real hash with the thermal paste and have residue around the bridges (IIRC the VCore options are around the L11 bridges). Worth a try and make sure the CPU is scrupulously clean.

Also, even if the mobo is playing up, if you use a CPU with a different spec VCore you may avoid the faulty part. (Extreme long shot this!)

Except (again IIRC) the VCore is read/set by the BIOS during POST and is not changed thereafter (unless you have a mobile or A-64)?

[mormakil]

Except (again IIRC) the VCore is read/set by the BIOS during POST and is not changed thereafter (unless you have a mobile or A-64)?

In fact, Vcore changes suddenly while running, some times from 1.65 to 1.85 and some times from 1.85 to 1.65, but when it changes it usually stays in the new value for hours. And now it's even reaching 1.9 and 1.95 values.

I've a Palomino 1800 (Vcore 1.85) that I'll try and I'll also check the thermal grease, however I don't think that can be related because it's not a permanent bad Vcore setting but a ramdomly changing one.