Possible Emergency? CPU at 1.475V rather than 1.35V after un-overclocking

[Chibigoat]

I overclocked my AMD Phenom II X6 1055t to 3.5 ghz from it's default 2.8ghz (3.3 turbo) about a month ago and it's been overclocked since, I just now took it back to the "Optimized Defaults" on the bios and made sure everything looked like it did previously. So it's back to 2.8ghz.

The problem is that the Voltage, according to Core Temp is sitting at 1.475 rather than its typical 1.35. The screenshot shows 1.408 because it seems to be jumping from 1.35 to 1.475 over and over again

https://imgur.com/a/pnIQ2

This should be what you need to see, let me know if you need more information.

This could be perfectly normal, I have no idea but I know that Core Temp sat at 1.35 before and during the overclock and now it's higher, is that just Core Temp's fault or what?
I'd just rather be safe than sorry.

Oh, and here is a cap of what Core Temp is saying:

https://imgur.com/oH3EH

 

[westom]

The problem is that the Voltage, according to Core Temp is sitting at 1.475 rather than its typical 1.35. The screenshot shows 1.408 because it seems to be jumping from 1.35 to 1.475 over and over again

Long before trusting those measurements, first a motherboard's digital voltmeter must be calibrated. A multimeter would do that. Measurements from the onboard voltmeter must be compared to what a digital multimeter also reports.

You know some numbers make no sense. -5 and -12 volts are clearly in error. Is that a defective divider (on motherboard)? Or is the PSU outputing a defective voltage? Calibrate each voltage. Only then can monitoring software report numbers that validate your concerns.

 

[keny]

Use CPUz or HWmonitor to get you're voltage readings and the jumping about of the vcore is normal, its coolNquiet doing its job of adjusting the vcore and CPU freq according to load, if you don't want it to do this you can dissable it in you're bios and then manually set you're vcore to as low as you can run prime at, but there's really no need to worry, as long as its not overheating you're ok.

 

[trents]

Optimized defaults would restore all the green stuff, probably. Cool N Quiet has already been mentioned but there is also C1E and Turbo that might be contributing to voltage fluctuation which as others have said is normal.

 

[Seebs]

CoreTemp is showing VID which is different to VCore.

VID (Voltage Identification)
The VID of your CPU is the default vcore your CPU needs in order to run at stock to run Intel’s standards for stability on any non-faulty motherboard. Some people have found that they can undervolt below their VID to still run at Stock speeds without failing stability tests.

Vcore (aka V-Core, CPU Voltage, Core Voltage)
Vcore determines how much voltage the CPU gets, and as you increase the CPU speed, you’ll eventually have to start increasing vcore as well.

The default vcore depends on your processor’s VID, which is different for every CPU and is set by Intel in the factory, but it’s usually around 1.1 – 1.2 V. The absolute maximum vcore voltage recommended by Intel is 1.4 V, so I’d recommend staying under that for everyday use.

So it is perfectly normal for CPUz to show you a VCore of 1.408V while CoreTemp shows you a VID of 1.475V.

Long before trusting those measurements, first a motherboard's digital voltmeter must be calibrated. A multimeter would do that. Measurements from the onboard voltmeter must be compared to what a digital multimeter also reports.

You know some numbers make no sense. -5 and -12 volts are clearly in error. Is that a defective divider (on motherboard)? Or is the PSU outputing a defective voltage? Calibrate each voltage. Only then can monitoring software report numbers that validate your concerns.

Nothing wrong with HWMonitor reporting the -5V and -12V lines. It's not a calibration error, or a software error... It's just how PSU/Motherboards manage voltages for all the different components.

 

[RGone]

VID (Voltage Identification)
is set by Intel in the factory, but it’s usually around 1.1 – 1.2 V. The absolute maximum vcore voltage recommended by Intel is 1.4 V, so I’d recommend staying under that for everyday use.

Seebs until you said that in the AMD CPU forum section everything was pretty good. His AMD cpu will not have that low of a VID about 90% of the time.

 

[trents]

CoreTemp is showing VID which is different to VCore.

VID (Voltage Identification)
The VID of your CPU is the default vcore your CPU needs in order to run at stock to run Intel’s standards for stability on any non-faulty motherboard. Some people have found that they can undervolt below their VID to still run at Stock speeds without failing stability tests.

Vcore (aka V-Core, CPU Voltage, Core Voltage)
Vcore determines how much voltage the CPU gets, and as you increase the CPU speed, you’ll eventually have to start increasing vcore as well.

The default vcore depends on your processor’s VID, which is different for every CPU and is set by Intel in the factory, but it’s usually around 1.1 – 1.2 V. The absolute maximum vcore voltage recommended by Intel is 1.4 V, so I’d recommend staying under that for everyday use.

So it is perfectly normal for CPUz to show you a VCore of 1.408V while CoreTemp shows you a VID of 1.475V.




Nothing wrong with HWMonitor reporting the -5V and -12V lines. It's not a calibration error, or a software error... It's just how PSU/Motherboards manage voltages for all the different components.

Did you mean to say Intel? This is the AMD CPU section.

 

[westom]

Nothing wrong with HWMonitor reporting the -5V and -12V lines. It's not a calibration error, or a software error... It's just how PSU/Motherboards manage voltages for all the different components.

-5 and -12V is determined by the PSU. But since -5 and -12 have almost no purpose, then software may be trying to read voltages from a motherboard voltmeter that does not measure those voltages.

Motherboard voltmeter cannot report accurate voltages until calibrated by something more accurate such as a digital multimeter.

If I remember, +12 volts also appeared marginal. Probably is fine. Unknown until motherboard voltmeter is calibrated.

 

[Seebs]

@ Trents and RGone

LOL. My bad. I copy/pasted that info from wikis. The "meat" of the matter still stands though. VID and VCore are two different things and therefore it is OK for CPUz to report one and CoreTemp to report the other. No need to worry about CoreTemp reporting 1.475V since that is not the VCore being supplied to the CPU (as shown in CPUz).


@ westom

Agreed.

I just don't know how one would calibrate the motherboard's voltmeters. If such a thing is possible; please do let us know how.

 

[trents]

I don't think there's a way to literally calibrate a motherboard voltmeter but conceivably there might be a way to build in an offset to the reporting software after measuring the actual voltages on the motherboard circuitry with a voltmeter.

 

[westom]

I just don't know how one would calibrate the motherboard's voltmeters. If such a thing is possible; please do let us know how.

Measuring many of those voltages is easy. The +5, +3.3, +12, and +5vsb are on wires from the PSU to motherboard. Red, orange, yellow, and purple wires accordingly. Compare numbers on the meter to numbers when reported by software.

Each voltage must be verified separately.

Measuring CPU voltages is difficult since how each adjacent CPU power supply works can be unique for each motherboard manufacturer. One way is to measure all power supply pins until finding a voltage is closest to what software reports.

Generally, if other voltages are slightly high or low, then that CPU power supply voltage should also measure slightly high or low proportionally

I was unaware that any CPU changes voltages on the fly. Generally, a CPU voltage was preset (by the CPU). And remained at that voltage.

 

[RGone]

Voltage to the cpu does certainly change on the fly. I have Cool N Quiet enabled and Win 7 set to Balanced mode and am doing next to nothing on the computer and my cpu speed and cpu voltage maybe 800Mhz and 1.275 Volts and load up an application and fly off at 4.1Ghz and 1.45Volts to the cpu.

And then put an FX procesor in the cpu socket and leave it operating in the manner AMD built it to operate at and the speeds and voltages will seem moving all the time with something like 7 P-states on the FX processors.

 

[keny]

All modern CPUs change voltage on the fly

 

[westom]

All modern CPUs change voltage on the fly

Modem CPUs do not. Some modern CPUs do. (how good is your eyesight?)

 

[keny]

It says moder"n" look closer

 

[Seebs]

Measuring many of those voltages is easy. The +5, +3.3, +12, and +5vsb are on wires from the PSU to motherboard. Red, orange, yellow, and purple wires accordingly. Compare numbers on the meter to numbers when reported by software.

Each voltage must be verified separately.

Measuring CPU voltages is difficult since how each adjacent CPU power supply works can be unique for each motherboard manufacturer. One way is to measure all power supply pins until finding a voltage is closest to what software reports.

Generally, if other voltages are slightly high or low, then that CPU power supply voltage should also measure slightly high or low proportionally

Well... Yeah. Those would be rather easy to monitor and calibrate for, but I think the whole issue with this thread was that OP noticed HWMonitor reporting a figure for VCore and then he saw CoreTemp showing a much higher figure. The issue there is not of "lack of calibration", but of the two programs reporting two different things.

I was unaware that any CPU changes voltages on the fly. Generally, a CPU voltage was preset (by the CPU). And remained at that voltage.

That's what the VID is. If you leave all voltage settings at AUTO in your BIOS; your motherboard will supply the CPU with whatever VID has been embeded into it. There are power saving features though and those will downclock/downvolt your CPU automatically... And then there is LLC; that will also make your VCore change up/down depending on the load put on the CPU.

 

[westom]

The issue there is not of "lack of calibration", but of the two programs reporting two different things.

Nothing is known until a voltage is confirmed by a multimeter. Maybe one program is reporting correctly. Which one? Meanwhile, another voltage (+12 volts) is marginal. Simplest answer involves simplest labor. Touch a probe to each wire. Read the voltage. Then know what actually exists.

Only reason to not do that? Nobody cares what voltages are.

 

[RGone]

Well "Chibigoat" the thread starter has not posted back at all. I guess he found his answer. Surely it was given above. Hope he got answer and maybe will come and add [RESOLVED] to his thread title. Luck to you man.

 

[Seebs]

Nothing is known until a voltage is confirmed by a multimeter. Maybe one program is reporting correctly. Which one? Meanwhile, another voltage (+12 volts) is marginal. Simplest answer involves simplest labor. Touch a probe to each wire. Read the voltage. Then know what actually exists.

Only reason to not do that? Nobody cares what voltages are.

I fear we're beating a dead horse into the ground here westom... I agree with you that the +12, +5, +3.3 lines would be easily "calibrated" for with a voltmeter, but the OPs question was about VCore and how two programs (HWM and CoreTemp) reported different figures... That is not a "calibration" issue; it's just two programs reading two different sensors on the cpu.

 

[westom]

That is not a "calibration" issue; it's just two programs reading two different sensors on the cpu.

The issue is accuracy. A motherboard voltmeter and how those numbers are reported by programs are both completely unknown. A motherboard voltmeter has a history of inaccuracy. Obviously at least one of two programs are also inaccurate. Which one? Multiple unknowns exist. All questions get answered simply by one minute with a multimeter. Which assumes the OP still wants an answer.