Voltage, which one is correct

[Rikus78]

I was using CPUID HWMonitor to see what was the maximum voltage that mi 3570K reach at 4.000 mhz, and in BF3 was 1,14v

But if i use HW iNFO 64, gives me 1,24v

Wich program gives the correct value?

Greetings

 

[dragonstongue]

odd, seeing as hardware monitor does not give me cpu voltage.

cpu-z is accurate and is using an Asus board, Ai Suite definately is.

What voltage did you set in bios?

 

[drescherjm]

Wich program gives the correct value?

I would say None. The voltage sensor hardware is cheap and highly inaccurate and needs to be calibrated to you motherboard and software.

 

[Teitoku]

I was using CPUID HWMonitor to see what was the maximum voltage that mi 3570K reach at 4.000 mhz, and in BF3 was 1,14v
But if i use HW iNFO 64, gives me 1,24v

Does the value vary/change within either program? I'm wondering if one is reporting your stock VID while the other is attempting to measure voltage usage.

 

[Fooshnik]

Probably this ^. One might be VID, the other vcore. Link.

 

[Rikus78]

What voltage did you set in bios?

1 and Offset Negative 0,050

Greetings

 

[dragonstongue]

HWinfo reports VID(which can be on the $ for accuracy or quite off, depends on how accurately it was calibrated) HWmonitor-CPUz and a few others read the actual voltage chip is recieving according to Intel and AMD specs(sometimes this can be through sence pins, other times through VRM similar to the VID)

Can they be accurate, yes, software never truly is, however, if properly calibrated(as digital is a rough approximation anyways voltage is more of an analog signal where digital is a more absolute) they are usually quite good, this does very much depend on the chip/motherboard and such in question though.

I remember that Phenom II has an offset of a few degress celsius and thier voltage is offset a bit by factory to keep temps and TDP within guidelines, and my E8400 also had to be calibrated anywhere from 1-6c to be "accurate"

so 1v + .050, see what cpu-z tells you and compare it, real temp and core temp(I forget which) can also tell the voltage, as does everest and a whole slew of other programs.

I generally go by what CPU-z tells me, if I set to say 1.4v in bios and cpu-z tells me 1.41v then I know everything is the way it should be, but if it "droops" to say 1.38v when loaded I know I need to adjust LLC or cpu-vdda for my Phenom II. it would be nice to see all the proper volts though to know you arent over or under volting for things such as core voltage-memory controller-Dram-QPI/VTT etc.

 

[dragonstongue]

mmm, and voltage sensors need to be calibrated to software, because they are cheap XD. They infact rely mostly on hardware which makes the software reading it flawed more often then not, so is this not a double negative or am I not understanding something.

From the factory according to specs, they have to read within a certain tolerance i.e +/- .2% or something like this, nothing at all is perfect, not even voltage meters, they as well have a +/- so you will always get within a certain ballpark. Considering the meters are built with circuits and such just the same as the motherboard and the like, I think that the VRM or even the cpu itself is at least as accurate as almost any "test" we can do. The newer motherboards are probably very accurate(given they use the best digital circuitry, multi-phase VRM and high tolerance caps) overall as it is essential to keep thier power consumptions, temperatures and such in line.

I am not 100% sure what polls the actual socket/pcb itself as far as the various software is concerned, I know Asus AI Suite II and cpu-z read very very close to the same as what shows in my bios, bios to windows you will always lose or gain some Voltage according to Intel and AMD spec and how the maker(motherboard) decided to implement it.

So yeh, whatever reads closest to what your BIOS actually reports I would go by that, CPU-Z is quite accurate for "loaded" voltages i.e in windows, and I use open hardware monitor for other temps-voltages.