X370 Killer SLI/AC Voltage jumping all over

I also suggest using HWiNFO64 in general over any other monitoring program. The author goes out of his way to accommodate every different mother board, CPU/chipset platform, sensor chips, you name it.

I forgot to answer a question you asked earlier, about using +60mv.

Millivolt, or mv, is 1/1000 of a volt, or 0.001V. That is just one millivolt.

Your setting of +60mv is equal to 0.060V. That is 60/1000 of a volt, or 6/100 of a volt.

Denroth wrote: So what is the difference between the voltage below the frequency value (in the Bios) and the lower one in the external cpu voltage? on hwinfo Voltage 1: vcore Voltage 2 (external voltage): CPU core voltage (SVI2 TFN)

Good question, I am not 100% certain about what HWiNFO64 is displaying for each of those values.

I'll give you my opinion, but first a question for you, do you get anything in HWiNFO64 that has an entry like:

ASRock X370 Killer SLI/ac (IRF IR35204)

With several lines below it with temperature, voltage, and power readings?

I do, that is the CPU VRM readings from the sensor chip.

Back to HWiNFO64, the CPU (SVI2 TFN) voltage matches the CPU VRM voltage output reading. The maximum voltage for the VCore and CPU (SVI2 TFN) voltage matches what I have set the VCore to be in the UEFI, off by 0.001V +/-, within the error and resolution for each reading.

When any or all of the power saving features I mentioned previously are active, the VCore will drop at low CPU load conditions. Since I don't mind the CPU voltage dropping, I have those feature enabled, and I see the VCore drop to very low values. But each reading in HWiNFO64 for or related to VCore will be different at the low levels. The CPU VRM (and CPU (SVI2 TFN) ) output only goes down to just under 1.0V, while the VCore will go below 0.400V. So these two readings are a bit complicated. But what I just described is not really what your concern really is.

You seem to want a rock solid, unchanging VCore while stress testing. Is that even possible? Or even something that is desirable? A little VCore VDroop is known to occur, and is considered normal.

How much variation do you get in the VCore? Do you know that CPU stress tests are not 100% constant with their loads. They change to different instruction being executed during a run, and switch between testing the integer and floating point arithmetic processing sections of the CPU too. So while the CPU "load" may always read 100%, different things are being tested at different times.

So the VCore can and will vary because some things require more VCore than others. Intel had a big controversy with the new AVX2 instructions in their processors, that when being executed cause a huge increase in the VCore compared to anything else.

Are you having failures during stress testing the CPU?

with offset at -50mv = no post at all.

man, i forgot how convenient having a clear cmos button or having dual bios was!

i haven't touched jumpers in years.

anyhow, cleared cmos and now back to 1.325 + 65mv offset.

nycalex wrote: wardog wrote: nycalex wrote: however, i only have 2 options for vcore offset, it's either auto (then voltages really jump high) or a set number starting with +60mv. this is the reason i went with +60mv. This confused me. Are you aware you can override Offset and Fixed (milli)voltage by typing in an appropriate and within limits value? there's no option to type in an amount. it's a drop down menu. so it's either AUTO or +65, +130, +200, +250, +325, +400, -50 and -100mv i'm currently testing stability with -50

Phew!

I thought you were implying having no optional choices. Drop-downs ........

Thanks for clarifying that for us.