SOLVED!Has anyone gotten to the bottom of CPU PLL OC/VCCPLL OC voltage directly affecting Core temps

Speedeu4ia

Limp Gawd
Joined
Aug 7, 2017
Messages
392
UPDATE!! changing this does make temps not accurate see end of post


Stumbled on this, pretty crazy actually, Z270 and Kabby lake for sure have this not sure on others yet

heres the breakdown with temp graphs for different CPU PLL OC voltages or in my BIOS(gigabyte) VCCPLL OC
https://communities.intel.com/thread/112262

Soo my personal test is on Z270 gigabyte gaming 5 itx board
my 7700k I OC'd to 4.8 at who knows what voltage. Im set to normal and -.065v offset in bios and LLC on normal. In HWINFO64 shows VCORE from 1.26-1.308V under load durring AIDA64

Anyway on AIDA64 with cpu/FPU(hottest I can get) I was getting 81C.

I only change VCCPLL OC in bios from AUTO(shows 1.25V) to 1.15V and now max is 76C. I could instantly tell this was working as I ran something that normally would keep my cpu fan 100% but now it was throttling the fan

My idle temps are down as well(22-23C), almost to ambient(21-22C), but mine underclocks to .7v and 1.7ghz

Does not affect stability

So the question is what is the correct voltage to calibrate the temp sensors? Currently I think mine are a tad low at 1.15

mine defaults to 1.25v, seems others are different and enabling XMP may change it also(havent tested mine)


Update: So I took matters into my own hands. disconnected all of my fans and CPU fan. Ran aida64 for heat. results are that Yes changing the voltage makes the temps Not Accurate!

1st test was 1.21v VCCPLL OC 1st pic. It throttled at 99c

2nd test was 1.13v VCCPLL OC 2nd pic, now the temps are lower but it still throttled at 96C

So my boards default of 1.25v is likely correct(can only adjust odd voltages anyway 1.21,1.23,1.25 ect)

I know everyone was hoping for it to be a solution to the high temps of Kabby Lake
 

Attachments

  • throttle.png
    throttle.png
    1.6 MB · Views: 43
  • throttle2.png
    throttle2.png
    1.1 MB · Views: 36
Last edited:
I feel like I'm reading this wrong because it reads like you are trying to say you can adjust the voltages on your temperature probes when, I believe, those get a set voltage (5v?) and then the resistance is changed by the temperature of what it's touching which is measured and you use that to determine the temperature. There should not be any temp sensor on your motherboard that can be calibrated.
 
I looked at a few other forums related to this topic and I believe what he is talking about is that if you lower the CPU PLL voltage you change the REPORTED temps to the MB, potentially making them very wrong (reading lower than they are) and the system fans and throttling won't function as they should possibly leading to damage or inexplicable instability.

In short: don't set the CPU PLL voltage lower than it is supposed to be or you may cause this... you aren't actually lowering your CPU temp.

Do I have this correct?
 
I looked at a few other forums related to this topic and I believe what he is talking about is that if you lower the CPU PLL voltage you change the REPORTED temps to the MB, potentially making them very wrong (reading lower than they are) and the system fans and throttling won't function as they should possibly leading to damage or inexplicable instability.

In short: don't set the CPU PLL voltage lower than it is supposed to be or you may cause this... you aren't actually lowering your CPU temp.

Do I have this correct?


its not CPU PLL its a different voltage called either CPU PLL OC on MSI boards and maybe Asus. On my board(gigabyte) its called VCCPLL OC.

But yes this voltage on my board auto sets to 1.25v on the MSI board its auto=1.20v

So what I'm saying is what is the correct voltage for this to have the correct temps? And why isnt it just calibrated and locked/invisible in BIOS

Lowering this voltage to 1.15v lowers temps reported in all software.
 
I feel like I'm reading this wrong because it reads like you are trying to say you can adjust the voltages on your temperature probes when, I believe, those get a set voltage (5v?) and then the resistance is changed by the temperature of what it's touching which is measured and you use that to determine the temperature. There should not be any temp sensor on your motherboard that can be calibrated.
Exactly my point I should not be able to change this random voltage and make my temps drop 5C at load and 2-3C at idle(or more if I change it lower). But this is exactly what happens on many peoples z270 boards with Kabby lake(havent came accross other reports yet)

https://forums.tweaktown.com/gigaby...pll-oc-bios-reported-temperature-correct.html

http://www.overclock.net/t/1621111/i7-7700k-temps-vs-ram-speed-on-z270-mainboard
 
Last edited:
In this overclocking guide for Z270 they mention this voltage referring to Cold bug and LN cooling

"VCCPLL OC: You’re looking at values around 2.0V. It helps if you’re hitting the CB earlier than you should (-130C or so). Some CPUs need higher but 2.0V is a good starting value. Keep this value under 3.0V or you might degrade your CPU."

https://overclocking.guide/gigabyte-z270-overclocking-guide/


half way down here
https://intel-openport-v7.hosted.jivesoftware.com/thread/110728?start=1035&tstart=0

This one has BS feedback from MSI saing it helps with OC(which its proven to not be. It also talks about the Intel spec for this voltage should be 1.0v+/-5%

it also changes the MB cpu reported temps which is why some believe its really lowering temps.

I have a feeling if I set mine to 1.0v I will have below ambient idle. Soo Intel spec says 1.0v but thats not correct and my board says 1.25v and I'm not sure thats correct and MSI seems to set 1.20v
 
Last edited:
"VccPLL_OC power rail (which in the MSI Z270 M7 is set through the parameter CPU PLL OC Voltage) should be sourced from the VDDQ VR. The connection can be direct or through a load switch, depending desired power optimization. In case of direct connection (VccPLL_OC is shorted to VDDQ, no load switch), platform should ensure VccST is ON (High) while VccPLL_OC is ON (High).



VDDQ is the DRAM Voltage, which explains why when CPU PLL OC Voltage is set to auto, it takes whatever value the DRAM Voltage parameter is set to (by default 1.20V).



Interestingly enough, according to MSI, any voltage starting at 1.31V is not "safe" for CPU PLL OC Voltage. That means that anytime we enable XMP, VccPLL_OC is set to a non-safe value."



Still no clue why we get that core temperature difference.

This seems to be the technical specs, should be same as DRAM voltage, although mines 1.25v when default is 1.20v and my ream is set at 1.35v still doesnt add up, and not sure why it would alter temps like it does
 
"VccPLL_OC power rail (which in the MSI Z270 M7 is set through the parameter CPU PLL OC Voltage) should be sourced from the VDDQ VR. The connection can be direct or through a load switch, depending desired power optimization. In case of direct connection (VccPLL_OC is shorted to VDDQ, no load switch), platform should ensure VccST is ON (High) while VccPLL_OC is ON (High).


VDDQ is the DRAM Voltage, which explains why when CPU PLL OC Voltage is set to auto, it takes whatever value the DRAM Voltage parameter is set to (by default 1.20V).


Interestingly enough, according to MSI, any voltage starting at 1.31V is not "safe" for CPU PLL OC Voltage. That means that anytime we enable XMP, VccPLL_OC is set to a non-safe value."

I realize I'm a little late to the game, but CPU PLL OC Voltage (AID64) seems to be at 0.592V (for an XMP Profile):


Voltage Values
CPU Core 1.264 V
+3.3 V 3.312 V
+5 V 5.000 V
+12 V 12.192 V
+3.3 V Standby 3.376 V
VBAT Battery 3.136 V
CPU PLL 1.240 V
CPU PLL OC 0.592 V
VCCMP 1.145 V
VCCPRIM 1.008 V
VCCSA 1.216 V
VCC Sustain 1.173 V
IMC 1.152 V
DIMM 1.344 V

How should I interpret that? Just add the value to CPU PLL? (If so, at 1.832v, things would be grossly overvolted somehow).
 
Back
Top