[SOLVED] Undervolting i9-12900k advice and help wanted [ASUS]

Status
Not open for further replies.

SilverChad

Prominent
May 24, 2022
18
1
515
Hi people, Lately I built a new PC, I got a new 12th Intel Core i9 12900K and my Motherboard is ROG STRING Z690-F Gaming WI-FI
My cooling system is MasterLiquid ML360 Illusion with an extra fan in back of the case and 3 on front, I use latest windows 11 with all drivers updated.

When I start rendering and my CPU utilizes 100% it easily hits 100c on P-Cores, Seller says it shouldn't hit 100c with this cooling system, but it does.
I watched some YouTube videos on undervolting and they set Core Voltage (In BIOS) to Adaptive -0.0500, When I set this value system remains stable but nothing changes in heat, Still 100c.
So I tried-0.0800 based on another article but it causes HYPERVIZOR_ERROR BSoD on early rendering stage. I tried -0.10000 based on seller suggestion, It works fine with some apps like Cinebench and maximum hit becomes 92c but in software I use like V-Ray, Corona and VUE it throws CLOCK_WATCHDOG_TIMEOUT BSoD, Interesting thing is Cinebench works fine on using all cores even in Windows 10 with my alder lake CPU but same crashing software like V-Ray, Corona and VUE don't they only use E-Cores on windows 10.

I'm totally confused it's 2 weeks I got this PC and I can't even use it.
I hope you guys can help me out on the issue and give me some advices and solution.

Here's some questions beside my issue :
  1. Without undervolting only some of my P-Cores hit 100c and most of time it's P-Core #3, And P-Core #1 always stays at 89c, Can this be caused by CPU Silicon?
  2. Can this be caused by My 3090ti connected Vertically?
Here's a image of HW Monitor that shows heating cores :
Untitled.png


Regards


EDIT 2 : Some new shots for detail

1 - CPU Z Screenshot
CPUZ.png



2. IDLE SYSTEM
CPU-FAN-IDLE.png


3. UNDER LOAD SYSTEM
CPU-FAN-UNDERLOAD.png
 
Last edited:
Solution
What is the make/model of your case?

1. Normally, the pump will come with a mylar protective film that must be removed before you apply paste and install the pump.

2. while running, touch a finger to the pump.
You should feel a slight vibration.
Also check that the pump is connected to a motherboard pump header.
It should be running at 100% all the time.


If at idle, you see temperatures about 10-15c over ambient, then it is likely that your cooler is installed well and functioning.

Remember that ambient does not mean room temperature, it is higher because the heat generated by your graphics card is increasing the temperature that is used to cool the cpu.
Verify that the front fans are actually intakes. Dangle a tissue in front...
Did you take plastic off the CPU cooler? Is pump actually working? Do you feel heat being expelled from radiator when under load?

1.Did you take plastic off the CPU cooler?
I don't see any plastics, Here's image of it
photo-2022-05-24-16-52-59.jpg


2. Is pump actually working?
How can I find out?

3. Do you feel heat being expelled from radiator when under load?
Yes a little bit
 
What is the make/model of your case?

1. Normally, the pump will come with a mylar protective film that must be removed before you apply paste and install the pump.

2. while running, touch a finger to the pump.
You should feel a slight vibration.
Also check that the pump is connected to a motherboard pump header.
It should be running at 100% all the time.


If at idle, you see temperatures about 10-15c over ambient, then it is likely that your cooler is installed well and functioning.

Remember that ambient does not mean room temperature, it is higher because the heat generated by your graphics card is increasing the temperature that is used to cool the cpu.
Verify that the front fans are actually intakes. Dangle a tissue in front to check.
Verify that the intake fans are running at a high rpm.

You could possibly mount the radiator in front to get cooler air for the radiator.
But, such a move will then increase the heat of the air used for gpu cooling as well as the air used to cool the motherboard vrm heat sinks.

You might possibly try to increase the rpm of the front intake fans to get more cooling airflow.

It is normal for there to be a few degrees of difference among the cpu cores.
Some are located more distant from the cpu heat spreader.
I would not worry about that.

What is your motherboard bios level?
CPU-Z motherboard tab will tell you.
Current is 1403
https://rog.asus.com/motherboards/rog-strix/rog-strix-z690-f-gaming-wifi-model/helpdesk_bios
Several updates address performance issues.


In the motherboard bios settings, you may actually be overclocking.
Sometimes "load optimized defaults" does that.
Until you know what you are doing(and I don't) I would stick with defaults.
Look for the user forum for your motherboard.
That is where you may get more knowledgeable advice.

As a possible temporary solution,
Go to power plan settings.
Select balanced, and set the cpu max performance to 99%
 
Solution
set the cpu max performance to 99%
This trick disables Intel Turbo Boost and will have a 12900K running at its base frequency which is only 3.20 GHz on the Performance cores. No one that spends top dollar on a high performance computer is going to be happy running it at only 3.20 GHz.

https://ark.intel.com/content/www/u...2900k-processor-30m-cache-up-to-5-20-ghz.html

@SilverChad
Your screenshot shows that your cores are reaching the 100°C thermal throttling temperature when maximum package power consumption is at 221W. For comparison, at the 230W level, my Corsair H115i cooler is able to keep my 10850K at 80°C while running Cinebench R23 at a steady 5000 MHz.

https://www.corsair.com/us/en/Categ...oling/iCUE-RGB-PRO-XT-Coolers/p/CW-9060044-WW

98QQvwo.png


When a smaller cooler outperforms a bigger cooler, that means either your bigger cooler is defective, it was not installed properly or it is simply not that great of a cooler.

The temperature sensors that Intel uses are not 100% accurate temperature monitoring devices. Intel generally rates these sensors to be accurate to +/- 5°C so some variation in the reported core temperatures from core to core is completely normal, even when all of the cores are at the exact same temperature.

I have also found that Cinebench R23 can run reliably with less voltage compared to when running other software. As you discovered, adjusting the voltage based only on the stability of Cinebench is not a good idea for overall stability.

You can try installing your heatsink again. Clean and replace the thermal paste when you do this. A heatsink needs to be snug to the CPU for proper heat transfer. If this does not solve your problem then you either need to replace your cooler or you can go into the BIOS and reduce the turbo ratio limits to slow your CPU down a little. Running your Performance cores at 4800 MHz will help the temperatures without reducing performance too much. This is a much better compromise compared to completely disabling Intel turbo boost.
 
What is the make/model of your case?

...

Thank you for your time and detailed reply.

I have this case : https://www.coolermaster.com/catalog/cases/mid-tower/masterbox-td500-mesh-2021/

1. I don't have any idea if the thin film is removed or not, I don;t know how to check kit as well, My PC is assembled very tight it's very hard for me to check things out.

2. I did, I couldn't feel any difference much but I guess I felt a very very weak vibration felt like cycling.

My room ambient is ~25-28c and CPU temperature is ~38-42c at idle

I did tissue test, Top and rear fans doesn't move the tissue much in idle but under load they throw it, Front fans takes tissue in but I can't feel many difference between idle and under load, It seems just a little bit stronger to take in.

3. My bios level shown in CPU-Z is 0403 I added more images in first post of CPUZ and cpu rate + Fans RPM in idle and under load.
Should I update my bios firmware?

I didn't see any overclocking in bios, My system goes only ~5100MHz in idel, While rendering it's ~4800MHz in full load! I have no idea why...

I did 99% power balance but it doesn't change anything.
 
Last edited:
Your pc is wide open, that's not tight at all, plenty of room. Just saying.

Follow the tubes away from the radiator. They end up at that pump head with 4 screws. Directly under that pump head is the cpu. When the cooler was shipped and sold, there was a very thin protective plastic tape stuck to the back of the pump head. It's not uncommon for ppl to not see it, or forget to remove it, even pro's like Linus have missed it.

If that plastic was not removed, it'll insulate the pump head from much of the cpu heat.

Checking is easy, but will require having more thermal paste on hand, as that'll need to be replaced. Simply undo those 4 screws in stages until fully loose, then lift the pump head straight up, use a lint-free rag/paper towel/coffee filter to wipe away the paste. If the plastic is still there it will be self evident.

Take a good look at the top of the cpu. Paste should have covered pretty much all of it from edge to edge. The very corner tips aren't an issue.

Replacement is just as easy. Put a good sized blob of paste dead center of the cpu, with 12thgen make it an oval type blob, or use a credit card/spatula or similar plastic/wooden item to gently smear the paste completely covering the cpu in a very thin but consistent layer. Put the pump head (minus plastic if it had it) straight down with a little wiggle side to side. Start all 4 screws first, then tighten in stages in an X pattern until snug or the nuts stop. Doesn't require cranking down, but does require a good finger tight pressure. Can check with a screwdriver to be firm/stopped.
 
  • Like
Reactions: KyaraM
Hi, Some updates.
So it turned out the vendor assemblers connected liquid cooling fans reversed, So instead of exhaling hot air it was inhaling outside air.

However it didn't change much heat, idle cpu temp reduced by ~10c and underload it still hits 100c, This time I flashed and updated BIOS to latest and no more BSoD happened, Now underload it's between 88 to 95c, No throttling happens and performance increased much, Cinebench score is now ~28000

Thanks for all your information and help
Cheers
 
My system has had the same issue, tried 2 different coolers to be sure.

From reading, adaptive just doesnt work right, it pumps up the volts to what it thinks it needs then offsets based on that.. it could be 1.38v for instance.

Im closer to getting stability at 5.1 myself, but with just "offset" and + 0.050.. or leaving everything on auto cores i can get stable around 1.26 volts (which is still high). Im still trying to be sure its fully stable but its been a pain and im waiting on the thermalright cpu bracket to come to hopefully help with temps. I had tried the washer mod before swapping motherboards to the unify-x, opted to not risk possible errors due to the mod. I think with the bracket i should be low 90s on my most demanding software at 5.1ghz fixed.. or just settle for override 1.26 and the temps are probably "close" already.
 
My system has had the same issue, tried 2 different coolers to be sure.

From reading, adaptive just doesnt work right, it pumps up the volts to what it thinks it needs then offsets based on that.. it could be 1.38v for instance.

Im closer to getting stability at 5.1 myself, but with just "offset" and + 0.050.. or leaving everything on auto cores i can get stable around 1.26 volts (which is still high). Im still trying to be sure its fully stable but its been a pain and im waiting on the thermalright cpu bracket to come to hopefully help with temps. I had tried the washer mod before swapping motherboards to the unify-x, opted to not risk possible errors due to the mod. I think with the bracket i should be low 90s on my most demanding software at 5.1ghz fixed.. or just settle for override 1.26 and the temps are probably "close" already.

I solved the issue using TechPowerUp Throttle stop, Limit TPL to 175, All stable, Under 90c
 
  • Like
Reactions: boraokur44
Status
Not open for further replies.