CPU is running hot at Idle or low loads

[kennyg169]

i have noticed this issue lately when playing games like star citizen, The Finals or COD, The CPU "package temps" get into 90's C. NZXT cam saying its in around 70c at its peak that's with having an AC unit blow at the computer to help give cooler air. It started to crash and give me the blue screen. I repasted the CPU a few days ago and it kind of help, but the idle temps are still higher then when i first water-cooled this a few months ago.  My GPU stays so cool and never gets above 55c at its peak during game play and 25C at low load or idle. anyone got ideas? photo of the loop order and iCue at idle. The loop order is pump/res to GPU, 480mm rad, CPU to motherboard, 420mm rad back to the pump.

 

motherboard ROG Maximus Z690 Formula one

I9-12900K

MSI gaming trio X 4080

1300w PSU

[c-attack]

And this is with the pump at ~4400 rpm?   I don't see any physical reasons in CUE for this.  Water and air temps all excellent.  Probably need to do a Vcore check.  That's the other piece of the puzzle.

 

The 13/14th gen CPUs are getting all the attention, but heavy handed default BIOS settings have been around for a good 10 years.  Try running HWinfo or HWmonitor while you're gaming to get a decent look at peak Vcore and average Vcore (motherboard section in those apps).  It's the peak value that will likely tell us what we need to know.  If you're seeing values near 1.50v, then you need to trim the voltage back in the BIOS.  That's not too hard on Asus boards.  

 

 

[kennyg169]

right now, my core voltage at idle is 1.385 V

Edited August 15 by kennyg169

[kennyg169]

I'm still having the issue with keeping my CPU from overheating and crashing games. This shouldnt be an issue at all with this build?

 

[c-attack]

TJ Max on a 12900K is 100C. With your baseline liquid temp at -25C, that’s a +75C differential. That’s an incredible value, way beyond what someone really pushing their overclock should see. It shouldn’t be possible when gaming. Are you sure the crashes are cpu temp related? 
 

Try testing in a controlled environment. Go back to cpu-Z and the stress test in the bench tab. It’s linear and below the power limit. Probably about 210-225W on a 12900K and this should have a negligible effect on the coolant. However, note the immediate cpu temps after the test starts. You only need to run it for a few seconds. Very high cpu temps suggest voltage or contact issues. 

[kennyg169]

this is when I'm playing the finals 

 

[kennyg169]

I repasted the CPU again with Thermal Grizzly - KryoSheet this time and no change. I'm about to throw away this damn computer

[c-attack]

Are you sure the Vcore is really only getting to 1.38v when gaming?  Use a program like HWinfo or HWMon to track the peak/avg.  You also can add the value to the CUE dashboard through the MB panel, but you might want to drag only the Vcore out into it's own box and trash the rest of the temps.  

 

Also, what frame rate are you targeting in games?  Someone reminded me how much extra CPU power you can use when running at 240 this morning.  Quite literally double what mine does at 120.  

[kennyg169]

So I thought about the FPS part and I did go from 240 to 170 and it made no change on my temps or how it was running.  Maybe the thermal sheet moved during the install of the water block. I might pull the block again to check it.

[kennyg169]

I also think im not getting enough pressure on the CPU from the water lock. When i used an AIO never had any over heating issue. Same motherboard and CPU. But when i would install the AIO block I would never bottom out the stud to the back plate. With this XC7 waterblock I will bottom out the stud to the backplate

[c-attack]

I think a remount is probably worth while.  If it's not voltage, then it has to be physical conductivity.  

[kennyg169]

I fixed it, the loop order was wrong. I needed to go to CPU first

[c-attack]

Loop order should have a minimal impact on component temp.  You might pick up 1-3C difference depending on loading.  It looks like your original went pump->GPU->side radiator-> CPU-> VRM->top radiator and back. That should be a very small 1C add to CPU temp when the GPU is running heavy.  My 4090 blocks goes directly into my CPU block without seeing temps like that.  Is it possible there was some type of pressure build up slowing the flow in the CPU block?  It looked like your "XC7 cold plate temp" was running about 4C above the coolant temp measured at the pump.  That's about what I get when I run a max load CPU test to try and heat up the cold plate sensor the most.  More like 1.5C at idle.  Your pump speed is maxed out so I discounted flow rate being a factor, but perhaps that was hasty.  

[LeDoyen]

you probably moved the block a bit when rearranging the loop order, or something else solved itself in the process. as said, loop order has such small impact you can basically disregard it unless your pump runs extremely slow in a restrictive loop, like when using quickconnects or something else known to kill the flow.

[kennyg169]

so, because in my loop I have pump/res combo and it goes to the CPU and MB then to a 480mm rad(thick boy) mounted on the back wall of the case then to the GPU then to a 420 rad mounted to the top then back to the pump/res. I changed the loop order after i repasted the block and tested it, still having the temp problem the block never moved. The GPU runs a few deg hotter than before but less impact the before.

[kennyg169]

so, because in my loop I have pump/res combo and it goes to the CPU and MB then to a 480mm rad(thick boy) mounted on the back wall of the case then to the GPU then to a 420 rad mounted to the top then back to the pump/res. I changed the loop order after i repasted the block and tested it, still having the temp problem the block never moved. The GPU runs a few deg hotter than before but less impact the before.

[kennyg169]

Update , come to find out the issue was that my VRM EK water block on the MB started to corroded inside the block, so the CPU block micro fins got blocked and that’s why I was having issues with the CPU over heating. So if anyone is useing the ASUS ROG Z690 formula one MB, DO NOT USE THE VRM BLOCK.

[LeDoyen]

what coolant do you use? just curious

[kennyg169]

The Corsair pre-mix clear

 

[LeDoyen]

i don't know if it's algae or corrosion, but the coolant shouldn't allow either. maybe a bad batch

[c-attack]

Yeah, unfortunately the XC7 Elite cooling fins are very dense, so even a small amount of build up will really affect CPU temp. You usually can hear the water fighting to get through the block when this happens. I had a very small amount on mine and had clean it out over the weekend. It was a10C drop in max load temps. 
 

Do you think that’s bio growth or copper oxide from the VRM tank?  It looks like the latter. 

[NkJ]

This is a known issue for specified board due to materials used in VRM block. There are multiple reports online for this board and EKWB block. 

Edited October 2 by NkJ

[kennyg169]

So come to find out that the EK water block used the wrong materials in the block, Asus even droppped a msg about the issue, next time I will do more research.