[SOLVED] I9-10900k is overheating and has volatile temps.

[NeeRix]

I know this is a overly asked question but no thread can seem to give a solid answer to this. So here goes, I have multiple questions.

To get things started, here's my spec list

I9-10900k
MSI MEG Unify
Arctic freeze 2 420 cooler (kryonaut paste and front intake mounted rad)
64gb Corsair ram
RTX3080
1000 watt corsair psu
Phanteks enthoo 719 (2 intakes on the bottom, 1 rear exhaust fan and 2 top exhaust fan.)
My current OC is 5.1ghz at 1.326 volts (this setup under normal day to day doesn't overheat even gaming, just under any heavy stress.)

Question 1. My i9 just ramps temps so fast, it goes from 33 idle to 90 instantly back and forth when put on load and equally as fast back to idle when not under load. Is this type of temp fluctuation this fast normal? I've never had an i7 do this.

Question 2. I cannot get my temps under load in control. I am a video editor and just 4k 10 bit play back in adobe premiere makes my CPU run at 100% for 2 seconds at a time, which isn't a problem until I noticed my temps hitting 100c at times. I started diagnosing what the issue could be and lowered my clock values. I just took a long shot to test because it seemed no variation of voltage or clock speed changes did anything. I set my CPU to 3.7ghz on all cores with 1.28volts and my temps dropped from high 90s on a stress test down to mid to high 80s with the same volatility as before. I have tried switching between three different coolers with no success in temp changes. Im dumbfounded that YouTube can get higher clock speeds with lower temps. My power consumption is set to unlimited. When at stock values the cpu just goes wild with voltage and boosts to overheating and back. It seems like its really difficult to setup the msi board, so any suggestions is greatly appreciated.

It should be noted that when its running at 90+ Celsius I dont feel any major heat coming from around the cpu socket.
Thanks!

 

[Darkbreeze]

IMO, the TDP and thermal ceiling for this CPU are so closely matched that in truth there really IS no headroom for overclocking unless you are lucky enough to get some kind of magical sample, and pretty much since the 10th Gen SKUs were released Intel has been scrapping the bottom of the barrel in terms of trying to eke as much performance out of the silicone as possible since they had been having so much trouble trying to move to a new process, so they just kept optimizing the existing process and scratching out a few more clocks with each new Gen, so really there AREN'T any magical or golden samples to be found in the last three gens.

I would recommend removing your overclock and running it at the stock configuration. Probably not what you want to hear, but with the TDP and the temps you are seeing, you are likely to burn that thing up if you don't. You should NOT be running that CPU under ANY kind of loads that take it beyond 85°C.

Please read the Intel temperature guide for more information and for some additional clarity on expectations with these Intel CPUs.

How To - Intel CPU Temperature Guide

Update: March 12th, 2024 Preface The topic of processor temperatures can be very confusing. Conflicting opinions based on misconceptions concerning terminology, specifications and testing leaves users uncertain of how to properly check cooling performance. This Guide provides an understanding...

forums.tomshardware.com

If it still spikes beyond thermal spec at the stock configuration, then you will need to determine if there is an issue with the cooler mounting, pump, location or fan orientation, or even something else possibly. And for what it's worth, if your CPU overheats doing ANYTHING, no matter what it is, including running Prime95 with all forms of AVX disabled, then it is not thermally compliant regardless of the fact that it "doesn't do it while gaming".

 

[drivinfast247]

I'd say you need to check the mounting of the CPU block. Have you verified the pump is actually working?

 

[Darkbreeze]

Pump has to be working else it would overheat even while "playing games" and I've never seen a system with a faulty pump that didn't have an excessively high idle temp. Plus, it would not drop back to idle temps "equally fast".

So I think we can rule out a complete pump failure, BUT that doesn't mean that it can't be a weak pump problem OR something like this going on. And this is rather important, so for ANYBODY who reads this and has an AIO cooler but has never watched this video it would be a VERY good idea for you to do so.

View: https://www.youtube.com/watch?v=BbGomv195sk

 

[uWebb429]

@NeeRix - Are you running a fixed CPU voltage? A negative offset voltage of -50 mV to -100 mV can help make the power consumption and heat output more manageable. Here is how to enter a -75 mV offset into an Asus bios.

Here is an example of a similar 10 core 10850K running Cinebench full load at 4900 MHz.

CPU voltage is a hair under 1.20V. A 10900K is a better CPU than a 10850K so you should be able to run at this speed reliably with similar voltage. I am using a Corsair H115i which should be similar cooling to what you have. Temperatures are easy to manage at the 180W power level.

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

When you start running these CPUs beyond 5000 MHz, voltage goes up and power consumption and temperatures go up significantly. You might want to consider using an AVX offset. When running heavy duty programs that use a lot of AVX instructions, the CPU will slow down automatically. I think reducing the turbo power limits is a better idea.

Set the power limits based on however much cooling you have available. The Intel default power limits are PL1=125W and PL2=250W.

The Intel default 125W PL1 is too low and will interfere with maximum full load performance. Maybe setting PL1 and PL2 to somewhere around 200W to 250W will allow the CPU to run at close to full speed when needed without getting too hot.

If you never want to see your peak CPU core temperature go beyond 90°C then set a PROCHOT Offset value of 10.

This tells the CPU to start thermal throttling at 90°C instead of the full Intel recommended 100°C. There is no reason to do this but you can choose to run your CPU however you want.

Intel has been using the same 100°C throttling temperature for virtually all of their Core i processors since they first started producing them in 2008. If CPU failures at high temperatures were a significant problem, Intel would have lowered this temperature many years ago but they didn't. Many 3rd Gen Core i CPUs like the 3770K had the thermal throttling temperature set to 105°C. Intel went back to 100°C for the 4th Gen through 12th Gen so perhaps Intel determined that 105°C was too high and 100°C was just right.