Processors
Intel® Processors, Tools, and Utilities
14395 Discussions

Intel Processor i9-13900K overheating and throttling

MetaHG
Beginner
21,805 Views

Hi!

I have just finished building a fully new computer with an Intel Processor 13900K. To check that everything was fine, I decided to run a stress test with the software AIDA64 Extreme. When idle, my processor runs at around 40°C. Right after starting the stress test, in the stability test phase, the processor immediately peaks to temperatures > 90°C to 100°C for all cores. The test indicates that the processor is throttling (seen up to 10% for the short amount of time that I have let the test run). As such temperatures are really high and quite constant, I immediately stopped the test to avoid damaging my hardware each time I have run this test. On a side note, I asked a friend to run the same test (with a slightly different build with fewer fans for cooling) and his CPU (12600K) temperatures never increased over 40°C during this test. Both of us are running a 360mm AIO for CPU cooling (from a different brand).
Additionally, note that my motherboard temperature remains at around 40°C during this test and does not skyrocket to high temperatures like my CPU. The power consumed by the CPU during the test (~350W) is also way less than the capabilities of my PSU (1300W).

A bit more information about my build. My build is not overclocked. I initially enabled XMP and then disabled it to verify if my issue was related to it, but the problem still occurs exactly the same way without XMP enabled.
I have 6 intake fans and 4 exhaust fans creating a positive pressure that should guarantee good cooling. Stress tests have been run in multiple configurations:
- With default cooling curves for fans as well as all fans running at full speed (2100 RPM each)
- Open case / closed case
My pump speed is running at maximum speed at any time (~3300 RPM) and I can feel hot air being pushed out from exhaust fans.
As I was afraid of having applied thermal paste not well enough (too much / not evenly enough), but it seems that it was ok. In any case, I cleaned everything up and apply thermal paste again. This decreased the idle CPU temperature by around 2°C, it but didn't help in any way for the temperatures reached by the stress test.

I have run Intel Processor Diagnostic Tool. No error or failure was detected (see attached results). However, CPU temperatures were constantly very high (> 90°C to 100°C) pretty constantly during two of the tests: Prime Number test and CPU Load test. I let the test run to completion even though I was worried about the temperatures of all my cores being that high for so long.

I have also run Intel SSU. Please find the scan result attached.

At the moment, I suspect a hardware issue with my processor. It could be the AIO as well but the numbers on that side look correct and I didn't notice any malfunction. Sadly, I am not able to swap any of these two components to identify the culprit as I don't own any other hardware. I am unsure if the motherboard could also be the source of this issue, but I didn't notice anything wrong here either except for a relatively high-pitched noise (very annoying to the ear) that seems to come from the motherboard cores which could be coil whine (maybe abnormally loud, currently unsure about this).

If someone can give me some guidance about the source of my problem, I would be really grateful. Does this look like a CPU hardware issue?
If you need more information about anything, please let me know!

MetaHG





Labels (1)
0 Kudos
13 Replies
Jocelyn_Intel
Moderator
21,772 Views

Hello, @MetaHG

 

Thank you for posting on the Intel® communities.  

 

I am sorry to hear what is happening to your processor, I will gladly assist you here. 

 

  1. Did you check your Cooler doesn't have any leaks, liquid loss, or pump failure? 
  2. Load the BIOS defaults in your motherboard. 
  3. Please use a validated tool like Intel XTU to test your processor temperatures, as we cannot guarantee the accuracy of third-party tools: 

 

Intel® Extreme Tuning Utility (Intel® XTU) 

Note: Click on "Stress Test" tab and "Start Testing" and then, attach a complete screenshot. 

 

Best regards,  

Jocelyn M.   

Intel Customer Support Technician. 


0 Kudos
MetaHG
Beginner
21,757 Views

Hi @Jocelyn_Intel ,

Thank you for your response.

I have checked my cooler for leaks or liquid loss and I could not find any. Regarding the pump failure, I am unsure of how to check this. Do you have any guidance on how to proceed? The only thing I can say is that the RPMs reported by my motherboard about the pump speed steadily remain at 3300 RPMs. I can also feel "tension"/"vibrations" in the tubes of liquid circulating in them similar to a working AIO I would say.

I have reloaded the BIOS defaults but I had to change a few parameters to be able to boot Intel XTU.

 

I have run several Intel-XTU 5-minutes stress tests:

  1. CPU stress test (5 minutes) - Folder "cpu-stress-00"
  2. CPU stress test (5 minutes) - Folder "cpu-stress-01"
  3. CPU AVX stress test (5 minutes) - Folder "cpu-avx-stress"
  4. CPU AVX2 stress test (5 minutes) - Folder "cpu-avx2-stress"

You can find the screenshots and logs in the zip file attached. The results look less dramatic than for the AIDA64 stress test, but we still quite often reach temperatures >90°C, less frequently and on shorter periods of time though.

 

Additionally, I would like to mention that I tried to run the FPU stress test part of AIDA64 only, and it seems to be this test causing immediate high throttling to the CPU. Not sure if this can help.

Best regards,

MetaHG

0 Kudos
Jocelyn_Intel
Moderator
21,719 Views

Hello, @MetaHG

 

Thank you for the information provided. 

 

I have checked all the images you sent and System Temperatures are fine, there are no overheating issues, and not all third-party tools are accurate in their results. If you want to test your processor, we recommend you using the Intel XTU as this is a validated tool. 

 

In any case, I noticed that you have Intel® XMP running on 6400 MHz. I would like to let you know that altering clock frequency or voltage may damage or reduce the useful life of the processor and other system components and may reduce system stability and performance. If the processor is used outside of its specifications, the manufacturer's warranty may be voided. Check the following website for further details about XMP, in the "Performance and Product information", it specifies the warning about this: What is XMP? 

 

Please set your system settings within the processor specifications. You can verify here that your processor supports DDR5 Up to 5600 MT/s and, for DDR4, Up to 3200 MT/s. Using XMP within unsupported memory speeds can cause system instability. Load the BIOS defaults or set it up manually. 

 

Best regards,  

Jocelyn M.   

Intel Customer Support Technician. 

 

0 Kudos
MetaHG
Beginner
21,703 Views

Hi @Jocelyn_Intel ,

It is possible that I had XMP enabled at the moment I performed the SSU scan. However, as I stated in my very first message, I disabled XMP as soon as I noticed thermal issues to avoid any bias in the results of the tests I run. In all the test results provided, XMP was disabled and my RAM was running at 4800 MT/s (the default of my RAM).

I agree that with Intel XTU, there does not seem to be an overheating issue. I also understand that you can't rely on third-party tools to evaluate issues with your products. I mentioned in my first message that I also encountered overheating issues with temperatures similar to the ones mentioned about the AIDA64 test while running your tool "Intel Processor Diagnostic Tool". I run this diagnostic tool again and thermal throttling was present during many tests. These temperatures do not look normal to me. Please find the screenshots and logs in the attached zip file.

The tests leading to thermal throttling were the following:

  • Prime number
  • Math
  • GPUStressW
  • CPULoad

I tried to take screenshots during or at the very end of the tests. Screenshot names match the corresponding test. I also attached more "general" screenshots that might give a better view of the Intel XTU graph.

Best regards,

MetaHG

0 Kudos
Jocelyn_Intel
Moderator
21,665 Views

Hello, @MetaHG

 

Thank you for the information provided. 

 

I will check the images, as soon as I have updates. I will post them here. 

 

Best regards,  

Jocelyn M.   

Intel Customer Support Technician. 


0 Kudos
Jocelyn_Intel
Moderator
21,569 Views

Hello, @MetaHG

 

Thank you for your time. 

 

The processor is working as expected, however, if you are still concerned about your processor, contact your local Intel Customer Support to evaluate a possible Warranty claim: 

 

U.S. and Canada: Intel Customer Support 

Europe, Middle East, and Africa: Intel Customer Support EMEA  

Asia-Pacific: Intel Customer Support APAC  

Latin America: Intel Customer Support LAR   

 

Best regards,  

Jocelyn M.   

Intel Customer Support Technician. 


0 Kudos
Jocelyn_Intel
Moderator
21,569 Views

Hello, @MetaHG

 

Thank you for your time. 

 

The processor is working as expected, however, if you are still concerned about your processor, contact your local Intel Customer Support to evaluate a possible warranty claim: 

 

U.S. and Canada: Intel Customer Support 

Europe, Middle East, and Africa: Intel Customer Support EMEA  

Asia-Pacific: Intel Customer Support APAC  

Latin America: Intel Customer Support LAR   

 

Best regards,  

Jocelyn M.   

Intel Customer Support Technician. 


0 Kudos
MetaHG
Beginner
21,537 Views

Hi @Jocelyn_Intel ,

Thank you for your feedback. It's quite an intense CPU I would say then! But good to know that it seems to be working as expected and that I just need to revise my expectations.

Best regards,

MetaHG

0 Kudos
SmartOne_2000
New Contributor I
21,552 Views

I see 2 possible solutions ...

 

1. Re-paste your CPU for optimum cooling.

 

2. Undervolt your CPU using XTU or directly in bios, though XTU is easier. First make sure undervolt protection is disabled in bios. I managed to reduce my 350W power dissipation to ~ 290W - 300W by undervolting by -50mV in my case. Others have gone lower to even achieve better power savings. All this resulted in high CB R23 scores of ~ 40,000, with max temps of 90 - 92C.

0 Kudos
MetaHG
Beginner
21,534 Views

Hi @SmartOne_2000 ,

Thanks for joining the discussion. I have read a bit more info about the CPU at different places on the internet and this is indeed one of the solutions that has been mentioned. I believe that undervolting is considered to be some sort of overclocking that can void the warranty of the CPU though, is it correct?

I have seen on other forums that as an alternative to undervolting, it is possible to simply limit the power usage of the processor with some limits called CPL or PL 1 and 2? I am not sure about the name. From what I have understood, this is not considered to be overclocking as you basically reduce the power limits closer to what Intel suggests for the processor. However, I suppose that in contrast to undervolting, this reduces the performance of the processor. Is it correct? If you have knowledge about these topics, it would be very nice of you to share!

Thank you for your response and your time!

Best regards,

MetaHG

0 Kudos
SmartOne_2000
New Contributor I
21,526 Views

No problem ...

 

1. Undervolting (UV) can never kill a CPU, unlike supplying it with excess voltage or higher than rated clock rates. Intel's own XTU tool supports UV, and so do all mobo BIOSes out there, whose bios was developed in conjunction with Intel. I can't see how UV can violate a warranty if Intel supports the feature in XTU.

UV prevents you from hitting the thermal throttling threshold temperature (100C, I believe) by running cooler; Hence you get to run at full clock speeds, allowing your CPU-intensive tasks to run at their fullest speeds. To me this is my favorite way to reduce power while maintaining top performance.

 

2. You can also limit the maximum power levels to any level you want in XTU and in mobo bioses. The levels supported by most mobos are Intel's own PL1 (stock at 253W) and PL2 (while running at Turbo speeds for a programmable shorter duration). This is a form of underclocking as a given clock speed runs at a given power level. Processor performance is reduced and its up to you if its worth it or not, whether its a reduction you can observe and live with or not. Depends on the apps you run...

--- Standard MS Office productivity apps will not notice a difference.

--- Heavy scientific apps might be noticeable and you determine if its a big deal or not.

--- Gaming you'll see a reduction in fps (for cpu bound games), but again that might not be such a big deal, especially if you are limited by your monitors resolution and refresh rate. For example, if your monitors maximum refresh rate is 144Hz, there's no point quibbling about a reduction of fps from 200Hz to 180Hz due to PL1 implementations.

Hope this helps!

0 Kudos
MetaHG
Beginner
21,338 Views

Hi @SmartOne_2000 ,

Sorry for my late reply, I have had a very busy week.

Thank you very much for all this detailed information, it's much appreciated! I have not decided yet which solution I will opt for, but hopefully, this will solve or mitigate my temperature issue.

Thank you again, and have a nice day,

MetaHG

0 Kudos
Penko
Beginner
3,638 Views

Im having thermal issues too after a year contactet support but get no answers. So please can somebody help me here? I still have guarantee and everything 

0 Kudos
Reply