GraptikBlizzard

Prominent
Jul 3, 2021
32
1
535
Just got my 10900K and paired it with a NH-D15S and I'm having huge temperature problems. I can't even run prime95 at stock settings without hitting 100c. By stock, I mean that I loaded optimised defaults for my board (Z490 MAG Tomahawk) and then disabled enhanced turbo. Really not sure what is going on and would appreciate some help.
Thanks!
 
Solution
... 10900K ... NH-D15S ... can't even run prime95 at stock settings without hitting 100c ...

What is your ambient (room) temperature?

Have you disabled all AVX test selections in Prime95?

There's a few important details concerning stress test utilities which must be pointed out and explained.

Don't thrash your system using P95. Stress test the system using Aida64 which is much better to determine thermals and rail voltages.

Although MeanMachine41's previous advice is excellent, the portion of his comments quoted above, when left as a blanket statement without further explanation, is misleading.

“Stress” tests vary widely in load levels, Power consumption and the Core temperatures they produce. As such...

GraptikBlizzard

Prominent
Jul 3, 2021
32
1
535
are you absolutely sure you've installed the cooler correctly?
my first thought would be you've either not tightened it sufficiently, not applied thermal paste correctly, or some other basic error.
Thanks for the reply! Yes I am sure it is installed correctly, I have re-installed it twice, following the exact steps in the instrustions and screwing it down tight, using a good amount of thermal paste. There is no peel on the heat plate and my case has good airflow. I just don’t know what I’m missing
 

GraptikBlizzard

Prominent
Jul 3, 2021
32
1
535
can you feel a lot of heat being blown away from the cooler's fins?

try setting the fan to 100% while attempting to run Prime95.
do you notice any difference in temperature this way?
I have tried setting the CPU fans (and case fans) to 100% as you mentioned and it did seem to lower the temperatures by a few degrees on some of the less intensive tests but I still reach 100c on prime95. Putting my hand in front of the exhaust fan, the air being exhausted is very warm.
 
The 10900k does run hot. I would feel uncomfortable with temps at 100C under load and would want it to be around 80C max.

Tj Max is 100C however your cooling system is of concern. Test with the system under load and with the front case panel removed. Point a desktop fan into the case and see if temps improve.

Although Noctua is a good air cooler at stock settings, you may want to consider an AIO water cooler especially if you intend to Overclock the CPU. I would recommend a H110I or equivalent for a medium Overclock.

Don't thrash your system using P95. Stress test the system using Aida64 which is much better to determine thermals and rail voltages.
With AIDA64 stress the CPU, FPU and Cache and also view HWinfo64 as a comparison. Run the test for 20Mins and stop the test if temps reach 80C. Take screen shots at the 20min mark.
The test should be run at stock settings in Bios and please report the results.
 

iTRiP

Honorable
Feb 4, 2019
929
87
11,090
It's just thermal throttling @ 100 degrees, nothing to be concerned with, under normal computer use it wont hardly ever reach that high, since you mention that it's only doing those high temps in a stress test.
 

alexbirdie

Respectable
Your case fans: Best would be 2 fans in front blowing in and 1 in the back exhausting.

If your configuration is like above ( front intake should be open, that means, only a dustfilter in front of them), than you should think about an AIO ( at least a 240, better would be a 360, fans of it should be exhausting through then radiator).
 

CompuTronix

Intel Master
Moderator
... 10900K ... NH-D15S ... can't even run prime95 at stock settings without hitting 100c ...

What is your ambient (room) temperature?

Have you disabled all AVX test selections in Prime95?

There's a few important details concerning stress test utilities which must be pointed out and explained.

Don't thrash your system using P95. Stress test the system using Aida64 which is much better to determine thermals and rail voltages.

Although MeanMachine41's previous advice is excellent, the portion of his comments quoted above, when left as a blanket statement without further explanation, is misleading.

“Stress” tests vary widely in load levels, Power consumption and the Core temperatures they produce. As such, it's critical to correctly configure stress test utilities, rather than run them at default values.

Stress tests can be characterized into two categories; stability tests which are fluctuating workloads, and thermal tests which are steady workloads. When configured correctly, Prime95 Small FFTs (AVX disabled) is ideally suited for testing thermal performance, because it conforms to Intel's Datasheets as a steady-state 100% workload with steady Core temperatures. No other non-proprietary utility can so closely replicate Intel's thermal test workload.

As per Intel’s Datasheets, TDP and Thermal Specifications are validated “without AVX.

Prime95 with AVX test selections enabled will impose a brutal and unrealistic 130% workload which can increase Core temperatures by up to 20°C. To correctly set Prime95 to run your CPU at 100% workload, simply click on the AVX test selections that are not greyed out so that all three AVX boxes are checked, as shown below.

XHNNFmQ.jpg


Utilities that don't overload or underload your processor will give you a valid thermal baseline. Here’s a comparison of utilities grouped as thermal and stability tests according to % of TDP, averaged across six processor Generations at stock Intel settings rounded to the nearest 5%:

doQfNIZ.jpg


Although these tests range from 70% to 130% TDP workload, Windows Task Manager interprets every test as 100% CPU Utilization, which is processor resource activity, not %TDP workload. Core temperatures respond directly to Power consumption (Watts), which is driven by workload. Prime95 Small FFTs (AVX disabled) provides the correct workload for testing thermal performance. If Core temperatures don't exceed 80°C, your CPU should run the most demanding real-world workloads without overheating.

AIDA64 has 4 CPU related stress test selections (CPU, FPU, Cache, Memory) which have 15 possible combinations that yield 15 different workloads and 15 different Core temperatures. The individual FPU test is about 115% TDP workload, the CPU/FPU combination is about 90%, all 4 tests combined is about 80% and the individual CPU test is only about 70%. All other AIDA64 test selections are fluctuating workloads which are suitable for stability testing, but not for thermal testing.

In recent games with AVX, as well as real-world apps with AVX such as those used for rendering or transcoding, the AVX code is less dense than Prime95 with AVX. So when heavy, fluctuating AVX workloads in games or apps spike to "peak" Power consumption, Core temperatures will typically approach, but not exceed P95 Small FFTs without AVX. The CineBench R23 Multi Core test shown above in the %TDP Scale is a good example of a utility which replicates heavy, real-world AVX workloads.

If OCCT's first test, called "CPU", is configured for Small Data Set, Normal Mode, Steady Load and SSE Instruction Set (no AVX), then it's a steady-state workload at more than 97% that's nearly identical to the steady-state 100% workload of Prime95 Small FFTs without AVX. A steady-state 100% workload is key for thermal testing so the CPU, cooler, socket, motherboard and voltage regulator modules (VRM) can thermally stabilize.

Although the Noctua NH-D15 and the NH-D15S are excellent air coolers, your 10 Core 20 Thread i9-10900K is a high power consumption processor which demands top-of-the-line cooling that can extend beyond big-air into the realm of 280, 360mm AIO or custom loop cooling.

CT :sol:
 
Solution

GraptikBlizzard

Prominent
Jul 3, 2021
32
1
535
GraptikBlizzard,

Have you disabled all AVX test selections in Prime95?
If not, then have you configured an AVX Offset in BIOS?

There's a few important details concerning stress test utilities which need to be pointed out.

“Stress” tests vary widely in load levels and the Core temperatures they produce. However, it's critical that you configure stress test utilities, rather than run them at default values.

Stress tests can be characterized into two categories; stability tests which are fluctuating workloads, and thermal tests which are steady workloads. When configured correctly, Prime95 Small FFTs (AVX disabled) is ideally suited for testing thermal performance, because it conforms to Intel's Datasheets as a steady-state 100% workload with steady Core temperatures. No other non-proprietary utility can so closely replicate Intel's thermal test workload.

As per Intel’s Datasheets, TDP and Thermal Specifications are validated “without AVX.

Prime95 with AVX test selections enabled will impose a brutal, unrealistic 130% workload which can increase Core temperatures by up to 20°C. To correctly set Prime95 to run your CPU at 100% workload, simply click on the AVX test selections that are not greyed out so that all three AVX boxes are checked, as shown below.

Thanks so much for your reply! I've been performing moderate overclocks on my CPUs for over 7 years now and there was some information there that I have never seen before so that is super helpful! Definitely saving that TDP percentage graph for future reference.. After doing some testing with non-AVX workloads as you suggested (e.g. Prime95 with AVX disabled), temperatures are looking much more reasonable. Maxing out at about 80c package temp with Prime95 small FFT. I also tried some of the other benchmarks on the TDP graph you posted. IntelBurnTest and AIDA64 max out around 75C which is perfect. However, ASUS Realbench brings me into the 90+ zone and after looking online, it seems that Realbench uses AVX. So now that I have temps at a level that I'm comfortable with, my question would be: How much of an idea of "real world" performance does this give me if I were to run into an AVX workload. Say for example that I do a lot of stability testing and the OC is stable, but then I run an application that utilises and AVX workload, would the CPU run into temperatures similar to those seen during the Prime95 run with AVX enabled? Or would real world use cases be much less stressful (and therefore less hot) than that?
Thanks!
 

CompuTronix

Intel Master
Moderator
I was editing my post just before you posted, so please go back and read it again. What I added was:

"In recent games with AVX, as well as real-world apps with AVX such as those used for rendering or transcoding, the AVX code is less dense than Prime95 with AVX. So when heavy, fluctuating AVX workloads in games or apps spike to "peak" Power consumption, Core temperatures will typically approach, but not exceed P95 Small FFTs without AVX. The CineBench R23 Multi Core test shown above in the %TDP Scale is a good example of a utility which replicates heavy, real-world AVX workloads."

Do you have an "AVX Offset" configured in BIOS? An AVX Offset "downclocks" the processor during AVX workloads, which will reduce Core voltage, Power consumption, and therefore, Core temperatures. An Offset of -2 (200MHz) or more may be needed during AVX stress tests to keep Core temperatures in check.

Although "Throttle" temperature is 100°C, as MeanMachine41 has correctly pointed out, it's not OK to run it that hot. If your hottest Core is near its specified Tj Max Throttle temperature, then your CPU is already too hot. The consensus among well informed and highly experienced reviewers, system builders and expert overclockers, is that it's prudent to observe a reasonable thermal margin below Throttle temperature for ultimate stability, performance and longevity. So regardless of environmental conditions, hardware configurations, software workloads or any other variables, Core temperatures above 85°C are not recommended.

As such, here's the nominal operating range for Core temperature:

Core temperatures below 80°C are ideal.

PdancCI.jpg


And of course, environmental conditions can make a huge difference. Core temperatures increase and decrease with ambient (room) temperature, for which the International Standard for "normal" is 22°C or 72°F. So for every degree your ambient is above or below normal, so are all computer temperatures, which includes your Core temperatures.

We have a Guide that covers this topic; it's one of the "Stickies" which are information resources located at the top of the CPUs Forum so they're always available for everyone's benefit. Just click on the link in my signature.

CT :sol:
 

GraptikBlizzard

Prominent
Jul 3, 2021
32
1
535
I was editing my post just before you posted, so please go back and read it again. What I added was:

"In recent games with AVX, as well as real-world apps with AVX such as those used for rendering or transcoding, the AVX code is less dense than Prime95 with AVX. So when heavy, fluctuating AVX workloads in games or apps spike to "peak" Power consumption, Core temperatures will typically approach, but not exceed P95 Small FFTs without AVX. The CineBench R23 Multi Core test shown above in the %TDP Scale is a good example of a utility which replicates heavy, real-world AVX workloads."

do you have an "AVX Offset" configured in BIOS? An AVX Offset "downclocks" the processor during AVX workloads, which will reduce Core voltage, Power consumption, and therefore, Core temperatures. An Offset of -2 (200MHz) or more may be needed during AVX stress tests to keep Core temperatures in check.

We have a Guide that covers this topic; it's one of the "Stickies" which are information resources located at the top of the CPUs Forum so they're always available for everyone's benefit. Just click on the link ion my signature.

CT :sol:
Thanks for the clarification! I gave Cinebench R23 a run and the temperatures do indeed match those of Prime95 no AVX (in fact they're actually about 4 degrees lower). I have it clocked at 4.9GHz @1.21v at the moment. If I run the 100% TDP stress tests shown on that image, I actually have a little thermal headroom that I could use to push to 5GHz but I will probably keep that headroom for peace of mind. 80C on Prime95 no AVX @4.9 is perfect for my usage. Thanks so much for your help, I was beginning to think I had a dud of a cooler!