Does your system have a fix for overheating without needing additional components?
Does your system have a fix for overheating without needing additional components?
Sorry, I didn't review all the responses. From your request alone, if you're achieving mid-80s performance on an AIO of that size while running 5GHz across all cores, it seems unclear what the problem might be. The 9900(K) is recognized for being hot, consuming a lot of power and using 14nm technology. I think you're doing well.
You've already reviewed the introductory video, and it clearly outlines the issues that will arise with your setup. The first minute warns against front mounting the radiator with hoses at the top, as this can lead to coolant loss or air lock problems. So yes, you should definitely take action, and hopefully the current configuration hasn't caused any permanent damage. Usually, flipping the radiator resolves the issue, though it doesn't always work.
If you're not interested in reading the entire discussion, it's better to skip adding an uninformed comment. It might seem worse than necessary, but after spending time here, you understand what I mean.
Additionally, the original poster mentions individual cores exceeding 80°C and the package approaching that temperature. This suggests they're not letting it reach those levels and have stopped the test routines. In my view, this implies they're preventing further damage and are likely adjusting the parameters—such as allowing up to 85°C for the 9900k during the Small FFT test—to ensure safe operation. Also, make sure you're not running any AVX instructions while performing the small FFT.
A shift in perspective is necessary when dealing with liquid cooling. The cooler doesn’t simply lower the CPU temperature; it moves power from the CPU to the coolant, which then dissipates it through the radiator. What remains is the residual heat from the processing activity. The cooler’s efficiency depends on how cold it stays—cooler coolants can absorb more wattage. Initially, when testing begins, the coolant temperature matches the ambient room temperature. As the coolant absorbs energy, it transfers some of that heat to the CPU block, tubing, and radiator, gradually increasing its own temperature until it balances with the load. This process happens slowly, usually taking about 30 minutes to stabilize. During stress tests, starting conditions are lower, but over time the temperature climbs steadily until equilibrium is reached. Stress testing isn’t the right approach here; instead, using Prime95 with small FFT and disabling AVX technologies provides a more accurate picture. P95 small FFT applies uniform load across all cores, unlike stress tests that alter core usage patterns, voltages, and other parameters. For Z370 mid-range motherboards like Gaming Plus, they weren’t built to handle a 9900K all-core 5GHz setup. The VRM components are struggling because such high power demands at a constant frequency aren’t feasible in standard builds. Changing airflow won’t significantly help the VRMs—it mainly reduces current draw, not temperature.
It appears there is a general agreement that additional new components are necessary.
I must decide whether to fit the radiator vertically through the tubes or along the top of my case.
The only vertical mount kit available in-store is found at the link: https://www.microcenter.com/product/5105...rtical_Kit.
It seems designed for two particular Lian Li models and cannot be used with my current setup (HDDs require installation).
It also costs as much as some other cases.
Therefore, I plan to look for a case in-store that either includes a built-in vertical mount kit or can accommodate my 360mm radiator on the top of the case.
I won’t expect my z370 motherboard to handle my CPU without triggering the thermal alarms.
I was hoping for guidance on how to undervolt my CPU while keeping stable clock speeds—perhaps around 4.6 GHz, similar to my previous CPU.
The only helpful comment I found was: “Changing airflow won’t make much difference for the VRM’s, just lowering the current.”
Thus, I’ll visit Micro Center later today.
According to their site, they have several z390 chipset boards in stock:
- https://www.microcenter.com/product/5120...otherboard
- https://www.microcenter.com/product...ga...otherboard
- https://www.microcenter.com/product/5127...otherboard
- https://www.microcenter.com/product/5126...otherboard
MicroCenter mentions Core i9 support only in the CPU SUPPORT section for the Gigabyte board.
I’m unsure if this indicates it’s the sole option to consider.
My main aim is to use this system for gaming and AI tasks that demand high performance, so I want to know if any of these boards can handle sustained 9900K operation at high or maximum boosts for extended periods under heavy load.
I’m not sure what thermal behavior to expect during a short test, but I’d like to avoid any board reaching 60°C.
You don't require a vertical mount kit, which is designed for mounting the GPU vertically. The radiator should be placed anywhere suitable for a 360mm fan. The aio performs well, though the varying tubing lengths can be addressed with an alternative aio like the Fractal Design S36, which uses standard custom loop G1/4 fittings. You can adjust the tubing lengths as needed. You might also opt for a custom loop setup, employing a copper radiator, a 3.2 ddc pump, and a small reservoir to boost cooling capacity without enlarging the size, or consider a thicker rad such as a 45mm version. You have several choices.
I believe a stronger Z390 based motherboard would better support an i9-9900K. Look into reviews for each of the four options, focusing on VRM cooling and performance with a 9900K overclock. As expected, you’ll probably find that all four are reasonably priced.
The case is quite good for air cooling. You should address the 360 AIO or its mounting method—it’s not working well at the moment. The Noctua NH-D15 cooler seems promising. Its twin 135mm radiator towers match the size of a 360 AIO with three 120mm units, offering comparable cooling capacity. The key difference lies in the heat exchange location. Currently, the hot CPU air flows over the motherboard, which reduces VRM efficiency. With proper airflow, this won’t be a problem. The Noctua website lists it as suitable for maximum turbo and overclocking.
https://ncc.noctua.at/cpus/model/Intel-C...-9900K-307
Two front 140mm fans should suffice for the required cooling. If needed, you can upgrade to higher airflow models.
The Noctua NH-D15 delivers at 250w±, while a 360mm AIO typically ranges from 300-350w. Performance depends not only on surface area but also on fin density, fin material, heatpipes, and other factors. Aircoolers depend heavily on the narrow channel where the heat pipe travels through the fin for effective cooling; AIOs offer larger channels with multiple fins, improving saturation and watt distribution. In short, a liquid cooling radiator outperforms a heatsink in efficiency, enabling higher capacity per square inch. This is why 120mm rads often match or exceed the performance of 40mm budget coolers, despite their thinner profile. The D15 can manage high temperatures like 9900k at stock settings, especially since most games don’t fully utilize the CPU’s core potential. Running a 5GHz OC pushes it slightly, causing some games to cap around the 90s range. Comprehensive 5GHz tests such as P95 are demanding, influenced by both core voltage and required wattage, making it a matter of luck.
Just finished watching the 26 minute video about aio mounting.
I gained a lot of knowledge. It was a serious conversation.
If anything, it made me even more wary about using an aio cooler unless absolutely needed.
It bothers me that a cooler should wear out over time.
One idea to consider:
On the replacement motherboard, there are usually a few slots left for a graphics card.
Typically, the top slot closest to the CPU is an X16 slot and is usually the best spot for mounting a graphics card.
But a lower slot can also run at X8, which still provides solid graphics performance.
My observations suggest that the difference in graphics performance is minimal except with the most powerful cards.
It might be worth trying to fit your graphics card into a lower slot so you can position it optimally with the tubes at the bottom.
If you really can’t mount the 360 aio properly, then replacing it would be the best option.
If a 280 aio is your limit, then the Noctua solution should definitely be considered.
Or, explore a different case.
All the motherboards I mentioned have reviews mentioning they don’t support the 9900K. Luckily, another MicroCenter nearby offers more choices, especially when you consider open boxes. I’ve reserved an ASUS Z390 ROG Maximus XI Hero WiFi for pickup. It seems like a strong option, though one review said they had to lower the boost speed to 4.9 GHz for stable operation. Still, 16 threads at 4.9 GHz is significantly better than my old setup with 12 threads at 4.4 GHz.
There’s a high likelihood I’ll go back to the Corsair H150i Elite Capellix. The decision will depend on what I discover while browsing their cases.
*I acquired the Intel Core i9 9900K to replace my Intel Core i7 8700K after it began overheating unexpectedly.
CPUs usually don't overheat due to hardware faults; instead, it's often linked to cooling setup issues. Over the years, I've personally seen only a handful of CPUs actually suffer internal defects leading to overheating. Three in total—hundreds and hundreds of units I've owned or worked on. Two of these had physical damage from being dropped or mishandled. One developed a short circuit. The rest were affected by faulty CPU cooler mounts, dried paste (which usually takes years to develop unless the installation was poor), malfunctioning fans, improperly oriented fans, incorrect motherboard settings, or even incorrect monitoring software. Using inappropriate thermal monitoring tools can also contribute. I advise sticking to trusted solutions like HWinfo, Core Temp, and specific monitoring options for certain CPU generations—such as AMD FX processors, AM3/AM3+, AMD Overdrive, or Ryzen master.