An aggressive thermal solution and absolute fps gain
Here is my thermal calculation for CPU/GPU AIO Intake
|1. At ambient 20 °C, air density is 1.2041 kg/m³, air heat compacity is 1005 J/kg.°C|
|2. The airflow of a 140mm Noctua NF-A14 is 140.2 m³/h, the airflow of a 120mm Noctua NF-F12 is 93.4 m³/h|
|3. The power draw of a 5.2GHz overclocked 8700K at full load with Prime95 AVX0 is 225W|
|4. The power draw of a 2.0GHz overclocked 1080Ti at full load with 3DMark Ultra is 350W|
It all starts with a question:
For CPU AIO intake alone, how much heat will rise if the CPU is doing a Prime95 test?
The heat of CPU is dissipated by a 280mm AIO cooler.
And the calculation is
225J/s ÷ ( 1005 J/kg.°C × 1.2041kg/m³ x 140.2 m³/3600s x2) = 2.387°C
The result is 2.4°C more heat dumped into the case when doing Prime 95 with an overclocked 225W 8700K CPU. And of course, the CPU is delidded. And the motherboard components such as VRM would be 2.4°C hotter, though the VRM tolerance is above 100C. And for a good motherboard such as Asus Apex, the VRM is well under 80°C during Prime 95.
For 1 GPU AIO with a 120mm intake, the heat generation will be
350J/s ÷ ( 1005 J/kg.°C × 1.2041kg/m³ x 93.4 m³/3600s ) = 11.148°C
So it will be 11.2°C hotter for the motherboard under full load with max power draw
And for multi GPUs AIO intake setup, the heat generation will still be
N x 350J/s ÷ ( 1005 J/kg.°C × 1.2041kg/m³ x 93.4 m³/3600s x N ) = 11.148°C
No matter how many GPUs set as intake, the heat rise will be the same 11.2°C
So doing the CPU plus GPUs AIO intake setup, the overall VRM temperature would be roughly 14°C hotter under full load or under stability test such as Realbench (actually 8°C. Because the radiator and fan can't dissipate all the heat generated by the die on the CPU from that small surface area. And GPU power isn't maxed out in Realbench ). Plus this is not a daily load.
But if doing the N GPUs and CPU AIO as an outtake setup, it needs as much as 2X140mm and NX120mm equal intake airflow to reach the same cooling effect. To this point, not many cases are able to mount enough fans so the CPU and GPU will be a daily hotter compared to the intake setup.
I have been using intake setup for 6 months to do many torture tests and hit some benchmarks, plus 3 months of mining, the system is stable without a problem and I didn't shut it down for 6 months, motherboard VRM is well under 70°C. Btw, the exhaust fan is running at 100%, and other intake fans spin at least 80% regardless, so yeah, it is not a quiet setup.
As for dust concern, since the airflow is high enough plus the dust filter is available, there isn't much dust accumulated on the back of the GPU. It is even better than my previous normal empty intake fan setup.
Here are some of my benchmark videos with this machine. I have compared some X299/X99 HEDT gaming performance with different CPUs. 8700K is strong, in most cases straight up beating 6950X and 7960X. Can't deny It is currently the best CPU for gaming.
|CPU||Intel - Core i7-8700K|
|Motherboard||Asus - ROG MAXIMUS X HERO|
|Thermal Compound||Thermal Grizzly Conductonaut|
|Memory||Corsair - Dominator Platinum 64GB (4 x 16GB) DDR4|
|Video Card||EVGA - GeForce GTX 1080Ti FTW3 HYBRID|
|Power Supply||EVGA - SuperNOVA T2 1600W|
|Case||Corsair - Air 540|
|3DMark||overall||graphics||Graphics Test 1||Graphics Test 2||Physics Test||Combined Test|
|Fire Strike||36838||59994||295.27 fps||233.61 fps||73.35 fps||57.06 fps|
|Fire Strike Extreme||25250||30210||162.9 fps||110.03 fps||73.72 fps||55.93 fps|
|Fire Strike Ultra||14464||15173||86.6 fps||53.28 fps||73.57 fps||35.14 fps|
Prime95 CPU performance - Min/Max FFT 1344
|Core Clock||Cache Clock||Core Voltage||Max Temperature||Power Draw|
8K GPUs performance
|GPU Position||Core Clock||Memory Clock||Core Voltage||Core Temperature||Memory Temperature||Power Temperature||Real Power Target||Power Draw||Max Power Spike|