The Corsair Hydro X Custom Water Cooling Review, on a Ryzen 9 3950X
by Gavin Bonshor on February 5, 2020 9:00 AM ESTTest Bed and Setup
To test the performance of Corsair Hydro X, we compared two setups which we replicated the best we could with the hardware available. The Corsair Hydro X series single 240 mm radiator loop was tested with our ASRock X570 Aqua motherboard, while we used the ID-Cooling Auraflow 240 mm AIO on the similar spec ASRock X570 Creator for comparison. We used the exact same hardware across both systems including the same OS build, and same firmware settings. As the ASRock X570 Aqua and ASRock X570 Creator are nearly identical.
The Corsair Hydro X installed on our Openbench Table for performance testing
For our stock settings, we ran with default settings with the XMP 2.0 on our Corsair Vengeance RGB Pro DDR4-3200 CL18 memory kit enabled. Our overclocking settings include XMP 2.0 enabled on our memory, with a CPU VCore of 1.35 V and an all-core frequency of 4.3 GHz.
| Corsair Hydro X System Test Setup | |||
| Processor | AMD Ryzen 3950X, 105W, $329 16 Cores, 32 Threads, 3.5 GHz (4.7 GHz Turbo) |
||
| Motherboard | ASRock X570 Aqua (BIOS 1.40 - ABBA) - Corsair Hydro X ASRock X570 Creator (BIOS 1.70 - ABBA) - ID-Cooling Auraflow |
||
| Stock Settings | AMD Ryzen 3950X, Default Settings, PBO Enabled | ||
| Overclock Settings | AMD Ryzen 3950X, 4.3 GHz All-Core, 1.35 V CPU VCore | ||
| Cooling | Corsair Hydro X Series: Corsair XD5 Pump/Reservoir Corsair XR7 240 mm radiator Corsair Softline 10/13 mm fittings Corsair Softline 10/13 mm tubing Corsair XL5 clear coolant Corsair LL120 RGB 120 mm fans Corsair Commander Pro RGB hub ID-Cooling Auraflow 240mm AIO (as base comparison) |
||
| Power Supply | Corsair HX 850 850 Watt Platinum | ||
| Memory | Corsair Vengeance RGB Pro 32 GB (4x8GB) DDR4-3200 CL18-19-19-39 1T |
||
| Video Card | ASRock RX 5700 XT Taichi X 8G OC+ (1810/2025 Boost) | ||
| Hard Drive | Crucial MX300 1TB | ||
| Case | Corsair Cyrstal Series 680X | ||
| Operating System | Windows 10 1909 | ||
Thermal Performance
The AMD Ryzen 9 3950X is a 16-core 32-thread processor designed for the X570 desktop platform. With larger core counts typically comes more heat and as standard, the 3950X has a TDP of 105 W. While this is great, TDP doesn't play out as intended once motherboard vendors implement its tweaks to maximise performance. To keep the Ryzen 9 3950X cool, AMD recommends liquid cooling as standard from its marketing.
For the temperature testing, we took delta temperatures at idle and maximum load. For our load results, we ran the Prime95 to stress our AMD Ryzen 9 3950X processor and took the value after 30 minutes. Our ambient office temperature at idle was 21°C and at load, it was 22°C during testing.
At idle, the differences aren't that major at both default settings and overclocked at 4.3 GHz. The Corsair Hydro X has the benefit of running slightly cooler with 1.35 V applied on the CPU VCore.
Running an AMD Ryzen 3950X at full load with Prime95 for 30 minutes, and we start to see the gap open up between the Corsair Hydro X series custom loop and the ID-Cooling Auraflow 240 mm CLC. Although the gap at default settings between both solutions at stock is 3°C, and at load, just 5°C, the radiator size of both options is the same. Another variable to consider is that the Corsair Hydro X Series in our testing isn't just cooling the processor, but the power delivery and chipset of the ASRock X570 Aqua. This will naturally increase temperatures as more components are being cooled, but not by a drastic amount.
Facebook
Twitter
RSS
72 Comments
View All Comments
tamalero - Friday, February 7, 2020 - link
You're using the 3,000 RPM ones?Have you tried the 2000 ones?
FatBoyDiesel - Wednesday, February 5, 2020 - link
I got two EK Vardar Ferocity fans instead of Corsair's LL120 or ML120 fans for $43.98. The Corsair fans were $40-$45 each with roughly half (or less) the static pressure. The choice was obvious to me.FatBoyDiesel - Wednesday, February 5, 2020 - link
I meant Furious, not Ferocity. Autocorrect fail.DanNeely - Wednesday, February 5, 2020 - link
I paid $18 each for the 6 Phanteks 140mm x 25mm PH-F140XP-BK fans in my current build 5 years ago. Compared to the $10 each I'd previously paid for a batch of 120mm Yate Loon fans, even that felt steep. $40+ per fan is insane; thankfully unlike most performance parts fans are also used in large numbers by 'boring' system builders so getting models without paying a surcharge for frag harder disco blights should still be possible when I build my next system.From an increase in overall noise I think one of the fans started to go bad a few months ago (and from a reduction in the amount of dust on the corresponding filter I think I know which one); the other 5 are still going strong. The increase in fan noise hasn't gotten high enough for me to confirm my suspicions by selectively disconnecting fans to see if the noise goes down.
12345 - Wednesday, February 5, 2020 - link
These radiators don't require that much static pressure. They're only 16fpi and the fins aren't louvered.eek2121 - Wednesday, February 5, 2020 - link
Corsair charges a lot for their fans, however, they also have some of the best performing fans on the market.sarafino - Thursday, February 6, 2020 - link
Noctua fans easily outperform Corsair's fan lineup.Daveteauk - Sunday, February 9, 2020 - link
You're obviously a Corsair fanboy as that is completely untrue!satai - Wednesday, February 5, 2020 - link
I would like to see a comparison (both performance and noise) to a high end air cooler.brontes - Wednesday, February 5, 2020 - link
Odd that the pics show 240+360 rads and he only uses the 240. General rule of thumb is 240/280 per cooled component.As for noise, I'm running 360x2 "thin" rads (28mm?, hwl gts, corsair's oem) and good fans (a12s, notably not corsair.) Full/artificial load on an oc 3950x and a 1080ti is somewhere between actual silence and a very subtle woosh, depending on how far below spec/stock the component temps are kept. So as far as noise goes, if that's your concern and what you build around, you can safely assume it will be imperceptible/effectively silent 100% of the time, with heavy ocs, with a larger case and 360x3 or 420x2 rads (and expensive fans.)
Air is great for cpus, especially for relatively lower wattage 7nm amd stuff, if you don't mind minor spinup/spindown from brief temp spikes.
But there is absolutely no comparison with putting the gpu under water. You simply can't move 300W of heat quietly even with a fancy axial fan 2.5 slot hsf. I'm not sure that I could ever go back to the obnoxious noise of an air cooled gpu.
A single 280(?!) rad setup, as in this article, is relatively pointless and likely sounds about the same as system on full air. Which is a little disappointing and even a bit unfair in a corsair showcase, as they're rebadging the literal best components in hwl rads and the d5 pump.