What We Learned
Liquid Cooling Will Always Come With Challenges
I’ve done multiple liquid-cooled builds over the years, soft tubing, hard tubing, copper, ITX ones, big empty Fractal Design ones, ones in crazy cases like the Phanteks Enthoo Elite, or InWin's 909. Ones with EKWB, with Alphacool, with Phanteks, you name it, and every single time, no matter how easy you think this one’s going to be, it never is. Something always comes up. Perhaps this is in itself an evolution of the builder as each system becomes more ambitious, cleaner, crazier perhaps, and so the challenges of yesterday, once difficult and now are easy, are supplanted by the desires of today.
Lessons Learned
Either way, sweat, blood (literally) and almost tears went into this thing. However in the end, what we’re left with is an excessively clean system. It’s rare to be so happy as we are with the build. There are of course a few niggles here and there we’d like to change.
For one, the internal lighting could be better. Perhaps a single LED tucked away lighting up the back would be nice. And then there’s the tubing run from the top radiator to the VRM waterblock; that could look a bit cleaner too. Our pump mounting technique also needs work. Hell, if we could secure that in place we’d love to, it’s not exactly rattling on anything, or particularly loud, but the fact that it’s not sitting down due to the pressure exerted on it by the myriad of fittings and tubing attached to it is nothing if not disconcerting.
And then there’s the hardware. Oh how we wished we’d have gone with a GPU that had a better power target limit, or a superior VRM solution. Having a more premium aftermarket 2080 Ti would’ve been a lot more fun in the overclocking department that’s for sure.
Performance Woes
So as you can see from those performance charts, this rig packs a serious punch when it comes to the computational rendering side of things. Our Core i7-8086K seriously packs a punch, especially when it comes to single core performance. We've decided to rein in the overclock a little bit for our day-to-day activities. It is possible to run it at 5.4 GHz fairly comfortably, however the necessary voltage for that (1.45v) doesn't make us feel too comfortable.
That said 5.3 GHz still nets us a CineBench R15 single core score of 231, and a multi core score of 1,701, which is close enough to a Ryzen 7 2700X at stock, even despite the lack of those two extra cores. That said, it would be tempting to run at stock given just how low those temperatures are. 48 Celsius under load? Yes please.
The big problem however, lies with the GPU. We've got some form of throttling issue with the VRMs on the card as, even at stock, and despite the low temps, we're seeing performance anywhere between 15-22% lower than what we'd expect. It's a problem for sure. Liquid-cooling brings massive benefits in regards to both noise and cooling performance, however when it comes to diagnosing technical issues, especially ones that don't show up until benchmarking, it requires draining the entire loop, dismantling it partially, and removing the problem part, before reassembling, and filling all over again. It's arguably the biggest caveat against liquid-cooling, and although there are easier solutions out there (quick-disconnect valves for instance), you just can't compete with a traditional air cooled card.
We've been in touch with both Asus and Alphacool, and another RTX 2080 Ti is on its way to us, however to say it isn't frustrating is definitely a lie.
Temperature Testing
| Row 0 - Cell 0 | CPU Stock | CPU @ 5.3 GHz |
| CPU: Prime 95 Blend's Test | 46 °C | 66 °C |
| CPU: CineBench R15 Multi-threaded Test | 48 °C | 75 °C |
| GPU: 3DMark: Fire Strike | 57 °C | 58 °C |
| GPU: 3DMark: Port Royal (Ray Tracing) | 57 °C | 58 °C |
| CPU & GPU: Fire Strike Combined Test | 49 °C / 59 °C | 72 °C / 59 °C |
CPU Performance Testing
| Row 0 - Cell 0 | CPU Stock | CPU @ 5.3 GHz |
| CineBench R15 Single Core (Index) | 217 | 231 |
| CineBench R15 Multi-threaded (Index) | 1,409 | 1,701 |
| Tech ARP x264 (fps) | 30.84 | 37.61 |
GPU Performance Testing
| Row 0 - Cell 0 | 1440p Avg fps | 4K Avg fps |
| Total War: Warhammer II (Ultra) | 75 | 44 |
| Middle Earth Shadow of War | 87 | 56 |
| Assassin's Creed Odyssey | 54 | 35 |
| 3DMark: Fire Strike (Index) | 22,503 |
| 3DMark: Fire Strike Ultra (Index) | 6,801 |
| 3DMark: Port Royal (Index) | 5,967 |
| 3DMark: Timespy (Index) | 9,775 |
Storage Performance Testing
| AIDA64 Memory Bandwidth (MB/s) | 45,683 |
| AIDA64 Memory Latency (ns) | 50.3 |
| SSD 1: CrystalDiskMark 6.0 Sequential Read/Write (MB/s) | 3,522 / 2,237 |
| SSD 2: CrystalDiskMark 6.0 Sequential Read/Write (MB/s) | 3,518 / 3,244 |
| SSD1: CrystalDiskMark 6.0 Q32 Random 4K Read/Write (MB/s) | 406 / 314 |
| SSD2: CrystalDiskMark 6.0 Q32 Random 4K Read/Write (MB/s) | 425 / 360 |
Did We Succeed?
Well that’s the question you’ll have to answer for us. In our opinion, this thing is a miniature dream machine. A build that takes advantage of that RGB phenomenon in a clean, and sophisticated manner. It’s cool too, literally, at stock, the CPU won’t hit more than 50 Celsius, and the GPU sits pretty at 60 too. The fans are permanently running at an RPM where I can’t even hear them, and the only rattle from the machine is that D5 pump humming away in the background.
Yes, RGB may be here to stay, and yes I’m sure there are many out there who appreciate the wiley charm of that cornucopia of color, but we hope at the very least that this antithesis of a build inspires you and others to perhaps venture away from the confines of the rainbow unicorn, to build something of your own, that has character, has style, and is unique to you. Long gone are the days of brands having their own unique color schemes, but perhaps that’s just an evolution of this industry, perhaps it’s time we had our own color schemes instead.
MORE: Best Gaming Desktops
MORE: How To Build A PC
MORE: All PC Builds Content
Photo Credits: Tom's Hardware
