Conversion complete
The problem with this type of build is that reseating the CPU to troubleshoot a problem involves draining the rig entirely of – in this case, particularly expensive 6nm – coolant, cleaning out as much as you can, getting enough microfiber cloths and paper towels together, then disassembling the entire top half of the loop just to get to the processor. Then you have to reassemble it all, refill it, and see if it works. If this is your only system, it’s an absolute nightmare.
The reality is, although it was an interesting challenge to convert this build, it’s not a very practical system. Swapping out blocks and coolant at the same time led to problems. What we should have done was swap the blocks earlier on, tested it again, long before the photography dates, drained it after we knew it was all working, then filled it again with the XTR coolant for our experimentation.
To be really thorough, it would have been better to have gone for a far simpler system. The Shift X is an interesting chassis, and to liquid-cool in it is as challenging as builds like this go, but it’s not very representative of a normal liquid-cooled system, nor is it particularly easy to maintain, drain, or even move around.
The results
So, how much difference did XTR make? There was a significant improvement, but not as much as we’ve seen from others who’ve used it. Performance was solid across the board – temperatures dropped significantly, in some cases by up to 6 C compared to the pastel coolant. That’s not quite as high as we’ve seen some beta testers get (there have been reports of a 9–10 C drop compared to deionised water), but we have some theories as to why our results might be lower.
The biggest difference we saw was when we were using the Shift X build with the rear glass panel and dust filters in place. This is our sort of baseline test, for what this system ought to look like, and where it gets hottest. In this setup, the temperature drop across the GPU and CPU varied 4–6 C from test to test. Just for a pure coolant swap, that’s pretty solid. However, remove the glass panel and dust filters, and the difference falls to about 2–3 C. Still an improvement on what we had previously, but not quite as dramatic.
We have a hypothesis as to why this is the case. Firstly, the system isn’t fully flushed. There’s still a lot of the old pastel pink sitting in the loop. That coolant had been in there for almost a year when we drained it, so it wasn’t exactly fresh, and that means the XTR coolant is effectively diluted. That’s a problem on our part.
Secondly, we believe a lot of XTR’s performance depends on the maximum thermal capacity of the loop. With liquid cooling, we recommend that you have a minimum of 120mm of radiator surface area per component you want to cool; double that if overclocking. Combine that with fans, airflow, internal air setup, pump speed, and reservoir capacity, and you end up with the loop’s maximum thermal capacity. Removing our rear glass panel and dust filters dropped temperatures dramatically compared to if we ran with the case sealed. This suggests that our positive pressure setup with the Phanteks fans doesn’t exhaust heat fast enough. Take the side panel off, and all that heat can escape, as the positive pressure from the intake fans pushes it out – all of a sudden, temperatures drop significantly.
The theory is that XTR is far more efficient at transferring heat into the radiators than standard coolant. When you’re at your thermal limit within a loop, the best thing you can do is move heat to a radiator as fast as possible, then transfer it into the fins to be expelled or cooled by the fans. With our panel off, the internal temps were far lower naturally, thanks to that release of internal hot air. With the panel on and internal temps far higher, XTR moves heat away from the blocks and into the radiators at a far quicker pace than a standard coolant, so we see a bigger drop in temperature.
That means if you’re looking to cool a 12-core processor and an RTX 2080 Ti off of a single 240mm skinny radiator, XTR should far outperform a standard coolant, or even water, because at that level, the temps should be much higher and the coolant should perform much better.
On to the next tests
We’re not satisfied with our results just yet; we want to see how it responds when the system’s flushed fully, and is free of pastel residue. To that end, we’re going to blitz the system and crack open another bottle of the white stuff, and see how this coolant performs over long periods under load. Rumour has it that it improves with age, and temps drop further the longer it’s in a system, so we’ll have to check that out, too. We also we want to build the “Impossible Machine”: An ITX chassis, a single skinny 240mm radiator, a tiny pump and reservoir, a 16-core processor (if not more), and the best/hottest GPU we can get our hands on. It will take a while to get the parts together, but once done, it should prove once and for all whether Mayhems’ XTR is truly the king of coolants.