The Sapphire Radeon 6900 XT Toxic's water cooling block is only touching the GPU.
The person was using this specific video card to try and prove why water cooling is vastly superior. It's not exactly a fair comparison when the water cooling block is only touching the GPU while air cooled designs still have to cool off everything. And a quick glance around say EVGA's water cooling video cards also suggest that the water block is only touching the GPU.
But overall the person used the Sapphire 6900XT Toxic as the data point to prove to me that water cooling was somehow vastly superior. And in this specific implementation yes, I can't deny that. But I'm not going to blindly believe that it was simply because the cooling medium switched to water. The card (and it seems like other factory water cooled cards) had three advantages that I'll reiterate:
- The exhaust air is pushed directly out of the case
- The water block is only cooling the GPU
- Better performing fans can be used because there's no more constraint on the space taken up by the card.
Again I would argue that the only thing liquid cooling by itself does is increase the thermal mass of the cooling system. Which only means that it takes longer for the cooling system to reach a steady state temperature. To me that doesn't necessarily mean it's better at cooling. If the steady state temperature is the same or within a range where it doesn't really matter, whatever that means to you, then it's not exactly better in the grand scheme of things. Also because of the higher thermal mass, it takes longer for the part to reach the idle temperature because the cooling system now has to cool off that water. I guess a silver lining here is that it lessens thermal shock, but how often is thermal shock a failure point?
I also want to bring up the point that regardless of what it's cooling, another main advantage to water cooling is that it shifts the heat exchanging part elsewhere, and when you do that, a lot of mechanical requirements no longer apply. Take the amount of material from a 360mm radiator and try to make say a CPU air cooler out of it. The closest we've gotten is the Noctua NH-15, but I'm sure few people have the patience to install one of those and cognizant enough to check for clearances (and the heat from the first heat sink tower is being pushed onto the second heat sink tower, lowering its cooling efficiency).
Overall, my point isn't to deny that water cooling is better than air cooling. The point I'm trying to make is that there are factors that lead up to what makes water cooling systems a better performer, and simply looking at the primary method to transfer heat away from the part (be it water, heat pipes, or vapor chambers, since few designs directly interface the fins to the part) while ignoring the rest is not productive to the discussion. To me, it's not enough to just look at a number and call it a day.
I'll just take what Gamer's Nexus goes with: