This discussion sure took a ... let's say interesting turn. I mean, arguing for introducing arbitrary limitations and/or advantages in order to "show" which is "best"? That sounds ... problematic. Should coke-v-pepsi taste tests be decided by adding mustard to the pepsi beforehand? Liquid metal can be used anywhere you want to. TIM is (mostly) universally applicable (just don't use LM on an aluminium cold plate). Cooler size is dependent on what the case can fit, regardless of whether that's air or liquid cooling. Etc., etc. Setting arbitrary limitations on any of these, whether it's only using LM on air cooling or not fully populating your radiator mount ... proves nothing except the
radical notion that worse cooling = worse cooling.
It's obvious we can't come to any hard conclusion discussing like this. I would argue that such a conclusion
is impossible, as there are too many variables in play - from CPU/GPU power, thermal density, case design (both in terms of cooler fitment/clearances as well as component layouts), case airflow, fans, pump speeds and flow rates, noise characteristics of various coolers (and cases and fan grilles and other relevant factors) - and a heap more.
But that's not the point either, is it? Is the interesting part of this discussion finding a universal and absolute conclusion? I would think the interesting part is
the discussion itself, hearing the arguments (and thus use cases, priorities and preferences) of others. There is a 100% chance that someone will bring up something that you didn't consider or consider important; that someone will have very different priorities than you. Isn't exploring these connections - between preferences, priorities, desires and the choices made - the interesting part here? This becomes especially poignant once the search for some absolute answer leads us into some absurd nit-picky defining of conditions like what we've seen here. Is that interesting? Not to me, at least.
I mean, if it's hard answers we're looking for it's easy to just list the pros/cons of each broad type of cooling. We've all seen a hundred lists like that. Each solution has both advantages and disadvantages. Isn't it far more interesting to see which of these people choose to prioritize, and why?
they prefer to run their CPUs at 80c rather than build a Custom Loop.
This is a perfect example! I
have a custom loop,
and run my CPU around 80°C. Why not? The CPU has absolutely zero problems delivering peak performance at that temperature. And doing so allows me to run a (hot-running!) 5800X (-10 CO, +150MHz PBO, slightly lowered PPT/EDC/TDC, ~120W peak) and a 6900 XT (mostly running at a ~180W UV+UC profile, but fully capable of running at 330W if I need the performance) in a 14.7l case. Part of the reason for the high CPU temps is my Nouvolo Aquanaut CPU block+DDC pump mount, which isn't the best thermal performer. But who cares? I get a great pump, that runs very quietly, in a tiny form factor, and my CPU isn't complaining. I'm not extatic that my CPU is that hot, but the 5800X is notorious for running hot, so ... meh. The pros by far outweigh the cons. The overall package is fantastic - a tiny PC with fantastic performance, great efficiency (especially with the GPU profile), low noise (the system is literally inaudible over ambient noise for desktop usage, and even with the pump and fans pegged at 100% it's reasonably quiet, yet cools a ~500W load admirably). But as I said above: if I was starting from scratch today, I'd likely go air, due to cost and simplicity - something like a Noctua U9s or C14S and a honking huge air-cooled GPU in a Dan C4 or CM NR200 (could fit an even larger air cooler there) would likely be my choice then. It would be a bit louder under load (though deshrouding the GPU would no doubt alleviate some of that), but it would still be great.
The point is: air can be great, AIOs can be great, custom loops can be great - it all depends on the situation and implementation. Isn't that
so much more interesting than some pointless trench warfare about which is "best"?
