If you have a CPU like the 7950x which is a hot cpu, id reco water cooled. For example, at 30-34C ambient (a rather realistic setting) mine sits at near 68-72C just doing very lite desktop work.
hot chips all right. It strikes me that an AIO would be better as it lends itself to changing the paste. The big air cooler is more work to take off and could lead to physical damage if not really carful.
For performance and especially when stock, in theory water would offer very little over air cooled primarily because:
a. both have the dreaded thermal resistance at the interface, as great as water is it still has it's own copper block which is at the mercy of poor conduction-which gives rise to alot of thermal resistance, which criples the heat flux, same for paste and the cpu top. In metals heat flows primarily via the same route as electric current, but instead of being forced along at each and every point by a force field (which is
powered from an energy source), its pushed along by nothing more than a diffusion gradient created by the temp differential and which drops off rapidly with material thickness, which inturn kills thermal conductivity or heat flux.
b. the bulk heat transfer mechanism inside the air cooler gives rise to extraordinarily efficient thermal conductvity at the interface, its not conduction (with solid, liquid, or gas), nor convection (liquid, gas) but mass transport or forced convection driven by the massive temperature differential itself, what conduction happens is over a very wide surface area (where the fins start) and where the temp gradient from the cpu is very large, as to not cause a bottleneck like with pure conduction and convection. Then, blowing a tad bit of air on the fins is way more than enough, as anyone sitting in front of a fan on a humid day knows. Interesting story, the 120mm fan on my arctic cooler cooling my intel i7870 stopped spinning due to a bad mobo header, this was during a 5hr sha512 chk sum that absolutely pinned this cpu down, out of curiosity i checked the reported temp and the cpu was sitting at 88C being cooled solely by the copper block itself just relying on conduction with the air, when I touched the cooler it gave me smart burn. The single rear case fan did almost nothing as it was spinning at like 600rpm. Still better than my stock intel P4 cooler which had my stock P4 3.8ghz hitting 90C in gaming on a moderately warm day
. By comparison, water doesn't care, it's molecules are constantly vibrating, and it's bonds in between them absorb the copper's heat very quickly and then move onwards from the flow produced by the pump, with the motion of water constantly supplying fresh molecules.
Finally c. the nature of solid state physics (along with all mechanical machinery) is such that when you push a given technology beyond a certian point you hit the law of diminishing returns, where after that point it requires an ever accelerating amount of cooling that would violate some physical contraints and where the lifespan drops massively. And this is codified in the basic power equations of a transistor and shows why water isn't as good as you might think, it only buys you a tad bit more.
So I would try to find an AIO that had great attention on the cooling block and fans, meaning, just buying based on the radiator's size alone may very well be misleading. what counts is quality first above sheer capacity.
ou may have better GPU temps with an AIO, big tower cooler that are too close to the card can impact the GPU cooling.
Hmm i dont think so. On my AM5 system the 7950x with noctua nh-d15 even under heavy gaming doesn't really impact the GPU (4080s) in any noticeable way. Because the amount of heat the cpu generates is actually quite small compared to what the air coolers actual large mass can hold, the small crossection at the interface is a big bottleneck. Even under hard benchmarking, the entire copper is barely luke warm and that heat is easily transferred via the front fan and rear 120mm fan behind it. Second, the gpu fans themselves direct the gpu heat down at the base of the case where that warm air gets removed by the lower crossbreeze caused from front and rear fan. Here, a smaller case with the same sized fans and at same rpm would make the crossbreeze more effective as there is less distance for the air to travel which means more of a stronger breeze and less chance of the hot are getting directed to other parts of the case. But really, assuming there was no power density bottleneck, this cooler and case itself could arguable handle far far more heat.
Well then, you have a lot of flexibility if you want to go with an AIO.
. 
