Gaming Discussions

[nec_v20]

For the first time ever I saw a comparison between a good air cooler and an AIO running on a dual chip Ryzen CPU and as I suspected see Definitive guide to configuring the Ryzen 3900X , the air cooler outperforms the AIO.

This is because the air cooler has 100% of its cooling capacity implemented, that is covering the hotspots of the two chiplets; whereas an AIO only has at best 50% of its cooling capacity (where the heat exchanging fins are in the cooler head) covering those hotspots.

Source: Hardware Cannucks

The Noctua NH-U12S is not even the best 120mm cooler that Noctua has to offer, that would be the Noctua NH-U12A.

A Noctua NH-U12A would knock another 6-7°C off the two fan temp of the NH-U12S.

So I would take the recommendation of AMD to use at least a 280mm AIO to cool the 3950X with a HUGE grain of salt.

[ajlueke]

"A Noctua NH-U12A would knock another 6-7°C off the two fan temp of the NH-U12S."

Not sure that would be the case.  I have a 3700X with a NH-D15 on it, and the temps still hit about 73C on the AIDA64 stress test.  My temps aren't much lower than the 3950X despite the beefier cooler, and it is possible that they are only lower at all due to the 3700X being a 65W part vs 105W for the 3950X.

My 3900X is under an EKWB custom loop, and that maxes out around 68C.  Although, the 3900X has less heat density on the cores than either the 3700X or 3950X.  

After a while, the heat transfer from the cores to the IHS becomes the limiting factor, and improved heat dissipation does not equate to further reduced temperatures.  AMD seems to want to push the narrative that better VRMs and cooling will improve performance, but it likely doesn't.

I would love to see a breakdown of several different air, AIO, custom loop setups across X370, X470, and X570 motherboards with the identical CPU.  Odds are, they will all perform about the same.