presumably it cools exactly like a 5800X3D at any given PPT. It's the same silicon, running the same power. The clocks are irrelevant because the heat comes solely from the power being pumped through the chip.
I believe both the 5800X3D and 5700X3D share the exact same 142W PPT from their "105W" TDPs, so they will be cooled identically, or at least within margin of error.
For a non-X3D chip, I would say there's a chance that the faster CPU's boost clocks reach higher points on the freq/voltage curve, resulting in higher hotspot temperatures for 1/2-threaded low loads at max boost clock, but the X3D chips have a strict voltage ceiling, so I doubt there's much difference between the 5700X3D and 5800X3D, if any.
I had a 5800X3D running in my HTPC with a
Thermalright SI-100 for a while, where noise levels mattered a great deal in a relatively restrictive case. I ended up just setting a 75C throttle point which stopped the CPU from reaching its fastest single-core clocks, without lowering the PPT. When the heat is spread out over an all-core workload, 142W PPT isn't that hard to keep cool.
Doing some testing I found that single sample of 5800X3D basically hit diminishing returns at about 110W PPT, so if cooling is limited, you're still getting >95% of the performance from a 5800X3D at PPT=110, TDC=75, EDC=115
