Same motherboard, same card core and memory offsets, same driver version; just a CPU swap. For fun I threw in my ITX 3700x build with a 2070 Super on a Gigabyte X570 I AORUS PRO. I haven't had any time to tune the memory on either system fwiw.

 

https://www.3dmark.com/compare/spy/7636370/spy/7769466/spy/7753159

 

7529 for the 3900x, negligible difference in graphics tests, 51.5ms time per frame

6742 for the 2700x

4805 for the 3700x with a frame time of 79.1ms Doesn't seem to want to overclock the core at all which I suppose comes down to PCIe 4.0 teething pains

The CPU score is very impressive but I am rather puzzled by the GPU tests coming out the same between the 2700x and 3900X

Odd but the all the cards should be running the Same Driver Version and the GPU should be at the Same Clock alone with the CPU Clock.
At least this would give a better compare between the CPU and the GPUs.
Using Ti Cards is odd to compare to a 2070 Super, at least a 2080 Super or Non-Ti 2080 Cards but it is general a odd compare.
Run a DX11 Benchmark like Fire Strike and good compare across gaming would be 3DMark 11. Running Port Royal to Compare DXR or RT

The More Cores the Better Time Spy 
The Faster the Memory the Higher the Graphics Test 2 Score
 
Still nice overall benchmarks

As someone who has been waiting for Microcenter to get their 2nd round of 3000 series CPUs in stock, I'm very intrigued by this thread. Curious to see you re-run the tests based off bcavnaugh's suggestions above, keeping all variables aside from processor constant.