Since you apparently know precisely how everything works, I'm not sure why we're involved at this point.
Huhuhuhu. Incorrect, comrade.
This is why we are involved:
The question you ask is whether pcie 3.0 will be split to x8 / x8 if you use the second slot instead of first.
Depends on mobo but it will have no affect on your graphic card performance. Consider this: a rather intensive graphic card like 1080 ti has a bandwidth of about 1.3 GB/s. A 2x connection of pcie 3.0 will suffice not to bottleneck it (4.0 GB/s). So 8x is still an overkill.
Consider another sitution, which I have experience in: you have a mobo that triggers pcie 3.0 on second slot ONLY if you put your graphics card into the second slot, else is pcie 2.0, and worse, it will only run at 4x speed even if it opens up pcie 3.0. This is still a good slot to work for a power heavy intensive graphic card nevertheless (as of 2021), as the bandwidth overhead of pcie 3.0 4x is 1GB/s x4 = 4 GB/s.
Why would anyone use a second slot instead of a first slot for graphic cards? Sometimes it is broken, or sometimes the slot is just too small because of RAM placement. But nowadays, this is an option if you want to run an NVME at full speed but your second slot only supports pcie 2.0 x4 (like a H97 PRO), this is one way to go about forcing the mobo to provide a pcie 3.0 x4 (or 8x) on the second slot. Flipping the graphic card into the second slot will certainly reduce bandwidth but yet not bottle the graphics card AT ALL. Then you get best of both worlds- a non-bottlenecked nvme blazing at 4GB/s for big file transfers and migrations using NVME, and also a graphic card that still works at pcie 3.0 speed, albeit at 4x (if you mobo is a better one, it would be 8x).
The same strategy will work on mobo with pcie 4.0 only for the first slot but has a pcie 3.0 second slot that can work at 4.0 at specific situations like I mentioned, and you want to run a gen 4 NVME at read/write ~8,000 MB/s.
Never underestimate the intelligence of the hive mind (community).