This is an incorrect statement, I'm sorry I didn't see it earlier. PCIe versions can ABSOLUTELY affect your performance. If you've bought a PCIe 3.0 card and expect it to perform in a PCIe 1.1 or PCIe 2.0 slot you will be disappointed. While it's not the only thing to affect performance, it can be dramatic. Per lane, PCIe 1.X slots are rated for up to 2.5 Gbit/s, PCIe 2.0 are rated for up to 5.0 Gbit/s, PCIe 3.0 - 8.0 Gbit/s, PCIe 4.0 - 16 Gbit/s, PCI e 5.0 - 32 Gbit/s, etc. So the overall capability of a device can be hamstrung by the rated slot it's plugged into.
Once this is factored in, you can also factor in the number of lanes a CPU is capable of handling. For instance, an i7-12700F can only handle 20 PCIe lanes at up to PCIe 5.0 speeds. Knowing that not just a single video card with an X 16 lane slot is in use, that leaves only 4 lanes for anything else that is plugged into a mother boards. Not to mention, when two video cards are in use, often they will be split between the two slots where the first slot is an x 8 and the second is an x 4 or an x 16 and the second slot is an x 8. This might also explain why a second, third, or more graphics cards, regardless of the technology incorporated, will give you diminishing returns for overall performance in most cases (not including specialized mining or computing circumstances with custom builds).
TLDR; PCIe versions need to be matched up to the card to get the most performance, version differences DO matter.