Nope, all handled on the RTX card. The CPU load should definitely be proportional to your framerate in any direct comparison of RTX off vs RTX on. If you're also using DLSS and your framerate is going up, then the CPU usage should also go up.
I'm not sure what your issue is, but it's not the normal behaviour. There are a few notes of people using older CPUs complaining about similar things on Nvidia's own forums but it's not intended nor expected. Could be PCIe bandwidth related, there were a couple of hints that updating the motherboard BIOS fixes that. It may also be specific to one game; Not all RTX implementations are bug-free on all platforms.
For AMD IGPs, sure. That's because the AMD IGPs are fast enough that they lack bandwidth. Don't take my comment out of context though - it was a specific reply to DDR5 on the UHD 730 which is too slow to need all of the DDR4 bandwidth. Sure, if it was faster it would be able to take advantage of more bandwidth but the reality is that 32EU just plain sucks.
You don't have to guess or extrapolate, there are already several sites or channels that have investigated DDR4 vs DDR5 IGP scaling for Alder Lake that show no improvement at all. For the 96EU laptop Alder Lake models coming soon, I'm sure we'll see different results like we're accustomed to with the more compent AMD IGPs.
Gamers Nexus did a pretty solid investigation showing that the UHD 770 performance gains absolutely nothing from moving from DDR4-3200CL14 to DDR5 5200CL38. Although within 5%, the DDR4 IGP performance is better in all but one of the games tested, and that's likely to be from the improved latency of DDR4.