I believe what Mooncatt3953 was getting at is that the output pixels are rendered by the GPU, not by the CPU, and thus do not get “fed” from the CPU to the GPU. Geometry data (vertex positions, arrays of vertex indexes for each triangle, etc) is fed from the CPU to the GPU.
These have less of a clear relation to resolution (number of output pixels), except where using a higher resolution leads to loading higher resolution models or more objects.
You may be thinking of pixels in texture data, though, which has to get loaded via the CPU if you don’t have DirectStorage support (afaik this is not yet supported in MSFS? I haven’t heard of them actually saying it’s ready yet but might have missed it).