When Apple started supporting external GPUs via Thunderbolt (PCIe x4) there was a lot of discussion about PCIe bottlenecks. In the end most said it wasn't a big issue for most use cases. If you do a search on PCIe x16 vs x8 there are reports how doubling the bandwidth makes negligible difference in most apps.
Yes, unified memory helps but everything else Apple did for the M1's CPU and GPUs helps performance more.
Yes, unified memory helps but everything else Apple did for the M1's CPU and GPUs helps performance more.