The iMac Pro's relatively slow 2666MHz RAM can't be saved by the also relatively small (these days) cache in the Xeon-W it uses.
Today's desktops are as good as servers from 5 years ago. Today's laptops are as good as desktops 5 years ago. Today's phones are as good as laptops 5 years ago.
---------
The Apple M1 makes a leap with a huge L1 cache, very wide execution, at a very low clock rate (2.8 GHz for power-efficiency).
But when you pair such a leap with a 7-year technology jump, yeah, things look quite silly.
To be clear though, it doesn't matter in the end if the _result_ is more productivity. But it is worth noting that under the hood, the comparison is not entirely apples to apples.
Yes I have seen claims from people along the lines of 'x86 code has more optimization paths in mainstream compilers for historic and other reasons'. Not sure about that, certainly can't imagine such a significant difference.
Plus ARM64 is not really RISC in a meaningful way - it'll still be executing microcode under the hood.