And as I mentioned before that Fmax is a lower bound and you can likely do better with a better i/o and harness setup

I have a better I/O cell, do you want to do another Tapeout with my GPIO cell? In 3.3V VDDIO it has delay of 1ns-1.5ns

There's a couple of groups working on improved I/O cells, I'd be interested in characterizing them all