My first guess is that you've probably got crosstalk turned on. Additionally, the diff pair routing can take a bit of time to compute because we do a fairly exhaustive breakdown of the routed traces to make sure we have enough granularity in the simulation timestep so we don't miss any impedance discontinuities. This can slow the simulation down as well. If you don't have crosstalk turned on and it's still slow, try removing the strobe-to-clock skew measurement and see if that speeds it up at all.
I do have crosstalk on for the simulations, but it has never caused a problem in the past. In this particular design the Strobes are single ended so the only diff pair is the CLK.