Quan Gork: Hopefully they can do multiple iterations simultaneously, rather than one at a time sequentially. Seems to me this should be the case, but I don't know this for sure.

No, I don't think so, at least not for the same product.

These production runs in the foundry are very expensive, so you will want to do as few as possible and don't waste any money on "just trying" this or that. And you need the outcome of cycle A in order to be able to prepare cycle B. You need to characterize the cycle A chips to see what is working as expected and what is not. Based on that you can go and carefully adjust your parameters for cycle B.

You will also run your adjustments through your EDA simulation. However, you had done that before with your cycle A approach, and it turned out the the actual results deviate from the simulation's outcome. So you and your EDA tool provider (Synopsys) will have to understand the (assumed) reasons for the deviations and then refine the computer model accordingly to better suit the realities of physics.

So before you actually take a refined epitaxy (or whatever it is) to the foundry for cycle B, you will want to have everything done you realistically can do, you will want to be very, very well prepared and have everything double- and triple-checked in order to not waste any time and money.

After cycle B you will quite likely be at least closer to your target, and perhaps you even meet it. But this is something you will only know after characterization of the cycle B results. Only then you can decide whether a cycle C is required or not.