which model is better for a two-layer PCB in HFSS?

I am simulating a PCB board with two conducting layers on the top and bottom of a FR-4 dielectric layer. There are two ways to model the conductor layers
1) Draw two very thin boxes and assign PEC
2) Draw two rectangles and assign Perfect E boundary.

Which one is better in terms of the accuaracy of the solution and efficiency?

Thanks.

I doesn't make a big difference. But why don't you make two very thin boxes and assign finite condutivity and copper as material?
The closer to real thing you make your model the better.

Thank you very much