For 1100 X 1900 X 380 mm it is too large for common workstation or cloud to handle at um wavelengh due to hardware limitation and other factors.
If the cross section is relatively small, but the simulation is long along propagation such as metalens, it can simulate only a small distance and the result can be calculated using farfieldexact:
https://optics.ansys.com/hc/en-us/articles/360042160493-Bull-s-eye-aperture (please check the script inside the analysis group)
Depending on the frequency points you want, the farfieldexact might be a little slow.
If the cross section is large, you can also refer to the above examples but now the challenge is, you may need a lot of memory and the simulation is slow.
Baiscally the simulation will need to mesh the simulation space into grids and the time step is limited by the smallest mesh size, so for large dimensions more memory and longer simulation time are normal. If your computer is powerful enough, you can use as many as processes (cores) as possible, limited by your license.