Also, check the value of your lower frequency range.
For low-Q structures, such as your antenna here, reducing
the Fmin can influence positively (reduce) the simulation time.
I realised that when i was simulating one of my antennas,
between 1.5-3 GHz, the simulation time reduced when i tried
0-3GHz. I didnt even have to adjust the size of the bounding
box at all !
Also, from my experience, the meshline ratio limit plays a
dominant role, even when you have low (e.g. 10) lines/wavelength
setting. For my lower frequency antennas, e.g 400MHz, i
retain 10-14lines/λ, but i increase a lot the meshline ratio. This, i think,
can improve the accuracy in the meshing, without increasing
number of meshnodes in an exponential manner.
Let us know your outcome!