CST MWS results are different from FDTD results...

I heard that two simulation tools operate similary.
But, when I simulate the nano aperture in optical region,
these two results has different resonant frequency.
FDTD : 970nm
CST MWS : 1250nm
both of cases, I use the grid size of about 2~2.5nm (I think it is enough)
and same structure...
is it reasonable range? I don't think so.
Is grid size in the fdtd also same in CST MWS Transient solver?

Threr is a big difference in the mesh representation. If you optimize mesh the results will be more accurate and closer to each other.
FDTD usually use simple cartesian grid (you get something like sugarcubes). CST has better meshing at this case. If you want to see the difference, change the meshing in CST to hexahedral - staircase (in special mesh properties).
I simulated aircrafts with FIT (CST) and FDTD solvers (i think big and complex enough to see some difference). However if you do the meshing correctly the difference is relatively small and insignificant.