r.neighbors

Maybe someone can help me with a problem with the command
r.neighbors.
The operation method was "mode" and the neighborhood size
of 3x3 and 5x5 were used with the same result: the output
was identical to the input.
The input used was from a r.reclass. Also used for input
was a reclass that was run through r.support and r.resample.
Both resulted in no change to the output.
Any help would be appreciated.
Help could be sent to rvh@mn600g.mn.nrcs.usda.gov
Thanks