Yes, the rotamers are also identical in many decoys.
In case when I allowed for sidechains rotamers changes in the rest of the protein, the sequences of designed fragment are identical in 84 decoys (from 100 calculated),
and among them there are only three unique decoys.
I was wondering if it is not the problem with random seed, that seems to be the same for different runs, but I do not know if it used here…
We had similar problem with some version of Rosetta2, which we used for de novo modelling. When we run it on the computer cluster, from some reason the random seed was the same for different runs and we were obtaining many identical decoys. To avoid it we had to specify random seed manually for each run.