Is A for your inequality constraints or equality constraints?
Also, it may depend on the algorithm. If you have redundant inequalities, the active-set algorithm will pick and choose which inequalities enter the "active set". Perhaps, the redundancy improves the trajectory of the algorithm through the feasible region.
Are the rows identical (this includes the right hand side, b) or are they scaled versions? Perhaps, the scaling improves the numerics of the algorithm.
Overall, it's not clear from what I know.
