Ideally KS(-l) test is preferable for a sample size more than 2000. Some differences were observed from SAS and SPSS may be because SAS makes an adjustment on the sample size itself. On the other hand, if the two results are not significantly distinct, it is no sense to prefer one over another.