What dataframe operations do you find to be slow? Usually I'm able to get huge performance wins by rewriting my slow R code in a loop free way (*apply and friends).