That just means that you can only evaluate a smaller fraction of the data. If your goal is to do more than sample it, you've already lost.