[Freesurfer] Paradigm file stimulus onset time

Our group has been making paradigm files in which stimulus onset time is always a multiple of the TR (e.g. 0;2;4;8;10;12;16) even though, as a result of jitter in our stimulus presentation program, the actual stimulus onset time may be off by a couple hundred milliseconds (e.g. .08;2.13;4.21;7.82;10.02;12.30;15.86). I recently noticed this sentence in the freesurfer wiki: "The onset time does not need to be a multiple of the TR or linked in any way (e.g., the onset time could be something like 11.340)." Would it be preferable if we were to alter our stimulus onset column to reflect actual rather than intended stimulus onset time, or does the relatively low temporal resolution make it a moot point?