some benchmarks would help you decide ... for ten columns it's not much of a big deal w/  bind_columns ... because there are only 10 vars to track.

fetchall_arrayref will fetch out your entire dataset ... hence the slowdown, i'd think, talking about the number of records you're attempting to pull.

note that they are 2 different approaches ... one would entail a  while loop (  bind_columns ) while the other is a complete dataset fetch (  fetchall_arrayref ) with a  while loop for processing ....

on a side note, i'm more of a fan of  fetchrow_hashref than  fetchrow_arrayref because of the naming that's used. more extensible if you start fetching more columns ...