With 500MB files, I'd consider MLDBM if I were to go this route. What I'd probably do (if its an option for the OP) is stick with unix utilities (I'm assuming the fields are space delimited, and you don't mind sorted results):

sed 's/ /|/' file1 | sort >tmp1
sed 's/ /|/' file2 | sort >tmp2
join tmp1 tmp2 | sed s/|/ /' > result
[download]