How are you scanning through your file? It sounds as if you are reading the whole file into memory and then scanning through the lines, which will be slow. The following is a reimplementation of grep that quits after the first line matching the regular expression has been found:

while (<>) {
    /...whatnot.../ and print, exit
};
[download]

But the whole thing sounds to me as if you'd be better off with a database or at least a different data structure, like a tied hash - but all those suggestions are moot as you don't show any code to work with.