Here's a brute-force way of doing it:

use warnings;
use strict;

my $matching_line_loc = 0;
my $matching_line = '';

open (TEST_FILE, "<test.dat")
    or die "Couldn't open test.dat: $!\n";
while (<TEST_FILE>) {
    if (/333/) { #remember matching line and line number
        $matching_line = $_;
        $matching_line_loc = $.;
        last;
    }
}
close TEST_FILE;

if ($matching_line) { #don't bother with rest if no match!
    open (TEST_FILE, "<test.dat")
        or die "Couldn't open test.dat: $!\n";
    open (TMP_FILE, ">test.tmp")
        or die "Couldn't open/create test.tmp: $!\n";
    
    #print the matching line first
    print TMP_FILE $matching_line;
    #print everything BUT the matching line
    while (<TEST_FILE>) {
        print TMP_FILE $_ unless $. == $matching_line_loc;
    }
    close TMP_FILE;
    close TEST_FILE;

    #remove original file and rename rearranged file
    unlink "test.dat" or die "couldn't remove test.dat: $!\n";
    rename "test.tmp", "test.dat";
}
[download]

Note that this way (and others above) avoids the classic problem of 'slurping' the whole file into memory since it only deals with single lines at a time. Slurping would be pretty bad with a 60 meg file (though it would probably still work.)

This method also assumes that there will only every be one matching line.