It took 7 mins with your file. It seems to be related to the line ending not being normal for windows (they are LF only). After I 'processed' your file with this code it took less than 1 minute to scan.

#!perl
use strict;
my $t0 = time;
open FH, '<', "dict.txt" or die "$!";
open OUT,'>','dict1.txt' or die "$!";
while (<FH>) {
  print OUT $_;
}
close FH;
print time-$t0;
[download]

Original
Num. Line : 185866729 - Occ : 14900
421 secs

Converted
Num. Line : 185866729 - Occ : 14900
33 sec