It seems that you are using too many hashes, that are initialized nowhere: %hash is used but never declared, %path_file used and neither declared nor initialized.

Have you seen Scanning for duplicate files and Find Duplicate Files and the CPAN module:

File::Find::Duplicates
[download]

If used naively, that doesn't work out well for large files, because they have to be read from disc entirely.

If you care about performance, you might just want to hash the first 5% (or the first 1k or whatever) and see if there are any collisions, and if there are you can still look at the entire file.

I can agree. Another measure, taking in mind the performance, can be the filesize comparison before all.

use warnings;
use strict;
use File::Find;
  
  
   my $directory = '/mnt/music/very-good';
  
   find (\&wanted, $directory);
  
  my %path_file;
 
  sub wanted {
    my  $path = $File::Find::dir;
    my  $filename = $File::Find::name;
 
       $path_file{$path} = $filename;
 
  my %count;
  while (my ($key , $value) = each(%path_file)) {
 
      $count{$key} +=1;
 
         }
  }
[download]

There you go: the idea is to store for each filename in which paths it occurs.

#!/usr/bin/perl
use warnings;
use strict;
use File::Find;
use File::Spec;
  
  
my $directory = shift @ARGV || '/mnt/music/very-good';
  
find (\&wanted, $directory);

my %path_file;

sub wanted {
    my $path = $File::Find::dir;
    my $filename = File::Spec->abs2rel($File::Find::name, $path);

    push @{$path_file{$filename}}, $path;

}

while (my ($filename, $paths) = each %path_file){
    if (scalar @$paths >= 2){
        print "$filename occurs in these paths: ", join(", ", @$paths)
+, "\n";
    }
#    else {
#        print "$filename is uniq\n";
#    }
}
[download]

I have a question: Since %path_file is a hash why becomes an array here:

 push @{$path_file{$filename}}, $path;
[download]

Dereferencing the array @path_file or the hash %path_file

If you want to compare files by content (and not by name), you could have a look at the search and compare algorithm of http://www.fabiani.net/ -> Perl -> Downloads -> FindDuplicateFiles. It only runs under Win32 (because I only need it under Win32), but it is easy to port.